Split (1)

train · 150k rows

file stringlengths 27 28	text stringlengths 1 5.17M
PMC010xxxxxx/PMC10239708.txt	==== Front Int Rev Educ Int Rev Educ International Review of Education. Internationale Zeitschrift Fur Erziehungswissenschaft. Revue Internationale De Pedagogie 0020-8566 1573-0638 Springer Netherlands Dordrecht 10005 10.1007/s11159-023-10005-1 Editorial The future is not what it used to be Stanistreet Paul p.stanistreet@unesco.org grid.473814.d 0000 0001 2199 9798 UNESCO Institute for Lifelong Learning, Hamburg, Germany 5 6 2023 2023 69 1-2 113 © UNESCO Institute for Lifelong Learning and Springer Nature B.V. 2023 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. issue-copyright-statement© UNESCO Institute for Lifelong Learning 2023 ==== Body pmc “The old dear stories of possibility. No-one wanted to hear them anymore, but nothing had replaced them.” Joy Williams, Harrow Eight years ago, in 2015, United Nations (UN) Member States adopted the 2030 Agenda for Sustainable Development, “a shared blueprint for peace and prosperity for people and the planet, now and into the future” (UN 2015). Its 17 Sustainable Development Goals (SDGs) represented an “urgent call for action” to all countries, spanning a range of fronts – from climate change and poverty reduction to health and education – but also recognising their interdependence. Only by addressing these issues in the round, in an ambitious, inter-sectoral way, with partnership at its heart, the UN argued, could we hope to achieve the 17 goals and create a fairer, more just and sustainable future for everyone. Now, a little over halfway to 2030, it is obvious that we are failing to achieve these goals and that the defining promise of the 2030 Agenda to “leave no one behind” is, as the UN itself puts it in a new report, “in peril” (UN 2023, p. 2). Its preliminary assessment of the 140-plus targets found that only 12 per cent were on track, with close to half, while showing progress, “moderately or severely off track” and 20 per cent either regressing or showing no progress (ibid., p. 2). On current trends, the UN estimates, 575 million people (almost 7 per cent of the world’s population) will still be living in extreme poverty in 2030 (ibid., p. 7). Projections also show that approximately 670 million people will still be facing hunger in 2030; some 8 per cent of the world’s population, the same as in 2015 (ibid., p. 8). Remarkably enough, this is not the worst news. Action to address the climate crisis remains insufficient, the UN says, with the world “on the brink of a climate catastrophe”. Without “transformative action … to reduce greenhouse gas emissions deeply and rapidly in all sectors, the 1.5 °C target will be at risk and with it the lives of more than 3 billion people” (ibid., p. 19). SDG 4 on education and lifelong learning has a special cross-cutting role in supporting the achievement of the other 16 Sustainable Development Goals. It has an important part to play in enabling action to combat climate change (SDG 13) and in ensuring the sustainable use of oceans (SDG 15) and forests (SDG 15), for example, as well as in lifting people out of poverty (SDG 1) and reducing inequality (SDG 10), and in ensuring sustainable consumption and production (SDG 12). More generally, it has a critical role in generating informed public support and civic action and in encouraging political action for change. As the final report of the International Commission on the Futures of Education (ICFE 2021) recognised, education can help build a supportive, cooperative and more just social fabric that repairs past inequalities and injustices while redefining our relationship with the natural environment (ibid.). However, this promise is far from being realised and it is important to recognise just how far. The UN report estimates that, by 2030, 84 million children will be out of school and 300 million children or young people attending school will leave unable to read or write (UN 2023). To achieve SDG 4, “education systems must be re-imagined, and education financing must become a priority national investment” (ibid., p. 10). A Global Education Monitoring Report (GEMR) Policy Paper, published in April this year, found that the 79 low- and lower-middle-income countries are facing an average annual national financing gap of USD 97 billion per year to achieve their reformulated and less ambitious national SDG 4 benchmarks by 2030 (GEMR Team 2023). In the unlikely event of this gap being filled, these countries would still be well short of the previous set of benchmarks, now understood to be unachievable. The UN has put increasing emphasis on the need to “transform education”, recognising not only its contribution to “upholding people’s rights and human dignity and to the advancement of social, economic, political and cultural development” but also the certainty that this contribution would not be fully realised without genuinely “transformative action” (UN 2022a, p. 3). Education, it reminds us in its report of the Transforming Education Summit (TES) held in New York in September 2022, remains in “deep crisis”, a “crisis of equity, quality, and relevance” through which “[h]undreds of millions of the most vulnerable children, young people, and adults remain excluded” (ibid.), thus denying them their right to education and limiting their “potential to achieve other rights and freedoms” (ibid., p. 6). The numbers involved are truly shocking: 763 million young people and adults without basic literacy skills; 244 million children and young people out of school (ibid.). The UN Secretary-General’s vision statement, “Transforming education: An urgent political imperative for our collective future” (UN 2022b), calls for a fundamental rethink of the purpose and content of education, grounded in the principles identified by the UNESCO International Commission on the Futures of Education: ensuring the right to quality education throughout life; and strengthening education as a public endeavour and a common good (ibid.). The vision statement calls for more investment in education, targeting in particular “usually discriminated against” groups, such as “women and girls, ethnic minorities, persons with disabilities, indigenous populations, and those in protracted crises amongst others”, and for “the collective commitment and action of visionary political leaders at all levels, parents, students, teachers, and the public at large” (ibid., p. 32). Few would deny that there is a pressing need for transformation, in education and in other areas where the actions of countries are out of step with their policy commitments. But the gap between vision and reality is so great that it must surely call into question the purpose of such ambitious overarching goals and the ways in which they are useful. Do the Sustainable Development Goals, for example, represent an achievable framework for action by Member States or are they, rather, as the GEMR Policy Paper suggests (GEMR Team 2023), hopeful aspirations which, even if missed, will have served a purpose? Do reports such as Reimagining our Futures Together (ICFE 2021) create expectations of the future that have little or no chance of being realised? Their “promissory” or “anticipatory” nature, as it has been described,1 offers hope, a kind of philosophical orientation and a means of motivating and lobbying for change and galvanising political will. The authority of international organisations could, as Jens Beckert’s work2 suggests, be said to rest on the credibility of the promises they make concerning future outcomes. If hope is exhausted, “disorientation and discontent will arise” and “promissory legitimacy fails”, he writes (Beckert 2020, p. 3). Whether or not you agree that the frameworks of international organisations such as the UN, UNESCO or the Organisation for Economic Co-operation and Development (OECD) should be thought of as “promissory” in this sense, credibility is evidently threatened by commitments that everyone knows have little prospect of being delivered. The pledges made by nation-states will begin to look shallow and half-hearted, perhaps even dishonest. The interventions of international organisations will be seen as unrealistic and impractical, ill-fitted to the hard realities of real-world politics. Citizens will become sceptical and disaffected, more concerned with holding on to what they have, perhaps, than with creating something new and better – the essential task of our age. Beckert shows how the promissory legitimacy of neoliberalism has been eroded by its failure to deliver the social, economic and, indeed, educational gains it claimed would be achieved through privatisation, an increase in competition and the strengthening of the role of markets in different areas of life. Neoliberalism’s credibility is all but gone, kept alive only by a scattered group of zealots, gifted in magical thinking, who believe its failure can be explained by governments’ reluctance to go far enough in the way of deregulation and privatisation and the dark machinations of a mysterious cabal they call the left-wing economic establishment (Smith 2023). This failure, and the crisis of legitimacy it has created, is, for many, a source of hope and has, I suspect, given international organisations the courage and confidence to assert different values and to renew the language of solidarity, democracy and the common good, notably in the ICFE (2021) report and the UN’s Common Agenda (UN 2021). This is a gratifying and long-overdue turn of events, and it has been welcomed by many in the education community. However, it does not amount to (nor does it claim to be) a serious alternative vision to discredited neoliberalism. While neoliberalism lost credibility because it does not work and has been seen not to work (or, at least, it does not do what we were told it would), I think the international community faces a different credibility problem that threatens, nevertheless, to limit its influence. Although the vision offered by the UN and UNESCO is a compelling and attractive one, there is no roadmap or plan, and to the minds of most people little chance of realising it. Furthermore, there is little attempt in any of these documents and frameworks to understand or face up to the causes of the crises we face, in education and elsewhere, still less to address them head-on. The ambitions set out by the UN and UNESCO and other international organisations make little acknowledgement of the entrenched systems of power and inequality that resist change and uphold the discredited promise of neoliberalism. This is understandable, given the remit of these organisations and their modus operandi. But it becomes an issue in a context in which countries routinely make commitments against which they have little chance of delivering, and private corporate interests exercise undue, usually undisclosed influence over how public money is spent and wealth distributed. It is hard to shake the impression that, set against these interests, national politics lacks both substance and power. What we see, for the most part, is smoke and mirrors. There is an unbridgeable gulf between those who want to empower people to remake the world and those who wish to empower themselves to conserve their privilege and dominate others. Unhappily, it is the latter who direct national policymaking and shape the world in which we live. Even in countries in which representative democracy is quite well developed, party politics is distorted by the influence of wealthy donors and other interests, such as the media, still dominated by a few powerful individuals, and the formidably persuasive fossil fuel industry. Growing inequalities and the concentration of wealth have increased the influence of the very rich over domestic politics (Economist 2018). Kleptocrats act with near-total impunity, above the laws of nations (“[m]oney”, as Rachel Cusk has one of her characters say in the novel Outline, “is a country all its own”; Cusk 2014, p. 7). There is little prospect of persuading these interests to change direction or do something else less harmful since their power and wealth rests on doing precisely the things they have been able to do to date, thanks, in no small part, to their ideological cheerleaders and apologists in national government. There is no conversation to be had because there is no common ground. This is a potentially fatal problem when hopes for change rest on asking those in power to behave differently or to be a bit nicer. And it is exacerbated when politicians put their names to commitments they do not think of as binding or of the first importance and which, in many cases, they know they will not be able to honour, often because they consider their hands to be tied by other interests or by their perceived lack of real power. Would another set of political leaders do better? In some exceptional cases, perhaps, but in general, I suspect, no. It is more a matter of how politics is done than of who does it. If mainstream politics offers only roadblocks and resistance, where else might we look for the radical change of direction we so desperately need? There is a hint in the UN Secretary-General’s TES vision statement, which appeals not only to politicians but to “parents, students, teachers, and the public at large” to come together and act collectively (UN 2022b, p. 7). The crises we face cannot be resolved through more of the same, that much is clear. They demand informed civic engagement, critical thinking, and active participation across society. The causes of the crises must be understood and challenged. Hope lies in people wanting a radically different future and in their being persuaded that it is possible. The problem, and the reason this is so difficult to talk about, openly at least, is that it threatens the dominance of the very powerful, whose grip on mainstream politics and the media seems unassailable. ICFE’s appeal to a social contract is perhaps also an implicit acknowledgement that conventional politics needs to be transcended, to some extent at least. We cannot afford to address ourselves only to governments. How else can we agree “a shared vision of the public purposes of education” (ICFE 2021, p. 2) other than through “inclusive and democratic public dialogue and mobilization to realize it”? (ibid., p. 15). That dialogue needs to be far-reaching, informed, critical, plural, and self-consciously challenging, going beyond the naturally limited remit of the report and its commissioners to confront the causes of these crises. For real, sustainable change of the sort the ICFE report talks about, we need a society in which people find joy and fulfilment in civic participation rather than private accumulation. And we need education that promotes critical thinking and engagement and action in the public sphere and empowers people and communities. The public sphere has been hollowed out in many places in the world, with local democracy eroded and public activism discouraged and, in some places, criminalised. It is increasingly difficult to find spaces, physical or intellectual, in which to meet, think and act publicly, with others. Public discourse, for the most part, is animated not by hope and the expectation of a better world, but by fear of change and difference, ritual and idolatry, and a silly-sentimental attachment to an idealised past. Autocratic leaders conjure national mythologies and find in these myths the justification for contemporary crimes and incursions. While many people live lives of unbearable hardship, in other places ignorance and denial, and an attachment to a fantasised version of the past, have become badges of honour. Actions that silence dissent are applauded by the very people who stand to lose the most from the erosion of the right to protest and speak out publicly. We have become sluggish and introverted in defence of these rights, frequently thoughtless, and appallingly myopic, convinced that however bad things are they could not, in any case, be much better, whatever politicians did, and could almost certainly be much worse. Such quietism is, of course, to a degree, understandable and not at all surprising. Millions of people lead lives in which despair and resignation have become ordinary. But it is also deeply pernicious. As Hannah Arendt observed, collective action – the expression of our care for the world – is possible only through thinking together, in ways that acknowledge both that the world is uncertain and that we have the power to change it. The polis, the “sphere of freedom”, as Arendt describes it in The Human Condition (Arendt 1958), “is not the city-state in its physical location; it is the organization of the people as it arises out of acting and speaking together, and its true space lies between people living together for this purpose, no matter where they happen to be” (ibid., p. 198). This is territory into which the ICFE report tentatively ventures in its attempt to reframe education in terms of “a public endeavour and a common good” (ICFE 2021, p. 2) and to redefine pedagogy “around the principles of cooperation, collaboration, and solidarity” (ibid., p. 4). In the spirit of Arendt, we need to ensure that pedagogical approaches make room for difference and dissent and promote challenge and critique, and that curricula are open, adaptable, and co-created, shaped by a recognition of education as a public good aimed at promoting human flourishing in the widest civic sense. In the preface to Between Past and Future: Eight Exercises in Political Thought, Arendt (2006 [1961]) describes the aliveness felt by members of the European resistance during World War II – their “treasure”, she calls it – as they stripped off their masks and began “to create that public space between themselves where freedom could appear” (ibid., p. 4). They became actors, agents of change, challengers and disruptors, prepared to take the initiative and act together in the world, freely, as equals, and without direction from others. This hopeful treasure was fleetingly held by participants of the Occupy movement and animates the pages of Isabelle Fremeaux and Jay Jordan’s brilliant and inspiring memoir of the occupation and defence of 4,000 acres of French wetlands – the ZAD or zone à defender – on which an international airport was to be built (Fremeaux and Jordan 2021), two exercises in direct action which had education at their heart. It is only natural that they should. As Arendt wrote, education is “the point at which we decide whether we love the world enough to assume responsibility for it and by the same token save it from that ruin which, except for renewal, except for the coming of the new and young, would be inevitable” (Arendt 2006 [1961], p. 193). Collective solutions are elusive, however; they cannot be imposed and their legitimacy rests on the nature and extent of public engagement in their creation, as ICFE (2021) recognised in calling for public debate around the creation of a social contract. Much more important than the vision are the means. In the end, the future of humanity depends not on the creation of a vision to which the countries of the world can sign up but, rather, on the degree to which people feel a future different to the present is possible and are willing and empowered, through education, “to assume joint responsibility for the world”, as Arendt put it (Arendt 2006 [1961], p. 186). Believing that change is possible is more important than being able to visualise what it will look like. Although the challenges of the time point to the necessity of different futures, most people experience the world as impervious to change. No matter how they might imagine their world as fairer, better or more just, there is, they recognise, really nothing they can do about it (and nothing that their politicians are willing or able to do either). Mainstream politics offers them vanishingly little in the way of hope. There was a moment, at the start of the COVID-19 pandemic, when it seemed that radical change might be possible. Much was said and much was promised. But it soon became apparent that promises to “build back better” really meant more of the same: more inequalities of wealth and opportunity, more strongarm authoritarian government, and more private-sector profiteering. Better for some, perhaps, but not for the many. Nevertheless, the solidarity and agency fostered by the crisis offered some hope, highlighting not only our inter-dependency but also our capacity for active civic engagement and public-spirited sacrifice. People were able to embody the kind of change they would like to see in society, a change characterised by collective endeavour and civic action. What we need, more than anything, and in every area of life, from school to the workplace to government and the media, is a radical infusion of this kind of democracy, accompanied by a willingness to lead with and for others and for the common good. And for that, we need to see education not as a commodity or a private investment but as a public good without which we would be unable to think our way to any sort of worthwhile future. The 10 articles in this double general issue of the International Review of Education amply demonstrate the public value of education and its wide, cross-sectoral benefits, exploring its challenges and possibilities, notably for those groups typically excluded or marginalised, while taking in an exceptionally wide range of thematic and geographical areas. The first article, “Literacy: A lever for citizenship?”, authored by Anna Robinson-Pant, considers the important relationship between literacy and citizenship, finding it to be more problematic than is usually believed. While literacy is often thought of as a prerequisite for citizenship, the author attempts to go beyond conventional framings in terms of functional skills for civic engagement and knowledge of rights – literacy for citizenship – and instead analyses different models of citizenship to discern ways in which literacy learning can emerge through active citizenship. She draws on ethnographic studies of literacy in everyday life to analyse the symbolic and instrumental meanings of literacy in specific contexts, introducing a social practice lens to literacy and citizenship, and going on to explore the pedagogical implications for literacy within citizenship education, particularly in relation to informal learning of “real literacies”, critical digital literacy, and literature as a way of entering someone else’s experiences. She shows how “the notion of multiple and multimodal literacies can help to broaden and deepen understanding of hierarchies of literacies and languages in relation to people’s multiple and changing identities”, arguing that this stance focuses our attention on intersectionality and diversity, rather than assuming people have one dominant identity that shapes their aspirations and rights as citizens. In concluding her article, Robinson-Pant contends that the role of literacy within citizenship education “is not only to provide skills for representation, documentation and accountability, but also to bring people closer together by sharing experiences, values, voices and aspirations and facilitating deeper interactions through written, oral and digital texts”. The second article in this issue also explores the intersection between education, agency, identity and diversity. “Translanguaging as bona fide practice in a multilingual South African science classroom”, written by Erasmos Charamba, responds to the call to improve students’ academic achievement in science education in a context of increasing cultural and linguistic diversity in the classroom through a focus on “translanguaging”, a pedagogical approach in which more than one language is utilised within a classroom lesson. While translanguaging is a relatively young field of research, interest in it has grown over the past decade. It has become clear that teaching children and young people solely in a language of instruction different to the one they speak at home inhibits their learning and leads them to attach less value to their home language and culture. The article explores the role language plays in the academic performance of multilingual students at a primary school in South Africa. Adopting an ethnographic approach, the author collected qualitative data through lesson observations video-recorded in a fifth-grade science class, supplementing these with several interviews with the teacher. Analysis of these data indicates the importance of translanguaging pedagogy for effective learning in multilingual classrooms. The use of multiple languages in this science classroom enabled multilingual students to engage in a practice of generating and creating scientific explanations in their own voice, resulting in better academic performance. Participants used their linguistic repertoire to clarify and review the scientific content, to construct rapport and to boost their participation in the lesson, while also increasing their proficiency in the language of institution. The next article, written by Boadi Agyekum, is likewise concerned with creating an enabling learning environment for students, this time with a focus on distance learning in higher education. The article, “Challenges of learning environments experienced by distance-learning higher education students in Ghana”, explores the challenges experienced by distance-learning higher education students in the Greater Accra region of Ghana. The author interviewed students in two University of Ghana distance-learning centres, where they attended weekend face-to-face sessions, asking them to share their experiences with respect to classrooms, learning facilities both inside and outside the classroom, and access to library support services. These data were supplemented with interviews with staff. The author’s findings revealed students’ struggle with poor infrastructure conditions, with most reporting lack of access to power sources in the classrooms, IT labs, library space, a student hub, and support services as barriers to experiencing meaningful higher education as distance learners. Students stressed the importance of infrastructural support and services tailored towards their needs as distance learners, with an emphasis also on students’ physical, social and psychological well-being. The article concludes by calling for greater attention to be paid to students’ “emotional learning environment”: “Learning centres need to factor in the role that emotions play in the process of teaching and learning, to provide the opportunity for students to talk about their feelings and concerns, and to provide the resources that will enable students to develop their emotional or mental skills through interaction and collaborative learning.” The fourth article considers distance learning from a teaching perspective. “An exploratory study to understand faculty members’ perceptions and challenges in online teaching”, by Tausif Mulla, Sufia Munir and Vivek Mohan, focuses on the implementation of online teaching and learning in the United Arab Emirates (UAE), where technology is one of the main pillars of the government’s vision of moving to a knowledge-based society. E-learning has become a popular method of delivery across higher education institutions in the UAE, a response, on the one hand, to globalisation and the demand for information technology infrastructure and, on the other, to the accelerant of the COVID-19 lockdowns. The authors conducted a systematic literature review, encompassing the period from 1999 to 2020. They found that while existing literature on online learning focuses predominantly on student-specific challenges, there is little published work covering faculty members’ specific challenges in facilitating online learning in the UAE. The second part of their study, based on semi-structured interviews with 15 faculty members, attempts to address this. It focuses on stakeholders’ reflections on designing and delivering online courses, analysing faculty members’ perspectives on online teaching and learning in the UAE. The authors identify a number of teacher-specific challenges, which they group into five themes: (1) learners’ expectations; (2) culture; (3) lack of incentives for faculty members to engage in online teaching; (4) pedagogy; and (5) technology. Understanding these challenges, the authors conclude, will help academic institutions to better support their staff and improve the delivery of their online programmes. The next article also considers the professional development (PD) support needs of teachers, this time in Quebec, Canada. “Effectiveness of professional development for teachers in French- and English-medium public elementary schools in Quebec, Canada: A first descriptive survey”, written by Marie-France Boulay, Christine Hamel and Sandra Hamel, contrasts “effective” professional development with “traditional” professional development, arguing that the former employs elements known to support changes in teaching practices and in student learning. Characteristics of effective PD include collective participation, adequate duration, active learning, and specific content focus, all within a coherent development process. Based on these characteristics, the authors conducted a survey to assess the PD provided to Quebec elementary (primary) schoolteachers and to identify the content, the learning modes emphasised, the reasons why teachers participated, the perceived benefits and the impacts of participation, as well as the incentives for and potential barriers to participation. They found that teachers rarely participate in PD unless it has first been specifically offered to them. Although Quebec teachers have access to a relatively wide range of PD activities, through in-school or out-of-school workshops, conferences, teacher networks, professional learning communities, university courses, etc., the authors identify a need for improvement in terms of the inclusion of characteristics that are recognised as effective in changing teaching practices and bringing about student learning gains. They recommend the development of mechanisms to support a richer and wider variety of professional development activities that meet a range of teacher and school system needs and, crucially, incorporate effectiveness characteristics. Such mechanisms might include ensuring that information on effectiveness characteristics is more widely distributed; encouraging a collaborative examination of teacher professional development needs; supporting the development of structured long-term professional development plans; and emphasising teachers’ professional autonomy. Our sixth article, “Akan folklore as a philosophical framework for education in Ghana”, by Samuel Amponsah, asks how Indigenous philosophies of education might enable Ghana to develop and promote lifelong education. The author argues that Ghana cannot pursue a lifelong education agenda by relying on education that is entirely centred on foreign cultural values and favouring Western educational philosophies to drive its educational policies and practices. Ghana, he contends, needs to incorporate more elements of an authentic Ghanaian framework and to strengthen the connection between education, culture and development. He thus analyses the educational strengths of African folklore from the Akan ethnic group of Ghana. He concludes that aspects of Akan folklore, including its stories and proverbs, its kinship rights and rules, its moral codes, its corporate and humanistic perspective, present a viable alternative and complement to the country’s current westernised education. The author proposes an enhanced Ghanaian framework for education informed by Akan philosophy and pedagogy. This, he argues, will be beneficial in promoting quality and lifelong education in the country while enabling ordinary Ghanaians to make their voices heard. The next article concerns gender equality and focuses on education in Ukraine. “A pedagogy of freedom as a viable basis for implementing gender equality in Ukraine’s educational institutions”, written by Alla Rastrygina and Nadiya Ivanenko, reflects on the extent to which Ukraine has prioritised gender issues in education since the 1990s. In the three decades prior to the Russian invasion, independent Ukraine’s efforts to integrate into the European community led it to engage in efforts to restructure its educational institutions and processes on the basis of democratic principles free from any form of discrimination, including gender-related discrimination. These efforts have been promoted through joint projects with UN Women and other international organisations, and gender equality improvement strategies are now reflected in Ukrainian legislation, though they are not yet fully implemented on the ground. The authors offer an analysis of the current state of gender equality in the Ukrainian system of education before presenting their own concept of the pedagogy of freedom as a viable basis for achieving gender equality in Ukraine’s educational institutions. Analysing literature devoted to the problem of freedom and gender equality in educational policy, they argue that learners’ free self-determination, self-development and self-realisation can only be effective factors in realising gender equality if pedagogical activity and learning spaces are designed to support the development of learners’ full potential. The eighth article of this double issue, “A tripartite understanding of experiences of young apprentices: A case study of the London Borough of Hounslow” by Priscilla Hansberry and Trevor Gerhardt, takes as its starting point a pledge, made by the London Borough of Hounslow in 2019, to create 4,000 new apprenticeships and training opportunities to help young people into work. The article investigates the experiences of young apprentices in Hounslow before and during the COVID-19 pandemic. Exploring the perspectives of two apprentices, two employers and one training provider in a small-scale qualitative study, the authors identify the factors that hinder and support entry into and sustainability of apprenticeships, and progression towards professional employment. Their findings show that labour market entry was significantly hindered by competition (notably with peers who had better maths and English qualifications, for a small number of apprenticeships) and organisational barriers (such as managers with prejudices against young people and the still-prevalent stigmatising of apprentices and apprenticeships). Supportive factors identified include maths and English qualifications (critical in entry and sustainability, as well as in progression to work), personal characteristics (such as a positive mindset, enabling young people to persevere despite a disadvantaged socioeconomic background and lack of family support, for example) and supportive relationships (e.g., mentoring) between apprentices and their training providers or employers. The authors conclude their article by highlighting maths and English qualifications and a supportive environment as the key factors conducive to successful apprenticeship management. The final two articles in the issue concern the experiences of refugees and migrants in accessing education. “North Korean women entrepreneurs learning from failure”, written by Jinhee Choi and Esther Prins, focuses on how migrants learn from failure and how this shapes their social adjustment. The study examines, in particular, how North Korean migrants struggling for a foothold in South Korea sought to learn from failures in their workplaces and everyday life. The article draws on nine months of ethnographic research in South Korean restaurants and cafés employing North Korean migrants. Data sources include informal conversations and loosely structured interviews with five women who started, or planned to start, their own enterprise. The findings reveal that these migrants perceived failure in five interrelated spheres: financial, relational, physical, psychological and professional. Participants developed perspectives to understand failure as an integral part of learning in a new society and adopting unfamiliar role expectations and responsibilities. They also applied knowledge they had acquired through their failures to change their approach to their career and to strengthen their personal and business capacity to obtain a legitimate social position. Paradoxically, failures that were beyond their control, such as legal problems, created opportunities to receive practical support from, and increase trust in, South Koreans. The article challenges the “pervasive view of migrants as incompetent, inferior workers, and a social burden”, and shows how the predominant focus on migrants’ acquisition of language, literacy and workplace skills ignores “their invisible learning about the self, others, and their host society through new social relations and practices such as opening a business and building trust with host citizens”. The study offers a more nuanced perspective, demonstrating how migrants learn from failure – inevitable for anyone in this situation – and use these experiences as learning opportunities to transform themselves into active citizens able to contribute to their host society. The final article of this issue – “Interventions to improve refugee children’s access to education and quality learning: A scoping review of existing impact evaluations” by Júlia Palik and Gudrun Østby – addresses the challenges refugee children face in accessing quality education. These challenges are widely recognised, and numerous interventions have been promoted to address them. What is still lacking, however, is systematic evidence on what works to improve refugee children’s enrolment and learning. The authors of this article set out to find what robust quantitative evidence exists regarding interventions that seek to improve access to education and quality learning for refugee children. They conducted a scoping review of quantitative peer-reviewed articles evaluating the effect of interventions which aimed to improve access to education and/or quality learning for refugee children. While their literature search for the time-period 1990–2021 resulted in 1,873 articles, only eight of these met the authors’ selection criteria. This low number indicates the general lack of robust evidence as to what works to improve quality learning for refugee children. What evidence there is suggests that cash-transfer programmes can increase school attendance and that learning outcomes, such as second-language acquisition, can be improved through physical education, early childhood development programmes, or online game-based solutions. Other interventions, such as drama workshops, appear to have had no effect on second-language acquisition. The authors conclude their article by calling for more evaluation of educational interventions in an area where rigorous quantitative evidence is often scarce and inconclusive. These last articles demonstrate the limitations of our knowledge of what works in certain key areas of policy in education and lifelong learning and in the context, in particular, of the educational needs of excluded, disadvantaged or otherwise discriminated-against groups. Understanding this is crucial if we are to do what the UN urges us to and prepare educational governance and institutions for “sustainable and inclusive transformation”, while prioritising policies and investments with “multiplier effects across the goals”. There is a crucial intersectoral role in this for education, but it demands not only more and better evidence but also, as the UN indicates, a “surge” in financing and “an enabling global environment for developing countries” (UN 2023, p. 26). Getting there from where we are now seems almost impossibly difficult, but it is the minimum requirement if we are to “break through to a better future for all”, as the UN says we must (ibid.). As far as education goes, the main means of transformation is the “new social contract for education” (ICFE 2021). But we are left largely in the dark about the mechanisms through which this can be created or how we might challenge and overcome the forces responsible for current and past injustices. Of course, it is not the job of international organisations to do this. But the failure of nation-states to live up to their commitments or even to offer a roadmap as to how, at some future point, they might achieve this, is creating a gap in credibility and eroding people’s confidence in both politics and future-making. I do not see that there is much chance of things changing, at least not in any substantive way. The centrist realism embraced by national politicians around the world will, I suspect, be looked back on by later generations as an extraordinary and extreme form of denial, as well as an appalling and unprecedented dereliction of responsibility to the future. The old stories of possibility are, for now, all we have. What hope there is lies not in convincing nation-states to do better, but in generating genuine public engagement in these issues and reviving spaces in which, as Hannah Arendt put it, freedom can appear. Education, reframed in terms of civic engagement and critical thinking for future-oriented collective action and social solidarity, has an important role to play in this, though it can do little alone, and we should not waste our time waiting for those with the least to gain from the transformation of education to facilitate it. 1 See, for example, Beckert (2020) and Robertson (2022). 2 Discussed in Elfert, M. & Draxler, A. (2022). Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Arendt H The human condition 1958 Chicago, IL University of Chicago Press Arendt, H. (2006 [1961]). Between past and future: Eight exercises in political thought. New York: Penguin Books. Beckert J The exhausted futures of neoliberalism: From promissory legitimacy to social anomy Journal of Cultural Economy 2020 13 3 318 330 10.1080/17530350.2019.1574867 Cusk R Outline 2014 London Faber & Faber Economist (2018). As inequality grows, so does the political influence of the rich. The Economist, 21 July [online article]. Retrieved 22 May 2023 from https://www.economist.com/finance-and-economics/2018/07/21/as-inequality-grows-so-does-the-political-influence-of-the-rich Elfert M Draxler A The Faure report: 50 years on – editorial introduction International Review of Education 2022 68 5 637 654 10.1007/s11159-022-09981-7 Fremeaux I Jordan J We are “nature” defending itself: Entangling art, activism and autonomous zones 2021 London Pluto Press GEMR Team (Global Education Monitoring Report Team) (2023). Can countries afford their national SDG 4 benchmarks? Policy Paper 49, April 2023. Paris: UNESCO. Retrieved 22 May 2023 from https://unesdoc.unesco.org/ark:/48223/pf0000385004 ICFE (International Commission on the Futures of Education) (2021). Reimagining our futures together: A new social contract for education. Paris: UNESCO. Retrieved 22 May 2023 from https://unesdoc.unesco.org/ark:/48223/pf0000379707 Robertson S Guardians of the future: International organisations, anticipatory governance and education Global Society 2022 36 2 188 205 10.1080/13600826.2021.2021151 Smith, E. (2023). Former UK PM Liz Truss is blaming the left-wing “economic establishment” for ousting her. CNBC, 6 Feb [online article]. Retrieved 22 May 2023 from https://www.cnbc.com/2023/02/06/former-uk-pm-liz-truss-is-blaming-left-wing-economic-establishment-for-ousting-her.html UN (United Nations) (2015). The 17 goals [online resource]. Retrieved 22 May 2023 from https://sdgs.un.org/goals UN (2021). Our common agenda: Report of the Secretary-General. New York: United Nations. Retrieved 22 May 2023 from https://www.un.org/en/content/common-agenda-report/assets/pdf/Common_Agenda_Report_English.pdf UN (2022a). Report on the 2022 Transforming Education Summit. New York: United Nations. Retrieved 22 May 2023 from https://www.un.org/sites/un2.un.org/files/report_on_the_2022_transforming_education_summit.pdf UN (2022b). Transforming education: An urgent political imperative for our collective future. Vision Statement of the Secretary-General on Transforming Education. New York: United Nations. Retrieved 22 May 2023 from https://www.un.org/sites/un2.un.org/files/2022/09/sg_vision_statement_on_transforming_education.pdf UN (2023). Progress towards the Sustainable Development Goals: Towards a rescue plan for people and planet. Report of the Secretary-General (special edition, May 2023). New York: United Nations General Assembly Economic and Social Council. Retrieved 22 May 2023 from https://sdgs.un.org/sites/default/files/2023-04/SDG_Progress_Report_Special_Edition_2023_ADVANCE_UNEDITED_VERSION.pdf
PMC010xxxxxx/PMC10240294.txt	==== Front Clin Nurs Res Clin Nurs Res CNR spcnr Clinical Nursing Research 1054-7738 1552-3799 SAGE Publications Sage CA: Los Angeles, CA 37264854 10.1177/10547738231177395 10.1177_10547738231177395 Research Articles Clinical and Socio-Demographic Variables Associated With Long COVID-19: A Cross-Sectional Study Mahmoodi Zohreh 1 Bahrami Giti 1 Shahrestanaki Ehsan 2 Seddighi Hamed 345 https://orcid.org/0000-0003-1511-3772 Ghavidel Nooshin 1 1 Social Determinants of Health Research Center, Alborz University of Medical Sciences, Karaj, Iran 2 Non-communicable Diseases Research Center, Alborz University of Medical Sciences, Karaj, Iran 3 Department of Clinical Psychology and Experimental Psychopathology, University of Groningen, Groningen, Netherlands 4 Child & Family Welfare Unit, University of Groningen, Groningen, the Netherlands 5 Department of Health in Disaster and Emergencies, Faculty of Public Health, Shahid Sadoughi University of Medical Sciences, Yazd, Iran Nooshin Ghavidel, Social Determinants of Health Research Center, Alborz University of Medical Sciences, Karaj, 3198764653, Iran. Email: ngh_med50@yahoo.com 2 6 2023 7 2023 2 6 2023 32 6 947953 © The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Considering the importance of long COVID-19 (LC), this study aimed to investigate the relationship between clinical/sociodemographic factors and LC symptoms (LCS). This online cross-sectional study was conducted on 308 people infected with COVID-19 in Alborz, Iran, from April 1 to June 1, 2022. Multivariable logistic regression models were applied to measure the association between the LCS with other variables. Overall, 76.6% of participants had at least one LCS. Results from the multivariate logistic regression analysis showed that females (crude odds ratio [OR] [95% confidence interval (CI)]: 2.725 [1.42, 5.22]), educated persons (3.747 [1.58, 8.84]), people with a higher number of COVID-19 reinfection (2.280 [1.30, 3.97]), having an underlying disease (1.996 [1.01, 3.93]), and COVID-19 severity (3.321 [1.037, 10.635]) had higher odds of LC than others (all p < .05). Study findings provide additional clinical/sociodemographic data on risk for LC. These data may inform future research and clinical practice for potential risk identification and early intervention. COVID-19 long-term symptoms social determinants of health socioeconomic status typesetterts1 ==== Body pmcIntroduction During 2 years of the COVID-19 worldwide pandemic, scientific researchers have focused on long-term morbidity from COVID-19. Although the scientific community’s understanding of long COVID-19 (LC) is still evolving, and understanding of the diagnoses, phenotypes, and epidemiology is nascent (Bell et al., 2021), LC’s health impacts and symptoms have been assessed by a broad range of studies (Aiyegbusi et al., 2021; González-Islas et al., 2022; Nalbandian et al., 2021). It is suggested that more than 20% of people positive for COVID-19 develop post-COVID-19 syndrome (Iqbal et al., 2021). According to the National Institute for Health and Care Excellence (NICE) guidance, the average recovery time from COVID-19 is 2 to 3 weeks and the long-term effects of COVID-19 are often described as LC. Clinical definitions for COVID-19 at different times include acute COVID-19: signs and symptoms of COVID-19 up to 4 weeks, ongoing symptomatic COVID-19: signs and symptoms of COVID-19 from 4 to 12 weeks; and post-COVID-19 syndrome: signs and symptoms of COVID-19 infection that persist for more than 12 weeks and are not explained by an alternative diagnosis (Shah et al., 2021). The term “long COVID” is commonly used to describe signs and symptoms that continue or develop after acute COVID-19. It includes both ongoing symptomatic COVID-19 (from 4 to 12 weeks) and post-COVID-19 syndrome (12 weeks or more) (NICE, 2021). Evidence has shown that the most prominent LC symptoms (LCS) were fatigue, dyspnea, muscle/joint pain, headache, cough, chest pain, altered smell/taste, hair loss, cognitive/mental impairments, and diarrhea (Desai et al., 2022; Yong, 2021). Although the mechanisms of long-term complications of COVID-19 infection remain unknown, several pathophysiological mechanisms may be responsible, including direct cellular infection, endothelial damage, immune dysregulation, and coagulation (Desai et al., 2022). Literature has indicated that risk factors of LC may include female sex, more early symptoms, early dyspnea, specific biomarkers (e.g., D-dimer, C-reactive protein, and lymphocyte count), and prior psychiatric disorders (Yong, 2021). Previous studies have shown a higher risk of developing LC with increasing age, female sex, symptom burden, hospitalization during acute COVID-19, presence of comorbidities, smoking, obesity, socioeconomic deprivation, and belonging to an ethnic minority (Carvalho-Schneider et al., 2021; Emecen et al., 2023; Jacobs et al., 2020; Subramanian et al., 2022; Sudre et al., 2021). In addition, a study has indicated that the female sex is a worse predictor of LC, even if the disease is less severe during that phase (García et al., 2022). A prospective cohort study from Iran has shown a significant association between LC and some factors, including age, underlying diseases, acute phase symptoms, weight loss, and history of skin sequel in the acute phase (Larijani et al., 2022). There is a need for further research into risk factors and socioeconomic determinants related to developing LC for evidence-based service planning and prevention mechanisms, which the NICE guideline has highlighted as a research priority (Thompson et al., 2022). One of the most important things that need more studies in this field is the relationship between LCS and clinical/sociodemographic factors. Therefore, this study aimed to investigate the relationship between these factors and the LCS. Methods Design and Participants We conducted an online cross-sectional study to explore the association between long symptoms of COVID-19 and clinical/sociodemographic factors in Alborz, Iran, from April 1 to June 1, 2022. We used an online survey to collect data from the general population infected with COVID-19 since the beginning of the outbreak of COVID-19 with a positive polymerase chain reaction test, and at least 4 weeks have passed since the onset of their disease. The exclusion criterion was the age <15 and unwillingness to participate in the study. Using convenience sampling, we implemented the survey by posting a link in various social groups through WhatsApp and emails from the general population. A total of 308 participants completed the online survey. This study protocol was approved by the Ethics Committee of Alborz University of Medical Sciences (IR.ABZUMS.REC.1400.335). Respondent’s decision to complete the survey implied consent to participate. At the beginning of the study, respondents were informed that the survey was confidential and anonymous, and their participation was voluntary. They also announced that the survey results would be used for presentations or publications. Data Collection Data were completed using a sociodemographic checklist, COVID-19 disease checklist, and socioeconomic status scale (SES). Sociodemographic checklist included questions about age, gender, internationality, education level, marital status, occupation, income, living area, and number of family members, smoking, body mass index (BMI), and the history of underlying disease. COVID-19 disease-related checklist included the onset of the disease, the number of COVID-19 reinfections in a person, the number of COVID-19 vaccines, and the severity of COVID-19 disease (outpatient–hospital admission–intensive care unit [ICU] admission), LCS, and duration of LCS. LCS is defined as the symptoms of COVID-19 that last ≥4 weeks (NICE, 2021). LCS in the checklist included cough, dyspnea, fatigue and weakness, headache, hyposmia, hypogeusia, hair loss, sleep disorder, depression, anxiety, memory loss, attention disorder, intolerance to exercise, digestive system problems, chest pain, irregular menstruation, joint pain, and cutaneous signs. SES consisted of six questions, including education, income, economic class, and housing status, which are scored based on a Likert-type scale from 1 to 5, and a total score ranging from 6 to 30. Validity and reliability have been performed in Iran (Eslami et al., 2014). Statistical Analysis We used descriptive statistics to analyze survey data using SPSS IBM software, version 20. The demographic characteristics of the study population were summarized using descriptive statistics, mean, standard deviation (SD) for continuous variables, and frequency (%) for categorical variables. Univariable and multivariable logistic regression models were applied to measure the association between LC with other variables. Results are presented as crude and adjusted odds ratios (ORs) and 95% confidence intervals (CIs). A value of less than 0.05 was considered statistically significant. Results A total of 308 people participated in this survey. The majority of participants were female (61%), married (77.6%), bachelor (37.0%), and employed (76.6%). The mean ± SD age of participants was 42.56 ± 11.59 years ranging from 15 to 78 years. The mean ± SD of BMI was 26.54 ± 4.17. Overall, 36.7% of people had a history of underlying disease. Most were hypothyroidism 7.5%, hypertension 6.8%, diabetes 5.2%, and hyperlipidemia 5.2%, respectively. The sociodemographic characteristics of the subjects are shown in Table 1. Table 1. Sociodemographic Characteristics of Subjects. Variables No. (%) Sex Male 120 (39) Female 188 (61) Educational level ≤Associate degree 88 (28.6) Bachelor 114 (37.0) <Bachelor 106 (34.4) Occupational status Un-unemployed 72 (23.4) Employed 236 (76.6) Income, Rial ≤100,000,000 173 (56.2) >100,000,000 135 (43.8) History of underlying disease Yes 113 (36.7) Marital status Single 69 (22.4) Married 239 (77.6) Smoking Yes 43 (14.0) Severity of COVID-19 disease Outpatients 269 (87.3) Hospital admission 37 (12.0) ICU admission 2 (0.6) Long-COVID symptoms Yes* 236 (76.6) Note. ICU = Intensive Care Unit. * Yes (long-term symptoms) is a symptom of COVID-19 lasting ≥4 weeks. 76.6% of participants had at least one LCS. The most LCS were fatigue and weakness (43.8%), hair loss (28.2%), and cough (23.7%), respectively. The frequency of LCS is shown in Figure 1. Figure 1. The frequency of long COVID-19 symptoms. The mean (±SD) of the SES score was 14.59 (±3.108). After logistic regression analysis, no significant differences existed between the SES score and LCS of COVID-19 disease (p = .061). Results from the univariate logistic regression analysis showed that females (crude OR [95% CI]: 4.198 [2.40, 7.33], persons who were in bachelor/≤associate degree (3.066 [1.62, 5.79]), >bachelor/≤associate degree (4.20 [2.10, 8.39]), people with a higher number of COVID-19 reinfection during pandemic (2.816 [1.65, 4.80]), smokers (0.452 [0.22, 0.89]), and having the underlying disease (2.033 [1.01, 4.06]) had higher odds of LCS than others (all p < .05). After multivariate logistic regression analysis, females (crude OR [95% CI]: 2.725 [1.42, 5.22]), persons who were in education >bachelor/≤associate degree (3.747 [1.58, 8.84]), people with a higher number of COVID-19 reinfection during pandemic (2.280 [1.30, 3.97]), having the underlying disease (1.996 [1.01, 3.93]), and more severe COVID-19 (3.321 [1.03, 10.63]) had higher odds of LCS than others (all p < .05) (Table 2). Table 2. Association Between Variables and Long COVID-19 Symptoms: Logistic Regression Analysis. Variables Univariate logistic regression Multivariate logistic regression Crude OR [95% CI] p Value Adjusted# OR [95% CI] p Value Sex Male 1 1 Female 4.198 [2.402, 7.336] .000* 2.725 [1.422, 5.221] .003 Age 0.987 [0.965, 1.01] .543 — — Living area Urban 1 1 Rural 0.451 [0.074, 2.751] .388 — — BMI 0.971 [0.912, 1.033] .347 — — Occupation Un-unemployed 1 1 Employed 0.741 [0.385, 1.425] .369 — — Education ≤Associate degree 1 1 Bachelor 3.066 [1.623, 5.793] .001 2.089 [0.991, 4.405] .053 <Bachelor 4.20 [2.102, 8.391] .000* 3.747 [1.587, 8.844] .003 Marital status Single 1 1 1 Married 1.468 [0.802, 2.685] .213 — — Family number 0.969 [0.778, 1.208] .782 — — Income ≤100,000,000 1 1 >100,000,000 1.042 [0.612, 1.775] .880 — — SES 1.085 [0.996, 1.181] .061 0.972 [0.876, 1.079] .595 The number of COVID-19 reinfections 2.816 [1.650, 4.806] .000* 2.280 [1.307, 3.977] . 004** Smoking No 1 1 Yes 0.452 [0.228, 0.897] .023* 0.990 [0.434, 2.258] .981 Number of COVID-19 vaccine 0.923 [0.600, 1.421] .717 — — Severity of COVID-19 No hospitalization 1 1 Hospitalization 2.861 [0.980, 8.357] .055 3.321 [1.037, 10.635] .043* Underlying disease No 1 1 Yes 2.442 [1.323, 4.509] .004** 1.996 [1.013, 3.933] .046* Note. Adjusted# for all variables with p < .2 in the univariate model. OR = odds ratio; CI = confidence interval; BMI = body mass index; SES = socioeconomic status scale. * p < .05. p < .01. *p < .001. Discussion In the current study, we evaluated the long-term symptoms of COVID-19 and its association with clinical/sociodemographic factors. In our study, 76.6% of participants had at least one LCS. The most LCS were fatigue and weakness (43.8%), hair loss (28.2%), cough (23.7%), and hyposmia (18.5%), respectively. In line with our study, a meta-analysis study demonstrated that 80% of the infected patients with COVID-19 had one or more LCS. The most common symptoms were fatigue (58%), headache (44%), attention disorder (27%), hair loss (25%), dyspnea (24%), and anosmia (21%) (Lopez-Leon et al., 2021). Some LCS were different from our study, which may be due to differences in the study populations, methods, follow-up periods, and sample size. Another study revealed that fatigue, respiratory, and cognitive problems occurred in 51.0, 60.4, and 35.4% of LC cases, respectively (Hanson et al., 2022). In addition, a study in the United Kingdom showed 86% of patients had at least one residual symptom at follow-up. The most long-term symptoms were anxiety, fatigue, and myalgia (Sykes et al., 2021). Results from the multivariate logistic regression analysis showed that females, persons with higher-level education, people with a higher number of COVID-19 reinfection during the pandemic, persons with hospitalization due to COVID-19 disease and having underlying condition had higher odds of LCSs than others. In line with our study, Fernandez-de-Las-Penas found that women are more likely to develop symptoms of COVID-19 8 months after contracting the virus than men (Fernández-de-Las-Peñas et al., 2022). In addition, many studies have reported gender as a risk factor associated with post-COVID complications and greater stress after the disease (Iqbal et al., 2021; Mahmoodi et al., 2021). Another study revealed that gender predicts COVID-19 complications such as dyspnea, fatigue, chest pain, and palpitation (Pelà et al., 2022; Yong, 2021). Knowing the different impacts of COVID-19 on men and women is a crucial step to understanding the pathophysiology and its consequences better to promote healthcare interventions. One of the leading causes of these differences is the biological and immunological disparity in the expression of angiotensin-converting enzyme receptors and interleukin-6 production between women and men (Fernández-de-Las-Peñas et al., 2022; Ortona et al., 2021). According to the present study’s results, education level was one of the predictors of LCS after COVID-19. Education is one of the essential socioeconomic factors. Most studies in infectious and non-communicable diseases have suggested the relationship between this factor and health outcomes and have suggested education as an independent factor for long-term complications of infectious diseases (Gupta et al., 2022). Other studies have shown that illiterate or low-educated people have more complications and deaths due to corona than other groups. Delay submitting reports and referring to medical centers is one of the leading causes of complications and deaths in these people. However, the relationship between low socioeconomic status with complications and death from COVID-19 is not specific (Clouston et al., 2021; Sharma et al., 2022). The recent difference in our results may be due to the more reporting and follow-up of people with a higher level of education compared to other groups. In a study, Darvishpour et al. (2016) found that the level of education has a significant relationship with the level of health literacy, and patients with sufficient health literacy seek treatment and control their disease symptoms more than others, and as a result, are more successful (Darvishpour et al., 2016). In our study, having underlying diseases and the number and severity of COVID infections in a person were one of the factors related to LCS. This finding is consistent with a retrospective cohort study by Ayoubkhani et al. (2021) conducted on COVID-19 patients. They found that the mortality and complications of the COVID-19 disease were higher in patients with a history of hospitalization than in others. However, the available evidence shows many changes in the estimates of the prevalence, incidence and influential factors of the LC due to the differences in the studied populations, survey methods, follow-up periods, and sample size; therefore, conducting more studies in this field is essential. Limitations Our study had some limitations: first, asking questions online about disease symptoms will have a high recall bias, which is one of the limitations of our research. Second, because the questionnaire was online, people with a higher socioeconomic level may have more access to participate in the study. Third, Most participants in our study were in the middle age period because older adults do not have enough knowledge of modern communication tools. Fourth, the number of samples in our study who were hospitalized, especially in the ICU, was minimal; further studies must be performed. Conclusion This study showed that in addition to the severity and number of reinfections with COVID-19 in a person, some sociodemographic factors, including gender, education, and history of underlying disease, are related to the LC. Therefore, to reduce the disease burden of this pandemic, focusing on these factors can effectively reduce the risks LC. In national policies, governments are advised to be active in other public sectors, such as the Ministry of Education, Ministry of Welfare, and other organizations alongside health organizations. In addition, the private sector and non-governmental organizations should also participate in designing and implementing policies and interventions. Author Biographies Zohreh Mahmoodi, associate professor, PhD in social determinants of health, Social Determinants of Health Research Center, Alborz University of Medical Sciences. Giti Bahrami, assistant professor, PhD in health and social welfare, Social Determinants of Health Research Center, Alborz University of Medical Sciences. Ehsan Shahrestanaki, candidate for epidemiology, researcher in the noncommunication research center, Alborz University of Medical Science. Hamed Seddighi, PhD, postdoctoral researcher, department of clinical psychology and experimental psychopathology, University of Groningen, Groningen, The Netherlands. Nooshin Ghavidel, MD and PhD in neuroscience, Social Determinants of Health Research Center, Alborz University of Medical Sciences. Author Contributions: NG was responsible for the acquisition of data. ZM and NG analyzed and interpreted the patient data. All authors contributed to the concept, design, drafting, and revising the article. All authors critically reviewed content and approved the final version for publication. Availability of Data and Materials: The datasets used during the current study are available from the corresponding author upon reasonable request. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. Ethical Approval and Consent: This study was performed by the latest version of the Declaration of Helsinki and with the approval of the research ethics committee of Alborz University of Medical Sciences (IR.ABZUMS.REC.1400.335). Respondent’s decision to complete the survey implied consent to participate. At the beginning of the study, respondents were informed that the survey was confidential and anonymous, and their participation was voluntary. They also reported that the survey results would be used for presentations or publications. Consent to Publish: Informed consent was obtained from all subjects who participated. ORCID iD: Nooshin Ghavidel https://orcid.org/0000-0003-1511-3772 ==== Refs References Aiyegbusi O. L. Hughes S. E. Turner G. Rivera S. C. McMullan C. Chandan J. S. Haroon S. Price G. Davies E. H. Nirantharakumar K. Sapey E. Calvert M. J. ; TLC Study Group. (2021). Symptoms, complications and management of long COVID: A review. Journal of the Royal Society of Medicine, 114 (9 ), 428–442. 10.1177/01410768211032850 34265229 Ayoubkhani D. Khunti K. Nafilyan V. Maddox T. Humberstone B. Diamond I. Banerjee A. (2021). Post-covid syndrome in individuals admitted to hospital with covid-19: Retrospective cohort study. BMJ, 372 , n693. Bell M. L. Catalfamo C. J. Farland L. V. Ernst K. C. Jacobs E. T. Klimentidis Y. C. Jehn M. Pogreba-Brown K. (2021). Post-acute sequelae of COVID-19 in a non-hospitalized cohort: Results from the Arizona CoVHORT. PLoS One, 16 (8 ), e0254347. 10.1371/journal.pone.0254347 Carvalho-Schneider C. Laurent E. Lemaignen A. Beaufils E. Bourbao-Tournois C. Laribi S. Flament T. Ferreira-Maldent N. Bruyère F. Stefic K. Gaudy-Graffin C. Grammatico-Guillon L. Bernard L. (2021). Follow-up of adults with noncritical COVID-19 two months after symptom onset. Clinical Microbiology and Infection, 27 (2 ), 258–263.33031948 Clouston S. A. Natale G. Link B. G. (2021). Socioeconomic inequalities in the spread of coronavirus-19 in the United States: A examination of the emergence of social inequalities. Social Science & Medicine, 268 , 113554.33308911 Darvishpour J. Omidi S. Farmanbar R. J. C. (2016). The relationship between health literacy and hypertension treatment control and follow-up. Caspian Journal of Health Research, 2 (1 ), 1–8. Desai A. D. Lavelle M. Boursiquot B. C. Wan E. Y. (2022). Long-term complications of COVID-19. American Journal of Physiology-Cell Physiology, 322 (1 ), C1–C11. 10.1152/ajpcell.00375.2021 Emecen A. N. Keskin S. Turunc O. Suner A. F. Siyve N. Basoglu Sensoy E. Unal B. (2023). The presence of symptoms within 6 months after COVID-19: A single-center longitudinal study. Irish Journal of Medical Science (1971-), 192 (2 ), 741–750. 10.1007/s11845-022-03072-0 Eslami A. Mahmoudi A. Khabiri M. Najafiyan Razavi S. M. (2014). The role of socioeconomic conditions in the citizens’ motivation for participating in public sports. Applied Research in Sport Management, 2 (3 ), 89–104. Fernández-de-Las-Peñas C. Martín-Guerrero J. D. Pellicer-Valero Ó. J. Navarro-Pardo E. Gómez-Mayordomo V. Cuadrado M. L. José A. Arias-Navalón J. A. Cigarán-Méndez M. Hernández-Barrera V. Arendt-Nielsen L. (2022). Female sex is a risk factor associated with long-term post-COVID related-symptoms but not with COVID-19 symptoms: The LONG-COVID-EXP-CM multicenter study. Journal of Clinical Medicine, 11 (2 ), 413.35054108 García B. A. Lema E. R. Rabuñal Rey R. R. (2022). Health perception among female COVID-19 patients. Comment on Fernández-de-las-Peñas et al. Female sex is a risk factor associated with long-term post-COVID related-symptoms but not with COVID-19 symptoms: The LONG-COVID-EXP-CM multicenter study. Journal of Clinical Medicine, 11 , 413. 10.3390/jcm11112999 35054108 González-Islas D. Sánchez-Moreno C. Orea-Tejeda A. Hernández-López S. Salgado-Fernández F. Keirns-Davis C. Galicia-Amor S. Trejo-Mellado E. Gochicoa-Rangel L. Castorena-Maldonado A. (2022). Body composition and risk factors associated with sarcopenia in post-COVID patients after moderate or severe COVID-19 infections. BMC Pulmonary Medicine, 22 (1 ), 223. 10.1186/s12890-022-02014-x 35676657 Gupta R. Joseph P. Rosengren A. Yusuf S. J. (2022). Location and level of care, education, availability of medicines and cardiovascular mortality. In Fuster V. N. J. Vaishnava P. Leon M. B. Callans D. J. Rumsfeld J. S. Poppas A. (Eds.), Fuster & hurst’s: The heart (511–526). McGraw Hill. Hanson S. W. Abbafati C. Aerts J. G. Al-Aly Z. Ashbaugh C. Ballouz T. Blyuss O. Bobkova P. Bonsel G. Borzakova S. Buonsenso D. Butnaru D. Carter A. Chu H. De Rose C Diab M. M. Ekbom E. El Tantawi M. Fomin V. . . .Vos T. (2022). A global systematic analysis of the occurrence, severity, and recovery pattern of long COVID in 2020 and 2021. medRxiv. Iqbal F. M. Lam K. Sounderajah V. Clarke J. M. Ashrafian H. Darzi A. J. E. (2021). Characteristics and predictors of acute and chronic post-COVID syndrome: A systematic review and meta-analysis. EClinicalMedicine, 36 , 100899.34036253 Jacobs L. G. Gourna Paleoudis E. Lesky-Di Bari D. Nyirenda T. Friedman T. Gupta A. Rasouli L. Zetkulic M. Balani B. Ogedegbe C. Bawa H. Berrol L. Qureshi N. Aschner J. L. (2020). Persistence of symptoms and quality of life at 35 days after hospitalization for COVID-19 infection. PLoS One, 15 (12 ), e0243882. Larijani M. S. Ashrafian F. Amiri F. B. Banifazl M. Bavand A. Karami A. Shokooh F. A. Ramezani A. (2022). Characterization of long COVID-19 manifestations and its associated factors: A prospective cohort study from Iran. Microbial Pathogenesis, 169 , 105618. 10.1016/j.micpath.2022.105618 35690233 Lopez-Leon S. Wegman-Ostrosky T. Perelman C. Sepulveda R. Rebolledo P. A. Cuapio A. Villapol S. (2021). More than 50 long-term effects of COVID-19: A systematic review and meta-analysis. Scientific Reports, 11 (1 ), 1–12.33414495 Mahmoodi Z. Koshesh K. J. Almasi G. Pournajaf S. Rastad H. Ghavidel N. (2021). Association between structural determinants of health and posttraumatic stress symptoms during the COVID-19 pandemic. Current Psychiatry Research and Reviews Formerly: Current Psychiatry Reviews, 17 (4 ), 270–279. Nalbandian A. Sehgal K. Gupta A. Madhavan M. V. McGroder C. Stevens J. S. (2021). Post-acute COVID-19 syndrome. Nature Medicine, 27 (4 ), 601–615. 10.1038/s41591-021-01283-z NICE. (2021). COVID-19 rapid guideline: managing the long-term effects of COVID-19. National Institute for Health and Care Excellence, London, UK. Retrieved November 11, 2021 from https://www.nice.org.uk/guidance/ng188/resources/covid19-rapid-guideline-managing-the-longterm-effects-of-covid19-pdf-51035515742 Ortona E. Buonsenso D. Carfi A. Malorni W. J. C. D. D. (2021). Long COVID: An estrogen-associated autoimmune disease? Cell Death Discovery, 7 (1 ), 1–3. Pelà G. Goldoni M. Solinas E. Cavalli C. Tagliaferri S. Ranzieri S. Frizzelli A. Marchi L. Mori P. A. Majori M. Aiello M. Corradi M. Chetta A. (2022). Sex-related differences in long-COVID-19 syndrome. Journal of Women’s Health, 31 (5 ), 620–630. Shah W. Hillman T. Playford E. D. Hishmeh L. (2021). Managing the long term effects of covid-19: Summary of NICE, SIGN, and RCGP rapid guideline. BMJ, 372 , n136. Sharma A. K. Gupta R. Baig V. N. Singh V. T. Chakraborty S. Sunda J. P. Dhakar P. Sharma S. P. Panwar R. B. Katoch V. M. (2022). Educational status and COVID-19 related outcomes in India: Hospital-based cross-sectional study. BMJ Open, 12 (2 ), e055403. Subramanian A. Nirantharakumar K. Hughes S. Myles P. Williams T. Gokhale K. M. Taverner T. Chandan J. S. Brown K. Simms-Williams N. Shah A. D. (2022). Symptoms and risk factors for long COVID in non-hospitalized adults. Nature Medicine, 28 (8 ), 1706–1714. Sudre C. H. Murray B. Varsavsky T. Graham M. S. Penfold R. S. Bowyer R. C. . . .Canas L. S. (2021). Attributes and predictors of long COVID. Nature Medicine, 27 (4 ), 626–631. Sykes D. L. Holdsworth L. Jawad N. Gunasekera P. Morice A. H. Crooks M. G. (2021). Post-COVID-19 symptom burden: What is long-COVID and how should we manage it? Lung, 199 (2 ), 113–119. 10.1007/s00408-021-00423-z 33569660 Thompson E. J. Williams D. M. Walker A. J. Mitchell R. E. Niedzwiedz C. L. Yang T. C. . . .Steves C. J. (2022). Long COVID burden and risk factors in 10 UK longitudinal studies and electronic health records. Nature Communications, 13 (1 ), 3528. Yong S. J. (2021). Long COVID or post-COVID-19 syndrome: putative pathophysiology, risk factors, and treatments. Infectious Diseases, 53 (10 ), 737–754. 10.1080/23744235.2021.1924397 34024217
PMC010xxxxxx/PMC10241648.txt	==== Front Clin Oncol (R Coll Radiol) Clin Oncol (R Coll Radiol) Clinical Oncology (Royal College of Radiologists (Great Britain) 0936-6555 1433-2981 Published by Elsevier Ltd on behalf of The Royal College of Radiologists. S0936-6555(23)00215-7 10.1016/j.clon.2023.06.002 Letter Impact of the COVID-19 Pandemic on Radiotherapy Interruptions and Patient Outcomes Hsu C.X. ∗†‡1∗ Li T.C. §1 ∗ Division of Radiation Oncology, Department of Radiology, Far Eastern Memorial Hospital, New Taipei City, Taiwan † School of Medicine, College of Medicine, National Yang Ming Chiao Tung University, Taipei, Taiwan ‡ Department of Biomedical Imaging and Radiological Sciences, National Yang Ming Chiao Tung University, Taipei, Taiwan § Division of Medical Imaging, Department of Radiology, Far Eastern Memorial Hospital, New Taipei City, Taiwan ∗ Author for correspondence. Chen-Xiong Hsu, Division of Radiation Oncology, Department of Radiology, Far Eastern Memorial Hospital, No. 21, Section 2, Nanya S. Road, Banqiao District, New Taipei City, 220, Taiwan. 1 Chen-Xiong Hsu and Tien-Chi Li contributed equally as co-first authors to this work. 6 6 2023 6 6 2023 20 5 2023 2 6 2023 © 2023 Published by Elsevier Ltd on behalf of The Royal College of Radiologists. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcSir — The insightful study by Grocutt et al. [1] captured our interest, as it highlighted a significant 10% decline in radical radiotherapy courses and a shift towards hypofractionated regimens during the first year of the COVID-19 pandemic in Scotland. While shedding light on the pandemic's indirect effects on radiotherapy services, the study also revealed unresolved issues that warrant careful attention. First, and foremost, the successful completion of scheduled radiotherapy exerts a profound impact on the ultimate survival outcomes of cancer patients, regardless of whether the treatment is curative or palliative [2,3]. Although the adoption of hypofractionated radiotherapy may have contributed to treatment completion, the full extent of how COVID-19 lockdown measures affected critical patient outcomes, including quality of life and overall survival, remains to be comprehensively elucidated [1]. Furthermore, a recent report by Gaudio et al. [4] explored the impact of sociodemographic factors on radiotherapy interruptions in an urban population before and during the COVID-19 pandemic in the USA. Despite an overall decrease of 36% in patient volume, the rates of interruption remained stable. However, economically disadvantaged populations undergoing longer treatment courses (>20 fractions) faced an increased risk of significant interruptions (five or more breaks) [4]. Moreover, the COVID-19 pandemic has significantly affected the transportation landscape, particularly for vulnerable patient populations with an elevated risk of infection [5]. When evaluating the fluctuations in radiotherapy activity throughout the pandemic, it is crucial to consider the factors contributing to transportation delays or cancellations due to lockdown measures. Given these concerns, future research should carefully examine the impact of the COVID-19 pandemic on radiotherapy interruption rates and long-term outcomes. We greatly appreciate the valuable insights provided by Grocutt and colleagues, as they enhance our understanding of the challenges and potential strategies to mitigate the pandemic's impact on radiotherapy services and patient outcomes. Author Contributions TCL was responsible for data curation and writing (original draft, review and editing). CXH was responsible for conceptualisation, Software, Validation and writing (original draft, review and editing). All authors contributed to the article and approved the submitted version. Funding None. Conflicts of Interest The authors declare no conflicts of interest. Acknowledgments The authors acknowledge the support for the COVID study (110293-E) provided by 10.13039/501100005866 Far Eastern Memorial Hospital . ==== Refs References 1 Grocutt L. Rutherford A. Caldwell D. Wilkinson C. Chalmers A.J. Dempsey L. The impact of COVID-19 on radiotherapy services in Scotland, UK: a population-based study Clin Oncol 35 2023 e227 e234 2 Hallemeier C.L. Moughan J. Haddock M.G. Herskovic A.M. Minsky B.D. Suntharalingam M. Association of radiotherapy duration with clinical outcomes in patients with esophageal cancer treated in NRG Oncology trials: a secondary analysis of NRG Oncology randomized clinical trials JAMA Netw Open 6 2023 e238504 3 Freischlag K. Sun Z. Adam M.A. Kim J. Palta M. Czito B.G. Association between incomplete neoadjuvant radiotherapy and survival for patients with locally advanced rectal cancer JAMA Surg 152 2017 558 564 28273303 4 Gaudio E. Ammar N. Gunturkun F. Akkus C. Brakefield W. Wakefield D.V. Defining radiation treatment interruption rates during the COVID-19 pandemic: findings from an academic center in an underserved urban setting Int J Radiat Oncol Biol Phys 116 2023 379 393 36183931 5 Rahimi E. Shabanpour R. Shamshiripour A. Kouros Mohammadian A. Perceived risk of using shared mobility services during the COVID-19 pandemic Transp Res F Traffic Psychol Behav 81 2021 271 281
PMC010xxxxxx/PMC10242155.txt	==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 Elsevier Ltd. S0264-410X(23)00658-8 10.1016/j.vaccine.2023.06.002 Article From high hopes to disenchantment: A qualitative analysis of editorial cartoons on COVID-19 vaccines in Canadian newspapers Pelletier Catherine a Labbé Fabienne b Bettinger Julie A. c Curran Janet d Graham Janice E. e Greyson Devon cf MacDonald Noni E. g Meyer Samantha B. h Steenbeek Audrey d Xu Weiai i Dubé Ève abj⁎ a Centre de recherche du CHU de Québec-Université Laval, 2400 avenue d'Estimauville, Québec, Québec G1E 6W2, Canada b Institut national de santé publique du Québec, 2400 avenue d’Estimauville, Québec, Québec G1E 7G9, Canada c Vaccine Evaluation Center, BC Children's Hospital Research Institute, University of British Columbia, 950 West 28th Avenue, Vancouver, British Columbia V5Z 4H4, Canada d School of Nursing, Dalhousie University, 5869 University Avenue, Halifax, Nova Scotia B3H 4R2, Canada e Department of Pediatrics, Dalhousie University, 5849 University Avenue, Halifax, Nova Scotia B3H 4H7, Canada f School of Population and Public Health, University of British Columbia, 2206 East Mall, Vancouver, British Columbia V6T 1Z3, Canada g Department of Pediatrics, Dalhousie University, 5980 University Avenue, Halifax, Nova Scotia B3K 6R8, Canada h School of Public Health Sciences, University of Waterloo, 200 University Avenue West, Waterloo, Ontario N2L 3G1, Canada i Department of Communication, University of Massachusetts Amherst, N308 Integrative Learning Center, 650 N. Pleasant Street, Amherst, MA 01003, USA j Département d’anthropologie, Université Laval, Pavillon Charles-De Koninck, bureau 3433, 1030 avenue des Sciences Humaines, Québec, Québec G1V 0A6, Canada ⁎ Corresponding author at: 2400 avenue d’Estimauville, Québec, Québec G1E 7G9, Canada. 6 6 2023 6 6 2023 13 12 2022 21 4 2023 1 6 2023 © 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In Canada, the first COVID-19 vaccine was approved for use in December 2020, marking the beginning of a large vaccination campaign. The campaign was not only unprecedented in terms of reach, but also with regards to the amount of information about vaccines that circulated in traditional and social media. This study’s aim was to describe COVID-19 vaccine related discourses in Canada through an analysis of editorial cartoons. We collected 2 172 cartoons about COVID-19 published between January 2020 and August 2022 in Canadian newspapers. These cartoons were downloaded and a first thematic analysis was conducted using the WHO-EPIWIN taxonomy (cause, illness, treatment, interventions, and information). From this, 389 cartoons related to COVID-19 vaccines were identified under the treatment category. These were subjected to a second thematic analysis to assess main themes (e.g., vaccine development, campaign progress, etc.), characters featured (e.g., politicians, public figures, public) and position with respect to vaccine (favorable, unfavorable, neutral). Six main themes emerged: Research and development of vaccines; Management of the vaccination campaign; Perceptions of and experiences with vaccination services; Measures and incentives to increase COVID-19 vaccine uptake; Criticism of the unvaccinated; and Effectiveness of vaccination. Our analysis revealed a shift in attitudes toward COVID-19 vaccination from high hopes to disenchantment, which may reflect some vaccine fatigue. In the future, public health authorities could face some challenges in maintaining confidence and high COVID-19 vaccine uptake. Keywords COVID-19 Vaccine Editorial cartoons Canada Qualitative research ==== Body pmc1 Introduction The COVID-19 pandemic has caused substantial deaths and disruption of social and economic life worldwide. Once the novel virus was identified, scientists started to work on the development of vaccines and treatments. In Canada, the first COVID-19 vaccine was approved for use in December 2020 [1], marking the beginning of a vaccination campaign of unprecedented scale. Due to the limited number of vaccine doses in the beginning, federal and provincial public health authorities first deployed vaccination strategies that targeted specific populations (e.g., elderly people, people with high-risk conditions, healthcare workers) to be vaccinated in priority [2], with mass vaccination campaigns following once vaccine doses were more available. Since the beginning of the pandemic, social media have been a major source of information on COVID-19 and vaccines [3], [4]. Therefore, many studies investigating public attitudes and perceptions about COVID-19 vaccines and determinants of vaccination intention used social media as data source [5], [6], [7]. Although social media are increasingly part of people's information habits [4], mainstream media (e.g., newspapers, broadcast media) are still widely used and play a critical role during a public health crisis by shaping public perceptions and behaviors [8]. One of the ways used by newspapers to convey messages is through editorial cartoons. Editorial cartoons, also known as political cartoons, are drawings that address and provide a perspective on complex events or phenomena [9]. Editorial cartoons are often considered as a “societal barometer” as they reflect cultural attitudes and mainstream values of a society in a particular place and time [10], [11]. As such, cartoon analysis offers an interesting and complementary perspective to social media analysis. Attitudes and perceptions of users who are active on digital platforms are not necessarily representative of the majority of the population (e.g., over 90% of Canadian adults have received the first two doses of COVID-19 vaccines while a high volume of negative content about vaccines was found in social media) [12], [13], [14]. In addition, a highly polarized discourse has been observed online during the COVID-19 pandemic, with an overrepresentation of extreme opinions about vaccination [14], [15]. The aim of this study was to describe COVID-19 vaccine related discourses as the pandemic evolved by analyzing editorial cartoons published on the subject in ten Canadian newspapers during the COVID-19 pandemic. 2 Methods This study is part of a larger qualitative research project conducted in four Canadian provinces (British Columbia, Ontario, Quebec and Nova Scotia). One of the objectives of this pan-Canadian project was to identify and describe online discourse related to COVID-19 at the national and provincial level by analyzing social media content and editorial cartoons [16]. In this paper, we present the results of a sub-analysis of COVID-19 vaccines cartoons, covering the period from January 1st, 2020, to August 31, 2022. 2.1 Data collection Data collection was done in two phases. For phase 1, which covers the period from January 1, 2020 to September 15, 2021, 10 Canadian newspapers (The Chronicle Herald [Nova Scotia], Le Devoir, Le Journal de Québec, La Presse [Quebec], The Toronto Star, The Toronto Sun [Ontario], The Vancouver Sun, The Times Colonist [British Columbia], The Globe and Mail and The National Post [national newspapers]) were selected. The newspapers were chosen according to certain criteria. In order to obtain content that covered news from both the different provinces and Canada, we selected newspapers with the highest circulation within the provinces and at the national level [17]. Political leaning was also considered to ensure a diverse sample, although analysis from a political perspective was not part of our objective. Finally, the newspapers selected were available in both print and digital formats, making it easier for the research team to collect and archive cartoons. Editorial cartoons were collected from the newspapers' websites. As the COVID-19 pandemic continued, a second phase of data collection was conducted to cover the period from September 16, 2021 to August 31, 2022. Due to the limited availability of editorial cartoons for some newspapers encountered at the end of the phase 1 (i.e., cartoons were no longer available online, delays in availability of archives, introduction of subscription-only access to newspaper content), this second phase was a retrospective convenience sample. Three of the initial newspapers were kept —The Toronto Sun, Le Journal du Québec and The Globe and Mail— in accordance with the selection criteria mentioned for phase 1. 2.2 Data analysis All editorial cartoons were downloaded from the newspapers websites, saved as PDF files and uploaded in a NVivo 13 project. In line with the objective of the main project, a deductive thematic content analysis was first conducted, using a coding scheme derived from the taxonomy designed by the World Health Organization’s Information Network for Epidemics (WHO-EPIWIN). This taxonomy proposes five categories (cause, illness, treatment, interventions, and information) with 35 subcategories (including one related to COVID-19 vaccines) [18]. This article presents a sub-analysis of the cartoons from the Vaccines subcategory. This second thematic content analysis was done inductively, and cartoons were coded based on main themes (e.g., vaccine development, campaign progress, etc.), characters featured (e.g., politicians, public figures, community) and position with respect to the theme (favorable, unfavorable, neutral). For both analyses, independent coding of a subset of data was performed by two research professionals to assess inter-coding reliability. The coding scheme and preliminary findings were discussed and validated by the team of researchers. 3 Results 3.1 Overview of COVID-19 vaccines editorial cartoons Of the 2,172 cartoons retrieved during the two phases, 389 related to COVID-19 vaccines. An average of 12 COVID-19 vaccine-related editorial cartoons were published monthly, representing 1.6 editorial cartoons per newspaper per month (Fig. 1 ). Between November 2020 and January 2022 when most editorial cartoons were published, this rose to 3.3 per newspaper per month. Through content analysis, the editorial cartoons were grouped into six main themes: Research and development of vaccines; Management of the vaccination campaign; Perceptions of and experiences with vaccination services; Measures and incentives to increase COVID-19 vaccine uptake; Criticism of the unvaccinated; and Effectiveness of vaccination (Fig. 2 ).Fig. 1 Average number of editorial cartoons about COVID-19 vaccination published per newspaper per month. This figure presents key vaccination-related events in Canada as reported by federal and provincial authorities [19], [20], [21]. Fig. 2 Key themes in editorial cartoons on COVID-19 vaccines. 3.2 Research and development of vaccines Twenty-eight editorial cartoons focused on research and development of COVID-19 vaccines. Most of these cartoons were published in late 2020 and early 2021, when pharmaceutical companies announced studies showing the experimental vaccines were effective. These cartoons mostly expressed positive feeling of hope and anticipation. Published towards the end of 2020, a Christmas theme was common, with vaccines portrayed as the main “wish” of the population and the long-awaited “present” (Fig. 3 ). Some editorial cartoons referred to the year 2021 as “the” year that would end the COVID-19 pandemic with vaccines.Fig. 3 Brian Gable, The Globe and Mail, November 2020. 3.3 Management of the vaccination campaign The vaccination campaign was a popular topic for cartoonists, especially in the early months of 2021; a total of 139 editorial cartoons relate to this theme. The editorial cartoons focused on the management of the campaign in the country, as the federal government was responsible for the supply and distribution of vaccines in Canada. The images were predominantly negative about Canada’s Prime Minister Justin Trudeau's management of the campaign, emphasizing the lack of preparation and absence of a realistic vaccination plan (Fig. 4 ). Delays in the delivery of vaccines and the slow rollout in spring 2021 were criticized, sometimes comparing Canada to low-income countries in need of vaccines. A few editorial cartoons specifically focused on the management of the vaccination campaign by some provincial governments (Ontario, Quebec). Once again, the cartoons demonstrated scorn for delays in vaccine procurement especially as these delays were viewed as an obstacle to the management of vaccination by the provinces. A few editorial cartoons denounced the lack of international collaboration for equitable access to vaccines worldwide. Although the pictures of the COVID-19 vaccination campaign in these cartoons were generally critical, even mocking at times, overall, vaccination was still depicted as desirable, but relatively unattainable, for Canadians.Fig. 4 Yannick Lemay, Le Journal de Québec, November 26, 2020. 3.4 Perceptions and experiences with vaccination services Overall, 57 editorial cartoons focused on the experience and perceptions of the COVID-19 vaccination services by the population. The majority of these cartoons were published during the spring of 2021, when the vaccines became more available. Some editorial cartoons addressed eligibility and prioritization of certain groups for vaccination, depicting “waiting” as the predominant theme. Despite the long delays in vaccine availability, vaccination was portrayed as largely positive in these cartoons, as the “right thing to do”. However, some cartoons had a more nuanced position, mostly related to the rare but severe side effects following the administration of AstraZeneca's vaccine. This theme appeared widely in April 2021, as the vaccine was rolled out and the rare side effect of Vaccine-Induced Immune Thrombotic Thrombocytopenia (VITT) became known. In several cartoons the AstraZeneca vaccine was described as less effective, “unwanted” by the population and a barrier to the success of the vaccination campaign, particularly in Quebec, where a large quantity of doses were available. With the growing availability of different vaccines, some cartoons illustrated the “choice” of vaccine type and brand (e.g., mRNA vaccines, viral vector-based vaccines) particularly for the second dose. Pictures often mocked the confusing recommendations and communications from public health authorities (Fig. 5 ).Fig. 5 Adrian Raeside, The Times Colonist, May 14, 2021. ©raesidecartoon.com. 3.5 Measures and incentives to increase COVID-19 vaccine uptake Fifty-five editorial cartoons brought into focus measures and incentives that were put in place to increase vaccination against COVID-19, with most cartoons published during the summer and fall of 2021. At that time, a new wave of COVID-19 cases due to the Delta variant was beginning in Canada and several provinces were considering new vaccination policies such as mandatory vaccination for healthcare workers and vaccine passport to increase vaccine uptake. As all Canadian jurisdictions had finally announced the implementation of a vaccine passport to access a range of non-essential activities and services, this was a frequent theme in the cartoons during this period. These editorial cartoons generally depicted individuals showing their vaccine passport in different contexts, sometimes accompanied by frustrated not-(yet) vaccinated individuals. Others clearly referred to the vaccine passport as a harmful measure. Some editorial cartoons showed the complexity of implementing and enforcing the vaccine passport, while others mocked their use, portraying them as required at private events (i.e., guests having to show their passport prior to coming in a home). Other editorial cartoons, mostly in Quebec, focused on the mandatory vaccination policy for healthcare workers that the Quebec government announced it would adopt in the fall of 2021. Even if non-vaccinated healthcare workers were implicitly blamed and identified as “anti-vaxx”, these editorial cartoons generally negatively portrayed this policy and highlighted its inconsistency in a context of a healthcare labour shortage and weakened healthcare system. Others portrayed the stubbornness of the Quebec government in pushing for mandatory vaccination for healthcare workers, which was finally abandoned due to public opposition (Fig. 6 ).Fig. 6 Yannick Lemay, Le Journal de Québec, October 28, 2021. Editorial cartoons also depicted initiatives that were being implemented to encourage people to get vaccinated (lottery, vaccination clinics at sporting or cultural events, school outings). Young adults were often represented as the target population of such initiatives, since vaccination rates were generally lower among this group. These cartoons pictured these initiatives as positive. Other editorial cartoons were more critical of these initiatives, giving exaggerated examples of incentives or portraying those taking advantage of them as capricious (Fig. 7 ).Fig. 7 Adrian Raeside, The Times Colonist, June 5, 2021. ©raesidecartoon.com. Finally, three editorial cartoons focused on the Quebec government's idea of imposing a “health tax” on non-vaccinated people. Announced in January 2022, this measure was quickly abandoned due to political and popular opposition. Of note, two of the three cartoons did not come from Quebec newspapers, which illustrates the media impact of this announcement and wide opposition. 3.6 Criticism of the unvaccinated Criticism of the unvaccinated was present in 76 editorial cartoons. Appearing in print as early as September 2020, most were published in the summer and fall of 2021, when the vaccination campaign moved into the general public. A large proportion of the editorial cartoons were focused on people who refused to be vaccinated, often identified as “anti-vaxxers”. These cartoons portrayed unvaccinated people as stupid, unreasonable and illogical (Fig. 8 ), putting their personal interests ahead of the wider public. Other editorial cartoons blamed “anti-vaxxers” for being responsible for the transmission of the virus or represented them as a hindrance to vaccination efforts. A few editorial cartoons illustrated the link between misinformation and opposition to vaccines, through their criticism of social media and of some Canadian politicians who contributed to the spread of false claims about COVID-19 vaccines.Fig. 8 Tim Dolighan, The Toronto Sun, July 3, 2021. Protests of people against vaccination and vaccine documentation (passports) were illustrated by some cartoonists. In particular, cartoons depicted protests against the vaccine passport in the fall of 2021 and against mandatory vaccination for Canadian truck drivers in the winter of 2022. Cartoonists either criticized such protests or mocked protesters that requested “freedom” in front a “dictatorship” imposed with measures such as the vaccine passport (Fig. 9 ).Fig. 9 Sue Dewar, The Toronto Sun, February 1st, 2022. 3.7 Effectiveness of vaccination Twenty-two editorial cartoons questioned the effectiveness of vaccination as the solution to end the pandemic. Covered in all the time periods analyzed, effectiveness as a theme was more apparent between April 2021 and January 2022. In 2021, these cartoons showed the threat to vaccination efforts posed by the emergence of new variants (Delta, Omicron), comparing vaccination to a “race” or a “marathon.” Most of these editorial cartoons expressed pro-vaccination sentiments (optimism that vaccines would remain effective against these new variants). At the end of 2021 and beginning of 2022, the new booster doses were presented in editorial cartoons, but representations became less positive. Fatigue related to the number of booster doses needed and the perceived ineffectiveness of vaccination was evident with the “variant-booster” sequence depicted as an endless cycle (Fig. 10 ).Fig. 10 Yannick Lemay, Le Journal du Québec, April 18, 2022. 4 Discussion Our analysis highlights the evolution of the COVID-19 vaccine editorial cartoon discourse between January 1st, 2020, and August 31, 2022. The number of cartoons on vaccine increased sharply in late 2020, following the approval of the first vaccines. After almost a year of public health measures (lockdowns, mask wearing, social distancing), vaccination was seen as a relief, as highlighted in the editorial cartoons, despite some critical views on the vaccine roll-out in Canada. This anticipation was reflected in intention by most Canadians to be vaccinated [22]. We identified six themes appearing in the vaccine-related editorial cartoons: Research and development of vaccines; Management of the vaccination campaign; Perceptions of and experiences with vaccination services; Measures and incentives to increase COVID-19 vaccine uptake; Criticism of the unvaccinated; and Effectiveness of vaccination. While in 2020, most of the cartoons were focused on Research and development, in the spring of 2021, as mass vaccination really began across the country, the focus of the discourse shifted to the perceptions and experiences of the public with the vaccination services. Some events shook the public's confidence in the safety of vaccines, such as the side effects of AstraZeneca's vaccine and the “mix and match” recommendation for the second dose, that may have led some individuals to want to “wait” before getting vaccinated [23], [24]. However, vaccination was still depicted in a positive manner in cartoons published at that time. The enthusiasm with vaccination started to fade in cartoons published from September 2021, when coercive measures such as mandatory vaccination and vaccine passport began to be implemented across different Canadian jurisdictions. While editorial cartoons published in early 2021 were neutral or positive towards a vaccine passport, as its use was confirmed or extended, it began to be represented as a tool to “punish” the non-vaccinated. As the vaccination campaign progressed, these individuals and those protesting against the public health measures were particularly targeted in the cartoons. This phenomenon has also been observed in the population, especially in traditional and social media [25], [26], [27]. Vaccination became a source of stigma, as non-vaccinated individuals were blamed for refusing vaccination and perceived as “stupid” or “selfish”. Since the beginning of the pandemic, stigma was widely present in the media [16], and measures such as the passport have not only helped to fuel it, but have also contributed to polarization [28]. Interestingly, this evolution of the discourse related to the vaccine passport in editorial cartoons is consistent with Canadian perceptions of the vaccine passport, whose acceptability largely declined between summer 2021 and summer 2022 [28]. This also reflects how the implementation of the measure was announced (i.e., as a temporary measure to control the pandemic) and the fact that the measure was finally maintained for several months after the Omicron wave. Some cartoons on COVID-19 vaccination focused on the perceived un(effectiveness) of vaccination. The emergence of variants has largely contributed to this theme, as well as the ongoing circulation of the virus and the extension of the vaccination campaign with recommendation to receive additional doses. The last cartoons collected reported fatigue, particularly towards the administration of a 3rd and 4th booster dose. The impact of this fatigue can be seen in vaccination rates, which show decreasing acceptability and uptake for each additional recommended dose [12]. Our analysis also shows that the number of cartoons published on the topic was strongly influenced by key events related to vaccination (e.g., vaccine approval, implementation of vaccine policies, recommendations). Despite a change in number of newspapers included in phase II, we did not observe a decrease in the mean average number of cartoons published per newspaper per month on the topic. The downward trend rather happened in January 2022, despite new recommendations and announcements related to the administration of additional booster doses and the availability of vaccines for children aged 6 months to 4 years. This trend has also been observed in Quebec social media, where the number of daily mentions related to vaccines − and the COVID-19 pandemic in general − has been decreasing since winter 2022 [29]. This could be a sign of COVID-19 fatigue, which have been defined by the World Health Organization as a “demotivation to engage in protection behaviours and seek COVID-19-related information” [30]. Many countries have reported a decrease in adherence to protective behaviors in their population as the pandemic progressed [30]. The unprecedented scale of the campaign, coupled with the high hopes that vaccines would end the pandemic that were communicated to the public at the onset of the campaign, and the subsequent waves of infections and recommendations for additional doses may also have contributed to increase the COVID-19 vaccine fatigue [31]. Furthermore, as all measures to contain the spread of the virus were withdrawn over spring and summer 2022 in Canada, many considered the pandemic over [32]. Finally, it is interesting to note that how COVID-19 vaccines were represented in the cartoons did not significantly differ across the different newspapers. Although a political analysis was outside the scope of this study, we did not find any major differences in the way vaccines were portrayed by the different cartoonists/newspapers. While some newspapers are known to be politically oriented (i.e., more left- or right-winged), it is recognized that Canadian news media tend to be “non-partisan”, especially when compared with the United States of America [33]. Similarly, no difference was noted between newspapers from different provinces, including Quebec, which is predominantly francophone. This is consistent with other research in Canada showing that Canadian newspapers tend to be congruent with each other, especially when the issue is important [34], [35], as in this case, a health crisis. 5 Limitations Our study has some limitations. First, only mainstream national and provincial English and French language newspapers with wide readership representing their regions were selected, resulting in a relatively small sample of newspapers. It is possible that we were unable to capture some opinions and perceptions published in newspapers with more extreme political orientations. Also, our results are based on an interpretive subjective analysis, and although different approaches such as independent dual-coding was used to ensure the trustworthiness of the data, there is a possibility that our interpretations are not necessarily those intended by the cartoonists. Finally, the change made in the methodology between the two phases of the data collection (i.e., number of surveyed newspapers) may not provide an accurate picture of the evolution of the discourse. However, validation analysis was conducted in NVivo to compare the coding themes using only the cartoons published in the three newspapers kept in Phase 2 for the Phase 1 (January 1, 2020 to September 15, 2021). No major differences were identified, suggesting that the changes in newspapers sources did not caused a major bias in the dataset. Also, to mitigate the impact of this potential issue, an average of cartoons per newspaper per month was calculated. Although imperfect, this allows a more accurate interpretation of the evolution of the number of cartoons. Finally, although the cartoons included in this study were published in digital format, their diffusion outside of the newspaper platforms (i.e., in social media) was not analyzed. It will be interesting in future studies to assess the “social life” of cartoons (e.g., which ones are shared, liked, criticized, become memes, etc.). 6 Conclusion The volume of information on COVID-19 vaccines (true, false and even misleading) exploded during the COVID-19 pandemic, leading early on to its coining as an “infodemic”, an epidemic of information [36]. While many researchers tried to make sense of these data through the use of artificial intelligence on big datasets [37], [38], our study used qualitative methods to analyze editorial cartoons. Editorial cartoons provide a perspective on timely topics that have already been established in the mainstream media as worthy of public attention, and can be seen as barometer of general public opinion [10], [11]. While understudied in public health, social sciences and communication scholars have analyzed editorial cartoons as representations of cultural attitudes and values of a society on a particular topic at a particular time [11]. By identifying common perceptions on a specific topic, the analysis of cartoons can provide insights for the development of health promotion intervention. Whereas factual educational interventions have limited impact on behavioral intention [39], editorial cartoons, as visual representations, could also be used to convey messages in a more compelling way [40], [41], [42]. Our analysis has revealed that general attitudes toward COVID-19 vaccination evolved from high hopes to disenchantment and disinterest in the topic, which may reflect some of the challenges ahead to maintain vaccine confidence and high level of uptake for COVID-19 (and routine) vaccination. Funding This study was funded by the Canadian Institutes of Health Research (grant #420096). CRediT authorship contribution statement Catherine Pelletier: Formal analysis, Investigation, Data curation, Writing – original draft. Fabienne Labbé: Formal analysis, Investigation, Data curation, Writing – review & editing. Julie A. Bettinger: Conceptualization, Writing – review & editing, Supervision. Janet Curran: Conceptualization, Writing – review & editing, Supervision. Janice E. Graham: Conceptualization, Writing – review & editing, Supervision. Devon Greyson: Conceptualization, Writing – review & editing, Supervision. Noni E. MacDonald: Conceptualization, Writing – review & editing, Supervision. Samantha B. Meyer: Conceptualization, Writing – review & editing, Supervision. Audrey Steenbeek: Conceptualization, Writing – review & editing, Supervision. Weiai Xu: Conceptualization, Writing – review & editing, Supervision. Ève Dubé: Conceptualization, Investigation, Writing – original draft, Funding acquisition. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available on request. Acknowledgements We wish to thank the newspapers and cartoonists who granted us permission to reproduce their work in this article. We also want to acknowledge the work of the research assistants who have contributed to the data collection for this article: Manvi Bhalla, Angèle Larivière, Florence L’Écuyer, Anjana Rajendran, Bobbi Rotolo and Ziwa Yu. This work was supported by the Canadian Institutes of Health Research [grant number 439799]. ==== Refs References 1 Government of Canada. Drug and vaccine authorizations for COVID-19: List of authorized drugs, vaccines and expanded indications, https://www.canada.ca/en/health-canada/services/drugs-health-products/covid19-industry/drugs-vaccines-treatments/authorization/list-drugs.html#wb-auto-4, 2022 [accessed 3 November 2022]. 2 National Advisory Committee on Immunization. Guidance on prioritization of initial doses of COVID-19 vaccines [2020-12-18], https://www.canada.ca/en/public-health/services/immunization/national-advisory-committee-on-immunization-naci/guidance-prioritization-initial-doses-covid-19-vaccines.html, 2022 [accessed 3 November 2022]. 3 Neely S. Eldredge C. Sanders R. Health information seeking behaviors on social media during the COVID-19 pandemic among American social networking site users: survey study J Med Internet Res 11 23 2021 e29802 4 DataReportal. Digital 2022: Global overview report, https://datareportal.com/reports/digital-2022-global-overview-report?utm_source=Global_Digital_Reports&utm_medium=Article&utm_campaign=Digital_2022, 2022 [accessed 3 November 2022]. 5 Cascini F. Pantovic A. Al-Ajlouni Y.A. Failla G. Puleo V. Melnyk A. Social media and attitudes towards a COVID-19 vaccination: a systematic review of the literature EClinicalMedicine 48 2022 101454 10.1016/j.eclinm.2022.101454 6 Boucher J.-C. Cornelson K. Benham J.L. Fullerton M.M. Tang T. Constantinescu C. Analyzing social media to explore attitudes and behaviors following the announcement of successful COVID-19 vaccine trials: infodemiology study JMIR Infodemiology 1 1 2021 e28800 34447924 7 Hussain A. Tahir A. Hussain Z. Sheikh Z. Gogate M. Dashtipour K. Artificial Intelligence – Enabled analysis of public attitudes on Facebook and Twitter toward COVID-19 vaccines in the United Kingdom and the United States: Observational study J Med Internet Res 23 4 2021 e26627 33724919 8 Pappas G. Kiriaze I.J. Giannakis P. Falagas M.E. Psychological consequences of infectious diseases Clin Microbiol Infect 15 8 2009 743 747 10.1111/j.1469-0691.2009.02947.x 19754730 9 Greenberg J. Framing and temporality in political cartoons: A critical analysis of visual news discourse Can Rev Soc 39 2 2002 181 198 10.1111/j.1755-618X.2002.tb00616.x 10 Gamson W.A. Stuart D. Media discourse as a symbolic contest: The bomb in political cartoons Sociol Forum 7 1 1992 55 86 10.1007/BF01124756 11 Giarelli E. Tulman L. Methodological issues in the use of published cartoons as data Qual Health Res 13 2003 945 956 10.1177/1049732303253545 14502960 12 Government of Canada. COVID-19 vaccination coverage. https://health-infobase.canada.ca/covid-19/vaccination-coverage/; 2023 [accessed April 17, 2023]. 13 Rotolo B. Dubé E. Vivion M. MacDonald S.E. Meyer S.B. Hesitancy towards COVID-19 vaccines on social media in Canada Vaccine 40 19 2022 2790 2796 10.1016/j.vaccine.2022.03.024 35370015 14 Council of Canadian Academies. Fault Lines. https://www.cca-reports.ca/wp-content/uploads/2023/02/Report-Fault-Lines-digital.pdf; 2023 [accessed April 17, 2023]. 15 Ojea Quintana I. Reimann R. Cheong M. Alfano M. Klein C. Polarization and trust in the evolution of vaccine discourse on Twitter during COVID-19 PLoS One 17 12 2022 e0277292 36516117 16 Labbé F. Pelletier C. Bettinger J.A. Curran J. Graham J.E. Greyson D. Stigma and blame related to COVID-19 pandemic: A case-study of editorial cartoons in Canada Soc Sci Med 296 2022 114803 10.1016/j.socscimed.2022.114803 17 Newspapers Canada. Circulation Report: Daily Newspapers 2015. https://nmc-mic.ca/wp-content/uploads/2016/06/2015-Daily-Newspaper-Circulation-Report-REPORT_FINAL.pdf; 2015 [accessed April 17, 2023]. 18 Purnat T.D. Vacca P. Czerniak C. Ball S. Burzo S. Zecchin T. Infodemic signal detection During the COVID-19 pandemic: Development of a methodology for identifying potential information voids in online conversations JMIR Infodemiology 1 1 2021 e30971 34447926 19 Canadian Institute for Health Information. Canadian COVID-19 intervention timeline. https://www.cihi.ca/en/canadian-covid-19-intervention-timeline; 2022 [accessed 29 November 2022]. 20 Statistics Canada. COVID-19 in Canada: A two-year update on social and economic impacts. https://www150.statcan.gc.ca/n1/pub/11-631-x/11-631-x2022001-eng.htm; 2022 [accessed 29 November 2022]. 21 Retail Council of Canada. Proof of vaccination requirements by province. https://www.retailcouncil.org/coronavirus-info-for-retailers/vaccination-requirements-by-province/#:∼:text=Since%20March%2012%2C%202022%2C%20the,no%20longer%20required%20in%20Quebec; 2022 [accessed 29 November 2022]. 22 Tang X. Gelband H. Nagelkerke N. Bogoch I.I. Brown P. Morawski E. COVID-19 vaccination intention during early vaccine rollout in Canada: A nationwide online survey Lancet Reg Health Am 2 2021 100055 10.1016/j.lana.2021.100055 23 Dunham J. (2021, May 4). NACI accused of contributing to confusion, hesitancy over ‘preferred’ vaccines guidance. CTV News. https://www.ctvnews.ca/health/coronavirus/naci-accused-of-contributing-to-confusion-hesitancy-over-preferred-vaccines-guidance-1.5413074. 24 Driedger SM Capurro G Tustin J Jardin CG I won’t be a guinea pig”: Rethinking public health communication and vaccine hesitancy in the context of COVID-19 Vaccine 41 1 2023 1 4 10.1016/j.vaccine.2022.11.056 36460532 25 Cho H. Li W. Cannon J. Lopez R. Song C.C. Testing three explanation for stigmatization of people of Asian descent during COVID-19: maladaptive coping, biased media use, or racial prejudice? Ethn Health 26 1 2021 94 109 10.1080/13557858.2020.1830035 33059486 26 Bagcchi S. Stigma during the COVID-19 pandemic Lancet Infect Dis 20 7 2020 782 10.1016/S1473-3099(20)30498-9 32592670 27 Li Y. Twersky S. Ignace K. Zhao M. Purandare R. Bennett-Jones B. Constructing and communicating COVID-19 stigma on Twitter: A content analysis of tweets during the early stage of the COVID-19 outbreak Int J Environ Res Public Health 17 18 2020 6847 10.3390/ijerph17186847 32961702 28 Bardosh K. de Figueiredo A. Gur-Arie R. Jamrozik E. Doidge J. Lemmens T. The unintended consequences of COVID-19 vaccine policy: Why mandates, passports and restrictions may cause more harm than good BMJ Glob Health 7 2022 e008684 29 Gagnon D, Dubé E, Pelletier C. Infodémie et vaccination contre la COVID-19 au Québec – Aperçu des conversations en ligne de novembre 2021 à mars 2022. https://www.inspq.qc.ca/publications/3260-infodemie-vaccination-covid-19-conversations-novembre-21-mars-22; 2022 [accessed 29 November 2022]. 30 World Health Organization. Pandemic fatigue: Reinvigorating the public to prevent COVID-19. https://apps.who.int/iris/bitstream/handle/10665/335820/WHO-EURO-2020-1160-40906-55390-eng.pdf; 2020 [accessed 29 November 2022]. 31 Su Z. Cheshmehzangi A. McDonnell D. da Veiga C.P. Xiang Y. Mind the “vaccine fatigue” Front Immunol 13 2022 839433 10.3389/fimmu.2022.839433 32 Lewis T. (2022, March 14). People, not science, decide when a pandemic is over. Scientific American. https://www.scientificamerican.com/article/people-not-science-decide-when-a-pandemic-is-over1/. 33 Hallin D.C. Mancini P. Comparing media systems beyond the Western world 2011 Cambridge University Press New York 34 Thibault S. Bastien F. Gosselin T. Brin C. Scott C. Is there a distinct Quebec media subsystem in Canada? Evidence of ideological and political orientations among Canada new media organization Canadian Journal of Political Science/Revue Canadienne de Science Politique 53 3 2020 638 657 10.1017/S0008423920000189 35 Gidengil E. Setting the Agenda? A case study of newspaper coverage of the 2006 Canadian election campaign Marland A. Giasson T. Small T.A. Political Communication in Canada: Meet the Press and Tweet the Rest 2014 UBC Press Vancouver 36 Pan American Health Organization. Understanding the infodemic and misinformation in the fight against COVID-19. https://www.paho.org/en/documents/understanding-infodemic-and-misinformation-fight-against-covid-19; 2020 [accessed 27 September 2022]. 37 Kolluri N. Liu Y. Murthy D. COVID-19 misinformation detection: Machine-learned solutions to the infodemic JMIR Infodemiology 2 2 2022 e38756 37113446 38 Cresswell K. Tahir A. Sheikh Z. Hussain Z. Domínguez Hernández A. Harrison E. Understanding public perceptions of COVID-19 contact tracing apps: Artificial Intelligence–eabled social media analysis J Med Internet Res 23 5 2021 e26618 10.2196/26618 33939622 39 Dubé E. Gagnon D. MacDonald N.E. SAGE Working Group on Vaccine Hesitancy. Strategies intended to address vaccine hesitancy: review of published reviews Vaccine 33 34 2015 4191 4203 10.1016/j.vaccine.2015.04.0401 25896385 40 Delp C. Jones J. Communicating information to patients: the use of cartoon illustrations to improve comprehension of instructions Acad Emerg Med 3 1996 264 270 10.1111/j.1553-2712.1996.tb03431.x 8673784 41 Kearns C. Kearns N. The role of comics in public health communication during the COVID-19 pandemic J Vis Commun Med 43 3 2020 139 149 10.1080/17453054.2020.1761248 32643470 42 Czerwiec M. Williams I. Squier S.M. Green M.J. Myers K.R. Smith S.T. Graphic Medicine Manifesto Vol. 1 2015 Penn State University Press Pennsylvania
PMC010xxxxxx/PMC10244858.txt	==== Front Chest Chest Chest 0012-3692 1931-3543 American College of Chest Physicians. Published by Elsevier Inc. S0012-3692(23)00313-6 10.1016/j.chest.2023.02.043 Correspondence Response Okin Daniel MD, PhD a Alba George A. MD a Bebell Lisa M. MD b Lai Peggy S. MD, MPH a∗ a Division of Pulmonary and Critical Care Medicine, Boston, MA b Department of Medicine, Massachusetts General Hospital, Boston, MA ∗ CORRESPONDENCE TO: Peggy S. Lai, MD, MPH 7 6 2023 6 2023 7 6 2023 163 6 e291e291 © 2023 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved. 2023 American College of Chest Physicians Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcTo the Editor: We thank the letter’s authors for their interest in our manuscript1 and for the opportunity to clarify our findings. In our manuscript, we referenced several studies focused on prone position ventilation (PPV) strategy, although we did not cite the 52-patient pilot randomized trial performed by Page and colleagues.2 In this trial, patients were randomized to a PPV session of 24 h vs 16 h, with duration of PPV as the primary outcome measure. The pilot trial was not powered to detect mortality differences between groups, although the trial did demonstrate safety of longer-duration prone positioning and highlighted its feasibility. Although in our observational study we defined prolonged PPV as a prone duration lasting at least 24 h before supination, in practice many clinicians at our institution left patients in PPV until there was evidence of clinical stability. As a consequence, the median duration of the first prone session in the prolonged PPV group was 40 h, whereas for the intermittent PPV group it was 17 h. There is heterogeneity in the literature regarding optimal prolonged PPV strategy—whether it should be based on strict time cutoffs for PPV or leaving a patient in PPV pending durable clinical improvement. Additional studies are necessary to address the remaining important questions regarding prolonged PPV, such as whether a prone duration longer than 24 h is beneficial, whether supination can cause harm, and metrics to identify the appropriate timing for supination. Funding/Support D. O. is supported by the NIH (T32HL116725). G. A. A. is supported by the NIH (5KL2TR002542-02). Financial/Nonfinancial Disclosures See earlier cited article for author conflicts of interest. Acknowledgments Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. ==== Refs References 1 Okin D. Huang C.Y. Alba G.A. Prolonged prone position ventilation is associated with reduced mortality in intubated COVID-19 patients Chest 163 3 2023 533 542 36343687 2 Page D.B. Vijaykumar K. Russell D.W. Prolonged prone positioning for COVID-19-induced acute respiratory distress syndrome: a randomized pilot clinical trial Ann Am Thorac Soc 19 4 2022 685 687 34491885
PMC010xxxxxx/PMC10244863.txt	==== Front Chest Chest Chest 0012-3692 1931-3543 American College of Chest Physicians. Published by Elsevier Inc. S0012-3692(23)00297-0 10.1016/j.chest.2023.02.028 Correspondence Extended Prone Positioning Duration, But After How Many Sessions? Walter Thaïs MD, MMath a∗ Hajage David MD, PhD b Ricard Jean-Damien MD, PhD ac a DMU ESPRIT, Service de Médecine Intensive Réanimation, AP-HP, Hôpital Louis Mourier, Colombes, France b Sorbonne Université, INSERM, Institut Pierre Louis d'Epidémiologie et de Santé Publique and AP-HP, Hôpitaux Universitaires Pitié Salpêtrière-Charles Foix, Département de Santé Publique, Centre de Pharmacoépidémiologie (Cephepi), Paris, France c UMR1137 IAME, INSERM, Université Paris Cité, Paris, France ∗ CORRESPONDENCE TO: Thaïs Walter, MD, MMath 7 6 2023 6 2023 7 6 2023 163 6 e286e287 © 2023 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved. 2023 American College of Chest Physicians Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcTo the Editor: We read with great interest the study published in CHEST (March 2023) by Dr Okin and colleagues1 examining whether prone positioning of extended duration (> 24 h) is associated with decreased mortality in mechanically ventilated COVID-19-related ARDS. In this study, patients who did not show clinical improvement after prone positioning of standard duration were left prone, sometimes for several consecutive days. Because prone position contributes to lung protection during mechanical ventilation, prolonging its duration might further maintain a more homogenous distribution of the gas-tissue ratio, thus delaying the increase in overdistention observed when patients are in supine position.2 Reducing the number of turning-over maneuvers also might avoid or reduce potential derecruitment associated with these maneuvers. There are, however, issues in the article that we would like to highlight and comment on. First, the main outcome of the study is 30-day all-cause mortality. However, the precise start of follow-up is not specified. Second, the definition of the two groups seems unclear. They are indeed defined as follows: “Prolonged prone positioning ventilation (PPV) was defined as a prone duration lasting at least 24 hours before supination. Intermittent PPV was defined as daily pronation and supination events. Patients who underwent a single PPV event during their ICU stay, or who had prone sessions on nonconsecutive days, were classified as intermittent if the longest prone session length was less than 24 hours.” As described, patients in the prolonged group may have been turned prone for a prolonged duration only after one or several prone positioning sessions of duration 24 hours or less. This definition of the exposure might induce an immortal time bias if the exposure assignment does not coincide with the start of follow-up. Finally, among prone positioning sessions of duration 24 hours or longer, the exact duration varies widely, between 24 hours and more than 300 hours for a single session. The criteria used to stop an extended prone positioning session are not provided. Because several studies have been published on extended prone positioning,3, 4, 5 being able to compare the different protocols is of prime importance for designing future randomized control trial to the best benefit of patients. Financial/Nonfinancial Disclosures The authors have reported to CHEST the following: J. D. R. reports that Fisher & Paykel has covered travel expenses and provides high flow devices to a multicenter randomized controlled trial which J. D. R. is conducting on the use of nasal high flow in patients with acute hypercapnic respiratory failure. None declared (T. W., D. H.). ==== Refs References 1 Okin D. Huang C.Y. Alba G.A. Prolonged prone position ventilation is associated with reduced mortality in intubated COVID-19 patients Chest 163 3 2023 533 542 36343687 2 Cornejo R.A. Díaz J.C. Tobar E.A. Effects of prone positioning on lung protection in patients with acute respiratory distress syndrome Am J Respir Crit Care Med 188 4 2013 440 448 23348974 3 Walter T. Zucman N. Mullaert J. Extended prone positioning duration for COVID-19-related ARDS: benefits and detriments Crit Care 26 1 2022 208 35804453 4 Douglas I.S. Rosenthal C.A. Swanson D.D. Safety and outcomes of prolonged usual care prone position mechanical ventilation to treat acute coronavirus disease 2019 hypoxemic respiratory failure∗ Crit Care Med 49 3 2021 490 502 33405409 5 Cornejo R.A. Montoya J. Gajardo A.I.J. Continuous prolonged prone positioning in COVID-19-related ARDS: a multicenter cohort study from Chile Ann Intensive Care 12 1 2022 109 36441352
PMC010xxxxxx/PMC10245362.txt	==== Front Drugs Drugs Drugs 0012-6667 1179-1950 Springer International Publishing Cham 37285013 1894 10.1007/s40265-023-01894-5 AdisInsight Report Glofitamab: First Approval Shirley Matt dru@adis.com grid.420067.7 0000 0004 0372 1209 Springer Nature, Mairangi Bay, Private Bag 65901, Auckland, 0754 New Zealand 7 6 2023 17 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Glofitamab (Columvi®) is a CD20 × CD3 T-cell-engaging bispecific monoclonal antibody being developed by Roche for the treatment of B-cell non-Hodgkin lymphomas, including diffuse large B-cell lymphoma (DLBCL). Glofitamab received its first approval (with conditions) on 25 March 2023, in Canada, for the treatment of adult patients with relapsed or refractory DLBCL not otherwise specified, DLBCL arising from follicular lymphoma, or primary mediastinal B-cell lymphoma, who have received two or more lines of systemic therapy and are ineligible to receive or cannot receive CAR T-cell therapy or have previously received CAR T-cell therapy. Glofitamab is also under regulatory review for relapsed or refractory DLBCL in the EU and USA and in April 2023 received a positive opinion recommending the granting of a conditional marketing authorization in the EU. Clinical development of glofitamab, as a monotherapy and in combination with other agents for the treatment of non-Hodgkin lymphomas, is continuing worldwide. This article summarizes the milestones in the development of glofitamab leading to this first approval for relapsed or refractory DLBCL. Supplementary Information The online version contains supplementary material available at 10.1007/s40265-023-01894-5. ==== Body pmc Digital Features for this AdisInsight Report can be found at 10.6084/m9.figshare.22825841. Glofitamab (Columvi®): Key Points A CD20 × CD3 bispecific monoclonal antibody is being developed by Roche for the treatment of B-cell non-Hodgkin lymphomas Received its first approval on 25 March 2023 in Canada Approved for use in relapsed or refractory DLBCL in adults Introduction Glofitamab (Columvi®) is a T-cell-engaging bispecific monoclonal antibody being developed by Roche for the treatment of B-cell non-Hodgkin lymphomas, including diffuse large B-cell lymphoma (DLBCL) [1, 2]. As a CD20 × CD3 bispecific antibody, glofitamab simultaneously binds CD20 expressed on the surface of B-cells and CD3 in the T-cell receptor complex expressed on the surface of T-cells [2]. Through this process of simultaneous binding, glofitamab is designed to redirect a patient’s existing T-cells to target and kill malignant B-cells in B-cell non-Hodgkin lymphomas. Glofitamab received its first approval on 25 March 2023, in Canada, for the treatment of adult patients with relapsed or refractory DLBCL not otherwise specified, DLBCL arising from follicular lymphoma (trFL), or primary mediastinal B-cell lymphoma (PMBCL), who have received two or more lines of systemic therapy and are ineligible to receive or cannot receive chimeric antigen receptor (CAR) T-cell therapy or have previously received CAR T-cell therapy [1, 2]. The marketing authorization for glofitamab has been issued by Health Canada with conditions, pending the results of trials to verify the clinical benefit of the drug [1, 2].Key milestones in the development of glofitamab. DLBCL diffuse large B-cell lymphoma, MCL mantle cell lymphoma, NDA New Drug Application, PDUFA Prescription Drug User Fee Act Glofitamab is available as a 1 mg/mL solution which is to be diluted prior to administration by intravenous infusion [2]. Glofitamab is administered in 21-day cycles. Seven days prior to the initiation of glofitamab, all patients must receive pretreatment with a single intravenous dose of obinutuzumab 1000 mg, which is used to deplete circulating and lymphoid tissue B-cells and to reduce the risk of cytokine release syndrome. Glofitamab therapy is to be initiated using a step-up dosing schedule commencing with glofitamab 2.5 mg 7 days after obinutuzumab pretreatment (i.e. on day 8) followed by glofitamab 10 mg on day 15. The recommended glofitamab dose after step-up is 30 mg, administered on day 1 of cycle 2 and of each subsequent cycle, with treatment to continue for a maximum of 12 cycles or until disease progression or unmanageable toxicity. Glofitamab carries a boxed warning for cytokine release syndrome, which may be serious or life-threatening [2]. In addition to pretreatment with obinutuzumab and the use of a step-up dosing schedule, to reduce the risk of cytokine release syndrome, patients receiving glofitamab should be well hydrated prior to infusion. Additionally, premedication with an oral analgesic/anti-pyretic and an anti-histamine should be administered prior to each glofitamab infusion, with an intravenous glucocorticoid also used in all patients prior to the first four glofitamab infusions and at all subsequent infusions for any patient who experienced cytokine release syndrome with the previous glofitamab dose. Patients should be monitored for ≥ 10 h following the first infusion of glofitamab and as clinically indicated for subsequent infusions. Severe or life-threatening cytokine release syndrome should be treated with tocilizumab, with or without corticosteroids. At least one dose of tocilizumab must be available at cycles 1 and 2 prior to initiating glofitamab infusion, with an additional dose of tocilizumab accessible within 8 h. If cytokine release syndrome occurs, glofitamab should be withheld until resolution. Alternatively, permanent discontinuation of glofitamab may be required based on cytokine release syndrome severity. Glofitamab must not be administered to patients with an active infection [2]. On 26 April 2023, the European Medicines Agency (EMA) Committee for Medicinal Products for Human Use adopted a positive opinion, recommending the granting of a conditional marketing authorization for glofitamab for the treatment of DLBCL [3]. Glofitamab is also under regulatory review for DLBCL in the USA, with the drug receiving a Fast Track designation from the US FDA in January 2023 [4]. Clinical development of glofitamab, as a monotherapy and in combination with other agents for the treatment of non-Hodgkin lymphomas, is continuing worldwide. Company Agreements Glofitamab was originated and developed by Roche. As of July 2022, Chugai Pharmaceutical in-licensed various drugs from Roche, including glofitamab, for their development and distribution in Japan [5]. Scientific Summary Glofitamab is a CD20 × CD3 T-cell-engaging bispecific antibody engineered with a novel 2 ꞉ 1 configuration of anti-CD20 ꞉ anti-CD3 [2]. Pharmacodynamics Glofitamab acts in the treatment of B-cell non-Hodgkin lymphomas by promoting T-cell–mediated lysis of CD20-expressing B-cells [2]. Glofitamab simultaneously binds bivalently to CD20 on B-cells and monovalently to CD3 on T-cells leading to the formation of an immunological synapse between CD20-expressing B-cells and CD3-expressing T-cells. Formation of the immunological synapse leads to T-cell activation and proliferation [2]. In addition to inducing the expansion of pre-existing intra-tumour resident T-cell populations, a preclinical study has shown that glofitamab also promotes the recruitment of peripheral blood T-cells [6]. Glofitamab treatment has also been shown to result in a dose-dependent and transient induction of proinflammatory cytokines (including interferon-γ, IL-6, IL-2, IL-8, IL-10, IL-15 and IL-17) [7]. Pharmacokinetics Glofitamab pharmacokinetics are linear and dose proportional over the dose range of 0.005 mg to 30 mg [2]. Maximum serum concentrations are reached at the end of the intravenous infusion, after which glofitamab concentrations decline in a bi-exponential fashion. The central volume of distribution is 3.33 L and the peripheral volume of distribution is 2.18 L, based on estimations from population pharmacokinetic modelling [2]. Glofitamab has an estimated half-life of 4–8 days. Although the specific pathways have not been characterized, as an antibody, glofitamab is expected to be principally cleared by catabolism.Features and properties of glofitamab Alternative names aCD20/CD3 TCB 2; aCD20/CD3 TCBs; anti-CD20 CD3 TCB; anti-CD20/CD3 bispecific monoclonal antibody; CD20-TCB; Columvi®; RG 6026; RG6026-2; RO 7082859 Class Antineoplastics; bispecific antibodies; immunotherapies; monoclonal antibodies Mechanism of action Antibody-dependent cell cytotoxicity; T lymphocyte stimulation Route of administration Intravenous infusion Pharmacodynamics Simultaneously binds to CD20 on B-cells and to CD3 on T-cells leading to the formation of an immunological synapse between CD20+ B-cells and CD3+ T-cells; promotes T-cell mediated lysis of CD20-expressing B-cells Pharmacokinetics After a single 10-mg dose: Cmax = 2.34 µg/mL; Tmax = 8.05 h, t½ = 106 h; AUCinf = 244 h·µg/mL; CL = 40.4 mL/h; Vz = 6180 mL Most common adverse events Cytokine release syndrome, neutropenia, anaemia, thrombocytopenia ATC codes WHO ATC code L01 (antineoplastic agents) EphMRA ATC code L1 (antineoplastics) Therapeutic Trials In Relapsed/Refractory B-Cell Lymphoma Pivotal NP30179 Trial A fixed-duration course of glofitamab monotherapy (following obinutuzumab pretreatment) was associated with a high response rate in patients with relapsed or refractory DLBCL in the open-label, multi-centre, multi-cohort phase I/II trial NP30179 (NCT03075696) [8]. At a median follow-up of 12.6 months, 39% (95% CI 32–48) of patients in the intent-to-treat population (n = 155) in the phase II part of the trial had had a complete response (as the best overall response) as assessed by an independent review committee (IRC; primary endpoint). For these patients, the median time to a complete response was 42 days [8]. In a prespecified subgroup analysis, the treatment effect was generally consistent among patients who had received previous CAR-T cell therapy (n = 52) and those who had not (n = 103), with 35% and 42% of patients in the respective groups achieving a complete response [8]. In other prespecified subgroup analyses, patients with relapsed disease (n = 23) had a complete response rate of 70% whereas patients with disease that was refractory (n = 132) to the last previous treatment had a complete response rate of 34% [8]. The objective response rate (complete or partial response) in the intent-to-treat population was 52% [8]. The median duration of objective response was 18.4 months (95% CI 13.7 to not reached). Complete responses were generally durable, with the median duration of complete response not reached (95% CI 16.8 to not reached) at data cut-off. Median IRC-assessed progression-free survival (PFS) was 4.9 months and median overall survival was 11.5 months [8]. In the pivotal cohort (n = 108), 35% (95% CI 26–45) of patients had a complete response as assessed by the IRC at a median follow-up of 9.0 months, significantly (p < 0.001) higher than the rate (20%) observed in a historical control cohort [8].Key clinical trials of glofitamab Identifier(s) Indication Phase Drug(s) Location(s) Sponsor(s) Status NCT03075696; NP30179 r/r DLBCL I/II Glofitamab, obinutuzumab, tocilizumab Multinational Hoffmann-La Roche Active NCT04657302 r/r DLBCL I Glofitamab, obinutuzumab, tocilizumab China Hoffmann-La Roche Active NCT04408638 r/r DLBCL III Glofitamab, obinutuzumab, tocilizumab, rituximab, gemcitabine, oxaliplatin Multinational Hoffmann-La Roche Recruiting NCT05335018 r/r DLBCL II Glofitamab, poseltinib, lenalidomide Republic of Korea Seoul National University Hospital Recruiting NCT05364424 r/r DLBCL I Glofitamab, obinutuzumab, tocilizumab, rituximab, ifosfamide, carboplatin, etoposide USA Hoffmann-La Roche Recruiting NCT04313608 r/r DLBCL or HGBCL I Glofitamab, mosunetuzumab, obinutuzumab, tocilizumab, gemcitabine, oxaliplatin Australia Hoffmann-La Roche Completed NCT04889716 r/r DLBCL or trFL II Glofitamab, mosunetuzumab, obinutuzumab USA Abramson Cancer Center at Penn Medicine; Genentech Recruiting NCT04703686 r/r lymphomas II Glofitamab, obinutuzumab France Lymphoma Academic Research Organisation Recruiting NCT03533283 r/r NHL Ib/II Glofitamab, atezolizumab, crefmirlimab, obinutuzumab, polatuzumab vedotin, tocilizumab Belgium, Denmark, Israel, Italy, Spain, UK, USA Hoffmann-La Roche Active NCT05533775; iMATRIX GLO r/r NHL I/II Glofitamab, obinutuzumab, tocilizumab, rituximab, ifosfamide, carboplatin, etoposide Denmark, Germany, Italy, Republic of Korea, Spain, USA Hoffmann-La Roche Recruiting NCT05169515 r/r NHL I Glofitamab, mosunetuzumab, obinutuzumab, tocilizumab, CC-220, CC-99282 Israel, Italy, Spain, UK, USA Hoffmann-La Roche Recruiting NCT04077723 r/r NHL I Glofitamab, obinutuzumab, tocilizumab, RO7227166 Australia, Belgium, Denmark, France, Italy, Spain, UK, USA Hoffmann-La Roche Recruiting NCT05219513 r/r NHL I Glofitamab, obinutuzumab, tocilizumab, RO7443904 Australia, Denmark, France, Italy, UK, USA Hoffmann-La Roche Recruiting NCT04970901 r/r NHL Ib Glofitamab, loncastuximab tesirine, mosunetuzumab, obinutuzumab, polatuzumab vedotin Belgium, Czechia, Italy, Spain, UK, USA ADC Therapeutics Recruiting NCT03467373; NP40126 Untreated DLBCL Ib Glofitamab, obinutuzumab, polatuzumab vedotin, tocilizumab, rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone Australia, Canada, Denmark, France, Germany, Italy, Spain, UK, USA Hoffmann-La Roche Active NCT04980222 Untreated DLBCL II Glofitamab, tocilizumab, rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone Denmark, France, Netherlands, Poland, Spain, USA Hoffmann-La Roche Recruiting NCT04914741; COALITION Untreated DLBCL or HGBCL I/II Glofitamab, polatuzumab vedotin, rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone Australia Peter MacCallum Cancer Centre; Hoffmann-La Roche Recruiting DLBCL diffuse large B-cell lymphoma, HGBCL high-grade B-cell lymphoma, NHL non-Hodgkin lymphoma, r/r relapsed or refractory, trFL transformed follicular lymphoma In the phase II part of the NP30179 trial, patients were ≥ 18 years old and had histologically confirmed DLBCL not otherwise specified (71% of patients), trFL (18%), high-grade B-cell lymphoma (7%), or PMBCL (4%), an Eastern Cooperative Oncology Group performance-status score of 0 or 1, and disease that had relapsed after, or was refractory to, two or more previous lines of therapy [8]. Patients had received a median of three prior therapies (range, 2–7), with 60% of patients having received at least three prior therapies [8]. Glofitamab was administered for 12 cycles (or until disease progression or unacceptable toxicity), according to the dosing schedule detailed in Sect. 1, including the pretreatment with intravenous obinutuzumab and the use of premedications to reduce infusion-related reactions (including cytokine-release syndrome). This recommended phase II dosing schedule was established in the phase I part of the trial [9]. Patients in NP30179 were hospitalized for administration of the first dose of glofitamab [8]. Other Trials Similar results to those observed in NP30179 were reported in a phase I trial (NCT04657302) of fixed-duration glofitamab monotherapy in 30 Chinese patients with relapsed or refractory DLBCL treated with two or more previous lines of therapy [10]. At a median follow-up of 10 months, 52% of evaluable patients (n = 27) in NCT04657302 had had a complete response as assessed by IRC (primary endpoint). For these patients, the median time to a complete response was 43 days. The objective response rate (based on IRC assessment) was 63%. At data cut-off, 79% (11/14) of complete responses and 76% (13/17) of objective responses were ongoing. Median IRC-assessed PFS was 8 months (95% CI 3 to not reached) and median overall survival was 11 months (95% CI 9 to not reached) [10]. Glofitamab in combination with polatuzumab vedotin has demonstrated promising efficacy in patients with relapsed or refractory B-cell non-Hodgkin lymphoma based on preliminary results from a phase Ib/II dose-escalation and expansion study (NCT03533283) [11]. At data cut-off, 59 patients had been enrolled, with 49 patients evaluable for interim efficacy. At a median follow-up of 3.7 months, 51% of evaluable patients had a complete response (as the best overall response) while the objective response rate was 80%. Patients in NCT03533283 had received a median of two prior lines of therapy, with 70% of patients being refractory to their last therapy. In the trial, glofitamab was administered as for NP30179 (including with use of pretreatment with obinutuzumab on cycle 1 day 1) and continued for 12 cycles. Polatuzumab vedotin 1.8 mg/kg was administered on cycle 1 day 2 and then on day 1 of each cycle up to cycle 6 [11]. In Newly-Diagnosed Diffuse Large B-Cell Lymphoma Glofitamab administered in combination with rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP) as first-line therapy induced high response rates in patients with newly diagnosed DLBCL in the phase Ib, multi-centre, dose-finding study NP40126 (NCT03467373) [12]. At data cut-off, 56 patients had been enrolled, with 46 patients having reached their scheduled end-of-treatment assessment and evaluated for efficacy. At a median follow-up of 5.6 months, 76.1% of patients in the end-of-treatment population had a complete response (as the best overall response) while the objective response rate was 93.5%. In NP40126, patients received 6–8 cycles of R-CHOP with intravenous glofitamab included from cycle 2 (with step-up dosing starting on cycle 2 day 8) [12]. Adverse Events Glofitamab has manageable tolerability in patients with haematological malignancies, based on currently available data [8–12]. In the phase II part of the NP30179 trial (Sect. 2.3.1.1), with glofitamab used as monotherapy in patients (n = 154) with relapsed or refractory DLBCL, 62% of patients experienced adverse events of grade ≥ 3, with these mostly being haematological events (e.g. neutropenia, anaemia, thrombocytopenia) [8]. In total, 9% of patients discontinued treatment due to adverse events, with 3% of patients discontinuing treatment due to adverse events that were considered to be related to glofitamab. Serious adverse events occurred in 47% of patients, most commonly cytokine release syndrome [in 21% of patients, per American Society for Transplantation and Cellular Therapy (ASTCT) criteria] and sepsis (4%). Eight patients (5%) died during the phase II part of the trial (not including deaths due to progressive disease); no deaths were considered related to glofitamab treatment [8]. The most commonly reported adverse event was cytokine release syndrome, which occurred in 63% of patients by ASTCT criteria [8]. Cytokine release syndrome was generally mild to moderate in severity, with the most common manifestations including pyrexia (99%), tachycardia (27%) and hypotension (24%). Grade 3 and grade 4 cytokine release syndrome was reported in 2.6% and 1.3% of patients, respectively; there were no deaths due to cytokine release syndrome. Events of cytokine release syndrome were most commonly observed in association with the first three doses of glofitamab, occurring in 53.9% of patients after the first dose (2.5 mg on cycle 1 day 8), in 32.5% of patients after the second dose (10 mg on cycle 1 day 15), and in 28.4% of patients after the third dose (30 mg on cycle 2 day 1). Few patients (2.0%) experienced cytokine release syndrome beyond cycle 3. All cases of cytokine release syndrome resolved except for one event which was ongoing at the time the patient died from progressive disease. One patient discontinued treatment due to cytokine release syndrome. Mitigation strategies used in NP30179 to reduce the incidence and severity of cytokine release syndrome included pretreatment with obinutuzumab and step-up dosing of glofitamab [8]. Management of cytokine release syndrome most commonly involved corticosteroids and/or tocilizumab. In a cohort of patients in the phase II part of NP30179 for whom pretreatment with dexamethasone was mandated (n = 40), cytokine release syndrome occurred in 48% of patients compared with 68% of patients pretreated with investigator’s choice of corticosteroid [8]. In the mandatory dexamethasone cohort, no grade ≥ 2 events of cytokine release syndrome were observed following the second or later doses of glofitamab. Infections of any grade occurred in 38% of patients, with infections of grade ≥ 3 severity occurring in 15% of patients [8]. The most commonly observed infections were Covid-19/Covid-19-related pneumonia (incidence, 9%; grade ≥ 3, 6%) and sepsis (4%, all grade ≥ 3). There were seven infection-related deaths, due to Covid-19/Covid-19-related pneumonia (five) and sepsis (two) [8]. Other adverse events of special interest occurring during treatment with glofitamab include neurological events, tumour flare and tumour lysis syndrome [2]. Neurological adverse events of grade ≥ 2 occurred in 15% of patients in the phase II part of NP30179; grade ≥ 3 neurological adverse events occurred in five patients (3.2%), including one patient with grade 5 delirium [8]. Tumour flare was reported in 11.0% of patients (grade ≥ 3 in 2.6%) [8], with 16 of 17 tumour flare events occurring during cycle 1 of glofitamab treatment [2]. Tumour lysis syndrome occurred in two patients, with both cases of grade ≥ 3 severity [8]. In addition to the manageable tolerability of glofitamab monotherapy demonstrated in NP30179, based on early data from phase I and II trials [11–13], glofitamab also appears to have manageable tolerability during use in combination with chemotherapy and other agents used or being investigated for the treatment of haematological malignancies. Ongoing Clinical Trials In addition to the trials discussed in Sect. 2.3, which are all ongoing, glofitamab (as monotherapy, or in combination with chemotherapy or other agents) is being investigated in several other trials in relapsed and refractory or previously untreated haematological malignancies. These trials include (but are not limited to):NCT04408638, a randomized, open-label, phase III trial evaluating the efficacy and safety of glofitamab plus gemcitabine and oxaliplatin versus rituximab plus gemcitabine and oxaliplatin in patients with relapsed or refractory DLBCL [14]. NCT04980222, a single-arm phase II trial evaluating the safety, efficacy, and pharmacokinetics of glofitamab in combination with R-CHOP as the first line of treatment in patients with circulating tumour DNA high-risk DLBCL. COALITION (NCT04914741), a randomized, open-label, phase Ib/II trial evaluating the safety and tolerability of glofitamab in combination with chemotherapy consisting of R-CHOP or polatuzumab vedotin and rituximab plus cyclophosphamide, doxorubicin and prednisone (R-CHP) as the first line of treatment for younger patients (aged 18–65 years) with higher-risk DLBCL or high-grade B-cell lymphoma [15]. iMATRIX GLO (NCT05533775), a single-arm, two-part, phase I/II trial evaluating the safety and efficacy of glofitamab, as monotherapy and in combination with a rituximab, ifosfamide, carboplatin and etoposide (R-ICE) chemoimmunotherapy regimen, in paediatric and young adult patients (aged 6 months to ≤ 30 years) with relapsed or refractory mature B-cell non-Hodgkin lymphoma. NCT05169515, a non-randomized, parallel-assignment, open-label, phase I/II trial evaluating the safety, efficacy and pharmacokinetics of glofitamab or mosunetuzumab in combination with ubiquitin protein ligase complex modulators CC-220 (iberdomide) and CC-99282 in patients with B-cell non-Hodgkin lymphoma. NCT04077723, a randomized, open-label, phase I dose-escalation trial evaluating the safety, pharmacokinetics and preliminary efficacy of RO7227166 (a bispecific monoclonal antibody targeting CD19 and 4-1BB) in combination with glofitamab in patients with relapsed or refractory B-cell non-Hodgkin lymphoma [13]. NCT05219513, a non-randomized, open-label, phase I dose-escalation trial evaluating the safety, tolerability, pharmacokinetics and preliminary efficacy of RO7443904 (a bispecific antibody-like fusion protein targeting CD19 and CD28) in combination with glofitamab in patients with relapsed or refractory B-cell non-Hodgkin lymphoma [16]. Current Status Glofitamab received its first approval on 25 March 2023, in Canada, for relapsed or refractory DLBCL [1]. Glofitamab subsequently received a positive opinion from the EMA in April 2023 recommending the granting of a conditional marketing authorization in the EU for relapsed or refractory DLBCL [3]. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 199 KB) Declarations Funding The preparation of this review was not supported by any external funding. Authorship and Conflict of Interest During the peer review process the manufacturer of the agent under review was offered an opportunity to comment on the article. Changes resulting from any comments received were made by the author on the basis of scientific completeness and accuracy. Matt Shirley is a salaried employee of Adis International Ltd/Springer Nature, and declares no relevant conflicts of interest. All authors contributed to the review and are responsible for the article content. Ethics approval, Consent to participate, Consent to publish, Availability of data and material, Code availability Not applicable. This profile has been extracted and modified from the AdisInsight database. AdisInsight tracks drug development worldwide through the entire development process, from discovery, through pre-clinical and clinical studies to market launch and beyond. ==== Refs References 1. Roche Canada. COLUMVI® (glofitamab for injection) receives Health Canada authorization with conditions for adult patients with relapsed or refractory diffuse large B-cell lymphoma [media release]. 25 Mar 2023. https://www.rochecanada.com/en/media/roche-canada-news/COLUMVI-Glofitamab-for-Injection-Receives-Health-Canada-Authorization.html. 2. Roche Canada. COLUMVI® product monograph. 2023. https://www.rochecanada.com/content/dam/rochexx/roche-ca/products/ConsumerInformation/MonographsandPublicAdvisories/columvi/Columvi_PM_E.pdf. Accessed 01 May 2023. 3. European Medicines Agency. Columvi. 2023. https://www.ema.europa.eu/en/medicines/human/summaries-opinion/columvi. Accessed 01 May 2023. 4. Roche. FDA grants priority review to Roche's bispecific antibody glofitamab for people with relapsed or refractory large B-cell lymphoma [media release]. 06 Jan 2023. http://www.roche.com. 5. Chugai Pharmaceutical Co. Conference on FY2022.12 2Q financial results. 21 Jul 2022. https://www.chugai-pharm.co.jp/cont_file_dl.php?f=FILE_5_60.pdf&src=%5b%250%5d,%5b%251%5d&rep=137,60. Accessed 01 May 2023. 6. Cremasco F Menietti E Speziale D Cross-linking of T cell to B cell lymphoma by the T cell bispecific antibody CD20-TCB induces IFNγ/CXCL10-dependent peripheral T cell recruitment in humanized murine model PLoS ONE 2021 16 1 e0241091 10.1371/journal.pone.0241091 33406104 7. Bröske AE Korfi K Belousov A Pharmacodynamics and molecular correlates of response to glofitamab in relapsed/refractory non-Hodgkin lymphoma Blood Adv 2022 6 3 1025 1037 10.1182/bloodadvances.2021005954 34941996 8. Dickinson MJ Carlo-Stella C Morschhauser F Glofitamab for relapsed or refractory diffuse large B-cell lymphoma N Engl J Med 2022 387 24 2220 2231 10.1056/NEJMoa2206913 36507690 9. Hutchings M Morschhauser F Iacoboni G Glofitamab, a novel, bivalent CD20-targeting T-cell-engaging bispecific antibody, induces durable complete remissions in relapsed or refractory B-cell lymphoma: a phase I trial J Clin Oncol 2021 39 18 1959 1970 10.1200/JCO.20.03175 33739857 10. Song YQ Zhang HL Huang HQ Glofitamab monotherapy demonstrates high complete response rates and manageable safety in Chinese patients with relapsed or refractory diffuse large B-cell lymphoma and ≥2 prior therapies [abstract] Blood 2022 140 suppl 1 12050 12051 10.1182/blood-2022-157544 11. Hutchings M Sureda A Terol MJ Glofitamab (glofit) in combination with polatuzumab vedotin (pola): preliminary phase Ib/II data supports the controllable safety and encouraging efficacy in patients with relapsed/refractory (R/R) diffuse large B-cell lymphoma (DLBCL) [title in German, English abstract] Oncol Res Treat. 2022 45 suppl 2 54 34818649 12. Topp MS Tani M Dickinson M Glofitamab plus R-CHOP induces high response rates and a favorable safety profile in patients with previously untreated diffuse large B-cell Lymphoma (DLBCL): results from a phase Ib study [abstract] Blood 2022 140 suppl 1 1775 1777 10.1182/blood-2022-157732 13. Hutchings M Carlo-Stella C Gritti G CD19 4–1BBL (RO7227166) a novel costimulatory bispecific antibody can be safely combined with the T-cell-engaging bispecific antibody glofitamab in relapsed or refractory B-cell non-Hodgkin lymphoma [abstract] Blood 2022 140 suppl 1 9461 9463 10.1182/blood-2022-157011 14. Hertzberg M Ku M Catalani O A phase III trial evaluating glofitamab in combination with gemcitabine plus oxaliplatin versus rituximab in combination with gemcitabine and oxaliplatin in patients with relapsed/refractory (R/R) diffuse large B-cell lymphoma (DLBCL) [abstract no. TPS7575] J Clin Oncol 2021 39 15_suppl TPS7575 10.1200/JCO.2021.39.15_suppl.TPS7575 15. Minson A Hamad N Yannakou CK Trial in progress: a multicentre, parallel arm, open-label trial of frontline R-CHOP/polatuzumab vedotin-RCHP and glofitamab in younger patients with higher risk diffuse large B cell lymphoma (COALITION) [abstract] Blood 2021 138 suppl 1 3571 10.1182/blood-2021-147636 16. Dickinson M Gritti G Carlo-Stella C Phase 1 Study of CD19 targeted CD28 costimulatory agonist in combination with glofitamab to enhance T cell effector function in relapsed/refractory B cell lymphoma [abstract] Blood 2022 140 suppl 1 3818 3820 10.1182/blood-2022-156808
PMC010xxxxxx/PMC10247294.txt	==== Front Am J Med Sci Am J Med Sci The American Journal of the Medical Sciences 0002-9629 1538-2990 Southern Society for Clinical Investigation. Published by Elsevier Inc. S0002-9629(23)01209-0 10.1016/j.amjms.2023.04.026 Online Images in the Medical Sciences Extensive thrombosis of the inferior vena cava and bilateral renal veins in a COVID-19 patient Songtanin Busara MD 1 Attaya Eman MD 2 Nugent Kenneth MD 1⁎ 1 Department of Internal Medicine, Texas Tech University Health Sciences Center, Lubbock, TX, USA 2 Department of Radiology, University Medical Center, Lubbock, TX, USA ⁎ Corresponding author. 8 6 2023 8 6 2023 19 6 2022 17 4 2023 © 2023 Southern Society for Clinical Investigation. Published by Elsevier Inc. All rights reserved. 2023 Southern Society for Clinical Investigation Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcCase presentation A 42-year-old man with a history of hyperthyroidism presented to the hospital with bilateral flank pain (left side worse than the right side) for 3 days with no alleviating or aggravating factors. The patient denied abdominal pain, gross hematuria, chest pain, shortness of breath, fever, dysuria, recent trauma, and a history of blood clots or intravenous drug use. Vital signs were unremarkable. Physical examination revealed bilateral costovertebral angle tenderness. Laboratory data showed normal renal function with BUN 11 mg/dL and Cr 0.7 mg/dL. Urinalysis revealed proteinuria with microscopic hematuria. Patient was COVID positive by PCR. Computed tomography (CT) of the abdomen and pelvis with contrast showed a thrombus likely originating from the infrahepatic IVC (Fig. 1A , arrow) extending into the bilateral renal veins, left greater than right (Fig. 1B, arrow), and the left lumbar vein (Fig. 1C, arrow). A wedge-shaped area of hypoenhancement in the upper pole of the right kidney was also noted and suggested renal infarction (Fig. 1D, arrow). His-d-dimer was 1289 ng/mL. Hypercoagulable work up, including Factor V Leiden, prothrombin gene mutation, antiphospholipid, was negative.Fig. 1 Figure 1 Thrombotic complications during COVID-19 infections occur frequently.1 Inferior vena cava (IVC) thrombosis is an extremely rare and life-threatening condition and can occur with compression from adjacent structures, trauma, a hypercoagulable state, or renal cell carcinoma. The most common predisposing factor for IVC thrombosis in the United States is an IVC filter.2 This patient, however, had no risk factors that could predispose him to develop thrombosis, and he likely had a hypercoagulable state from COVID-19 infection.1 SARS–CoV–2 uses the angiotensin-converting enzyme 2 to enter and infect endothelial cells. This causes tissue injury and endothelial dysfunction and releases von Willebrand factor antigen and factor VIII; these changes result in a prothrombotic state.3 Patients with COVID-19 have an increased frequency of deep venous thrombosis and pulmonary emboli, and these clinical events are associated with increased mortality. This patient likely had damage to his inferior vena cava during his COVID19 infection, and this resulted in extensive intra-abdominal thrombosis. Patient with IVC thrombus can present with lower extremity edema, abdominal, flank or back pain, or acute onset of deep vein thrombosis.2 , 4 Computed tomography can be used to visualize the location of thrombus; magnetic resonance imaging can help determine the extent of the thrombus. Management of IVC thrombosis includes removal of thrombus with pharmacological, endovascular, or surgical treatment. Anticoagulation is indicated to reduce the risk of embolization and propagation of thrombus from the time of diagnosis.4 Treatment of acute renal vein thrombosis in patients without acute kidney injury includes anticoagulation; patients with evidence of acute kidney injury should undergo removal of thrombus. Prompt administration of anticoagulation is needed because untreated acute renal vein thrombosis has a high mortality rate.1 Declaration of Competing Interest None. ==== Refs Reference 1 Avila J. Long B. Holladay D. Thrombotic complications of COVID-19 Am J Emerg Med 39 2021 213 218 33036855 2 Hollingsworth C.M. Mead T. Inferior vena caval thrombosis StatPearls July 10, 2021 StatPearls Publishing Treasure IslandFL 3 Sardu C. Gambardella J. Morelli M.B. Hypertension, thrombosis, kidney failure, and diabetes: is COVID-19 an endothelial disease? A comprehensive evaluation of clinical and basic evidence J Clin Med 9 2020 1417 10.3390/jcm9051417 32403217 4 McAree B.J. O'Donnell M.E. Fitzmaurice G.J. Inferior vena cava thrombosis: a review of current practice Vasc Med 18 1 2013 32 43 23439778
PMC010xxxxxx/PMC10247880.txt	==== Front J Am Med Dir Assoc J Am Med Dir Assoc Journal of the American Medical Directors Association 1525-8610 1538-9375 Published by Elsevier Inc. on behalf of AMDA -- The Society for Post-Acute and Long-Term Care Medicine. S1525-8610(23)00532-7 10.1016/j.jamda.2023.05.026 Original Studies The Benefits of Nursing Home Air Purification on COVID-19 Outcomes: A Natural Experiment Jutkowitz Eric PhD abcd∗ Shewmaker Peter MS a Reddy Ann MPH d Braun Joseph M. PhD e Baier Rosa R. MPH ad a Department of Health Services, Policy & Practice, Brown University School of Public Health, Providence, RI b Evidence Synthesis Program Center Providence VA Medical Center, Providence, RI c Center of Innovation in Long-Term Services and Supports, Providence VA Medical Center, Providence, RI d Center for Long-Term Care Quality & Innovation, Brown University School of Public Health, Providence, RI e Department of Epidemiology, Brown University School of Public Health, Providence, RI ∗ Corresponding Author: Eric Jutkowitz, Brown University School of Public Health, Department of Health Services, Policy & Practice, 121 South Main Street, Box G-S121-6, Providence, RI 02912. . (401) 863-3211 8 6 2023 8 6 2023 6 3 2023 25 5 2023 28 5 2023 © 2023 Published by Elsevier Inc. on behalf of AMDA -- The Society for Post-Acute and Long-Term Care Medicine. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objectives Improving indoor air quality is one potential strategy to reduce the transmission of SARS-CoV-2 in any setting, including nursing homes, where staff and residents have been disproportionately and negatively affected by the COVID-19 pandemic. Design Single group interrupted time series. Setting and Participants 81 nursing homes in a multi-facility corporation in Florida, Georgia, North Carolina, and South Carolina that installed ultraviolet air purification in their existing heating, ventilation, and air conditioning systems between July 27, 2020 and September 10, 2020. Methods We linked data on the date ultraviolet air purification systems were installed with the Nursing Home COVID-19 Public Health File (weekly data reported by nursing homes on the number of residents with COVID-19 and COVID-19 deaths), public data on data on nursing home characteristics, county level COVID-19 cases/deaths, and outside air temperature. We used an interrupted time series design and ordinary least squares regression to compare trends in weekly COVID-19 cases and deaths before and after installation of ultraviolet air purification systems. We controlled for county level COVID-19 cases, death, and heat index. Results Compared to pre-installation, weekly COVID-19 cases per 1,000 residents (-1.69, 95%CI: -4.32, 0.95) and the weekly probability of reporting any COVID-19 case (-0.02, 95%CI: -0.04, 0.00) declined in the post-installation period. We did not find any difference pre- and post-installation in COVID-19-related mortality (0.00 95%CI: -0.01, 0.02). Conclusions and Implications Our findings from this small number of nursing homes in the southern US demonstrate the potential benefits of air purification in nursing homes on COVID-19 outcomes. Intervening on air quality may have a wide impact without placing significant burden on individuals to modify their behavior. We recommend a stronger, experimental design to estimate the causal effect of installing air purification devices on improving COVID-19 outcomes in nursing homes. Key words nursing homes SARS-CoV-2 COVID-19 indoor air air purification ultraviolet ==== Body pmcFundingsources: This work was supported by RGF Environmental Group, Inc., manufacturer of the air purifiers. The evaluation was conducted independently and the contents are solely those of the authors. We thank RGF for funding; the nursing home corporation that installed the air purifiers for permission to use their installation data; and Smart Air Care, the distributor, for providing those data. The sponsor had no role in the design, methods, data collection, analysis or writing of the manuscript. CONFLICTS OF INTEREST: Joseph M. Braun was financially compensated for services as an expert witness for plaintiffs in litigation related to PFAS-contaminated drinking water. The other authors report no conflicts of interest. Brief summary: Improving indoor air quality is one strategy to reduce the burden of COVID-19 in nursing homes. In this pilot study, we found that installing ultraviolet air purification may reduce COVID-19 cases in nursing homes. Ethical Approval: Because this was a retrospective evaluation and did not involve any resident-level data, it did not require patient consent and was not considered human subjects research or subject to Institutional Review Board approval. Data availability: Public data used in analyses are directly available from the sources or the authors. We linked public data with private data from a multi-facility nursing home corporation in the southern US that gave permission for the distributor (Smart Air Care) of the ultraviolet air purification system that they installed to share facility-level data for this evaluation. An earlier version of the manuscript was published on a preprint server. Jutkowiz E, Shewmaker P, Reddy A, Braun JM, Baier RR. Pilot Study Demonstrates Benefits of Nursing Home Air Purification on COVID-19 Outcomes. medRxiv 2022.12.01.22282978; doi: https://doi.org/10.1101/2022.12.01.22282978 CONFLICTS OF INTEREST XXXXX was financially compensated for services as an expert witness for plaintiffs in litigation related to PFAS-contaminated drinking water. The other authors report no conflicts of interest.
PMC010xxxxxx/PMC10257511.txt	==== Front Clin Chim Acta Clin Chim Acta Clinica Chimica Acta; International Journal of Clinical Chemistry 0009-8981 1873-3492 Elsevier B.V. S0009-8981(23)00244-9 10.1016/j.cca.2023.117442 117442 Discussion Ethical issues in the use of leftover samples and associated personal data obtained from diagnostic laboratories Remes Lenicov Federico a⁎ Fink Nilda E. b1 a Instituto de Investigaciones Biomédicas en Retrovirus y SIDA (INBIRS), Universidad de Buenos Aires / CONICET, Buenos Aires, Argentina b Fundación Bioquímica Argentina, La Plata, Argentina ⁎ Corresponding author at: INBIRS (UBA – CONICET), Piso 11 Paraguay 2155, C1121ABG - Buenos Aires, Argentina. 1 on behalf of the Task Force Ethics of the International Federation of Clinical Chemistry (group members listed in the Acknowledgments) 10 6 2023 10 6 2023 11744224 2 2023 30 5 2023 9 6 2023 © 2023 Elsevier B.V. All rights reserved. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Diagnostic laboratories are an integral part of the research ecosystem in biomedical sciences. Among other roles, laboratories are a source of clinically-characterized samples for research or diagnostic validation studies. Particularly during the COVID-19 pandemic, this process was entered by laboratories with different experience in the ethical management of human samples. The objective of this document is to present the current ethical framework regarding the use of leftover samples in clinical laboratories. Leftover samples are defined as the residue of a sample that has been obtained and used for clinical purposes, and would otherwise be discarded. Secondary use of samples typically demands institutional ethical oversight and informed consent by the participants, although the latter requirement could be exempted when the harm risks are sufficiently small. However, ongoing discussions have proposed that minimal risk is an insufficient argument to allow the use of samples without consent. In this article, we discuss both positions, to finally suggest that laboratories anticipating the secondary use of samples should consider the adoption of broad informed consent, or even the implementation of organized biobanking, in order to achieve higher standards of ethical compliance which would enhance their capacity to fulfill their role in the production of knowledge. Keywords bioethics informed consent leftover sample remnant specimen biobank personal health information ==== Body pmc1 Introduction Diagnostic laboratories, in particular those within medical institutions, are an integral part of the research ecosystem in biomedical sciences. Their participation in research projects is specially appreciated for their proven expertise in the use of validated analytical methods and processes, including the critical aspects of sample identification, traceability and storage; all features that reassure confidence in the results. While this has been true for decades, it gained prominence by the amount and kind of research carried out during the COVID-19 pandemic. The COVID-19 pandemic not only triggered a large number of research projects involving human samples, but it impressed a sense of urgency to produce meaningful results [1], [2]. Diagnostic laboratories naturally became sources of samples for much-needed investigation of the new disease. This process was entered by laboratories with widely heterogeneous backgrounds in research, and thus with diverse expertise in the ethical management of human samples. The objective of this document is to present the current ethical framework regarding the use of leftover samples for research or method validation in clinical laboratories. While most studies will demand institutional ethical oversight and informed consent (IC) by the participants, there are a number of situations in which these could be waived off. This article will provide overview on various ethical concepts to understand these situations, discussing in particular the distinctive features of research carried out in diagnostic laboratories. This article was prepared taking into account recent reviews and presentations on the subject, which capture the current points of concern around the secondary use of leftover samples. Importantly, recommendations outlined in the article have been endorsed by the Task Force Ethics of the International Federation of Clinical Chemistry. Given the global scope intended for the recommendation, the article avoids discussing country-specific regulatory affairs. 2 Leftover samples and associated data Leftover sample can be defined as the residue of a sample that has been obtained and used for clinical purposes, and would otherwise be discarded [3], [4]. Fractions of samples that must be stored due to legal or technical reasons are not considered leftover, at least until the required storage period is over. Leftover samples can also be found in literature as remnant samples or residual biological material. In the context of this document, we will not discuss remnant fractions of samples from individuals enrolled in prospective studies, even if it involved clinically relevant tests, or samples taken from subjects participating in interventional trials. One important point to consider is that the biological fluid or tissue is only one part of what patients entrust to laboratory professionals. In medical laboratories, biological samples are always associated with various kinds of personal data. Associated data involves personal identifiers (name, address, for example), health-related information (chronic diseases, previous hospital stays) or other kinds of potentially sensitive data (travel history, family history). Both the nature and the accessibility of the associated data will be crucial to define the ethical requirements for the use of the samples. In the clinical laboratory, the use of leftover samples could be considered quality control, to generate sample pools or to assemble a panel of control samples [5]. Alternatively, the samples might be useful for research projects, either carried out by external investigators or by investigators affiliated to the lab. There are multiple advantages in the use of leftover samples and their associated data. First, their availability facilitates the advancement of research, particularly the fields of translational and personalized medicine. These samples represent more accurately the study population, since they are not subject to selection bias during recruitment, thus strengthening the validity of the results. Occasionally, leftover samples may represent experimental conditions which would be impossible to produce in prospective studies, such as old samples from untreated patients when a treatment is now available. Finally, from the ethical point of view, research designed to use leftover samples obviates the need to recruit new participants, and therefore avoids exposing them to the concomitant burden and risks. While these are compelling arguments in favor of secondary research on leftover samples, their use must be contemplated with caution. The individual that entered a medical laboratory did so with the specific objective of seeking clinically relevant results. Profound ethical implications derive from the use of their sample for other purposes. 3 Ethical concerns regarding the use of leftover samples Ethical concerns involved in the use of leftover samples can be analyzed in terms of the known principles of respect for autonomy, beneficence and justice. The first concern relates to the commitment to respect the autonomy of the individual providing the sample. This individual approached a health care institution for a clinically relevant test, and the implied consent is valid only for that test. The standard way to deal with the secondary use of this sample would be to ask for permission, usually in the form of an informed consent. There are instances in which it is difficult or impossible to obtain consent for the use of the leftover sample, and sometimes it might not even be considered ethically necessary. Nevertheless, efforts should be done to avoid this situation, given that the informed consent is the most important tool to assert the autonomy of the individual [5]. Another concern relevant to consider the use of leftover samples is the possibility of causing harm to the subject. In other words, the balance between benefits and risks must be satisfactory for the individual involved. In terms of risk assessment, breach of privacy is the most evident and therefore it is crucial to provide for the safety of the associated data. A person whose personal and medical data results exposed becomes at risk of stigmatization, discrimination and overall harm to their integrity. Risks could be related to the exposure of the initially recorded personal information, medical history, and also the newly obtained results. Anonymous samples (those with no recorded personal data) are hardly the case in medical laboratories, but the breach of privacy risk can be minimized when leftover samples are irreversible separated from identity data. Mitigation of the risk of harming an individual is a decisive issue regarding the possibility of using a leftover sample without asking for consent, as discussed in the next section. There is also the possibility of finding a clinically relevant result, other than the ones that those the patient came looking for; also known as incidental results. Risk of causing distress and anxiety to subjects derive from the possibility of notifying them unsolicited information regarding their identity, their health or even the health of their relatives. Management of this sensitive information obtained from leftover samples should be planned. Verification and notification of incidental results should be analyzed in terms of their clinical importance and the existence of a beneficial intervention [6], [7]. When the use of the leftover sample is planned, a question can be added to the consent process so that the researcher is aware whether the individual would like to know any unexpected result. This particular risk is also avoided when using irreversibly de-identified samples, so that it becomes impossible to assign the incidental result to any one subject. Importantly, another risk to avoid is the loss or deterioration of the sample itself as a consequence of its secondary use. For example, the performance of the clinically relevant test must not be jeopardized to save enough material for future research studies. In this regard, it is not Another related risk to be avoided is the inclination to extract more sample (for example, draw more blood volume) than what would be normally required for the diagnostic process. Hence, it is good standard practice in the laboratory to define the amount of sample before the extraction, as opposed to during the extraction. A word of caution must be said regarding arguments of beneficence to justify the use of the samples for research. It is incorrect to argue for the use of a leftover sample, either with or without consent, only by citing a “greater good” derived from the results, which will -eventually- benefit the human kind. This line of reasoning is dangerous, as resonant cases have proven before [8], [9]. Still, public interest in the area of health care is frequently cited as an argument for informed consent exemptions, a view that has been reinforced by sample usage and data sharing for the developing diagnostic tools, therapies and vaccines during the COVID-19 pandemic [10]. Finally, there are some considerations regarding the principle of justice. The selection of the samples must be based on technical criteria. Picking samples from vulnerable populations should be avoided. Vulnerable populations are defined as those including persons who have their autonomy already restricted by any other condition (imprisonment for example), or otherwise that it may be regarded that their eventual consent would not be valid (for example, because they are not able to correctly understand what is being asked for). If the objective of a research project is to study a vulnerable population, then the wiser option is to prepare a prospective study and obtain the specific informed consent. 3.1 Rationale supporting the possible exemptions to the requirement of informed consent (IC) When use of leftover samples is proposed for a project not planned at the time of sample obtention, consent can often be requested by contacting each subject, since laboratories usually retain personal contact information. Nonetheless, sometimes it could be impractical or impossible to look for consent. For instance, a research project may require to use a historical collection of samples stored from long ago. Samples may have lost, or been stored without, identification labels. Some projects may involve the re-analysis of thousands of records. In these and many other occasions, there are arguments to carry out the project considering exemptions to the IC requirement. First, it is important to establish that IC exemption does not mean lack of ethical oversight. Approval by an independent committee will still be needed, or at least advised, for the use of the samples and their associated data without individual consent. The rationale behind exemption of the IC requirement is that harm risks are sufficiently small as to overcome the loss of autonomy. Minimal risk is a reasonable assumption given that there is no interaction between researcher and study subject, and there are no physical or psychological interventions on the subject, who incurs in no extra burdens or activities by participating in the study. The main risk is related to disclosure of personal information leading to unintended negative consequences for the individual. Any such risk is certainly avoided when working with anonymous samples, in which identifiers were not recorded, they were erased irreversibly or, more rarely, the identity was never known. For example, this could entail transferring an aliquot of leftover blood to a new tube re-labeled with a non-linked code. Another example of this situation would be the pooling of samples for quality control. Method verification also poses no risks. Furthermore, in this case it can be argued that the implied consent by the patient includes repetition of the test, by the same or other technique. This practice also carries no risk of incidental findings, since the determination has been consented. Use of leftover samples under these conditions of minimal or absent risk can be considered to not require informed consent from the sample donor. Of note, special attention must be taken when the investigation will produce genomic data. Re-identification of subjects has been shown to be possible in genetic studies [11], [12], [13]. Indeed, risk assessment guides usually indicate the obtention of specific informed consent when the expected results of the secondary use involve this kind of data. On the other hand, clinical laboratories more likely retain personal records associated to samples. These samples can still be used for secondary research without IC, provided they are rendered non-identifiable before handing them to the investigator. Making a sample non-identifiable means dissociating the identifiers (name, address, contact information, personal ID number, and so on) from the biological specimen and associated databases, so that it would be not possible to ascertain the identity of any study participant. This process does not necessarily involve irreversible removal of information. Sometimes it is possible to maintain a link back to the personal information (coded samples). In such cases, risk must be minimized by setting up procedures making clear that the laboratory cannot share the key with the investigator who will use the leftover samples, and the investigator must not intend to re-identify them. Still, not all information needs to be concealed from the investigator. At an absolute minimum, for any such leftover sample to be useful, age and sex from the source individual must be known. Other pieces of information might be added as well, such as ethnicity, nationality or any georeference that allows localization of the origin of the sample (although not the personal address). However, more information increases risks. First, the possibility of re-identification of samples should be considered. Re-identification of individual subjects has been achieved using data from public sources, although this has only been done by experts in the field as a proof-of-principle that it be possible[13], [14], [15]. Secondly, by retaining demographic data to samples, there is the risk of singling out a population or group of people. For instance, results could show that a group is at high risk of having a genetic variant associated with a particular disease [16]. This situation underscores the advice to be thoughtful when working with vulnerable populations. Despite these risks, ethics committees are likely to grant IC exemptions when using leftover samples dissociated from personal information. In addition, clinical laboratories are used to deal with confidential information, and rely on procedures to warrant medical secrecy. Nonetheless, IC exemption would be difficult to claim when the project is managed by an investigator affiliated to the clinical laboratory, who likely has access to the laboratory database and the personal identification of the people included in the study. This situation does not comply with the requirement that the investigator cannot ascertain the identity of the subjects. In such a case, IC should be asked from the patient at some point. This would be preferably before the sample is obtained, but not necessarily, since it should be clear that the health care service (i.e. the clinically relevant test) will not be altered regardless of the patient’s consent to participation in research. 3.2 A debate regarding the relative weight of participant’s autonomy Current views on the requirement of informed consent for the secondary use of leftover samples follow the arguments discussed in the previous sections. Accordingly, rules and recommendations at the national or international level emphasize the importance of obtaining informed consent from research participants for any kind of study design, but usually incorporate exemptions based on the practicality of obtaining the IC, always on the condition of minimal harm risk [17], [18]. However, ongoing discussions have put forward the idea that minimal risk is an insufficient argument to allow the use of samples without consent [19], [20]. Besides the risk of breach of privacy, other ways of harming the individual should be considered in the case they find out their samples had been used without consent. For instance, some people may object the use of their bodily fluid on religious or other personal grounds. Some people may not have objections with the use of the sample per se, but instead may not want to contribute with research towards a certain aim. What is more, disclosure of unconsented use of specimens has led in the past to public scandals, such as the resonant cases of the development of HeLa cell line, the studies on Havasupai tribe samples and the organ collection at Alder Hey Hospital [16], [21], [22]. Admittedly, in these cases there were major deviations from ethical procedures, but they nonetheless reflect delicate public feelings regarding the use of remnant specimens. Investigators and laboratory managers should be warned to mind not only regulatory compliance but also expectations from individuals providing samples. Regarding people’s expectations, extensive empirical research shows that people prefer to decide whether to contribute the remnant of their samples to research [23], [24], [25]. Interestingly, survey participants had opinions on which topics they would rather refuse to contribute, such human cloning, studies involving indigenous populations and research with potential economic profit [26]. All in all, it appears that protocols for the use of leftover samples will incorporate these arguments in the future. Currently, secondary research and method development at clinical laboratories are mostly functioning under protocols based on IC exemption, but this practice should be revised in proposals to update regulations. Ideally, proposed approaches to incorporate informed consent for the use of leftover samples should be planned to avoid imposing restrictive burdens and costs, and also maintain some flexibility on research objectives. 3.3 The process of informed consent for secondary use of leftover samples The process of informed consent is the main tool to materialize the respect for people’s autonomy. In the case of leftover samples, the person should be asked whether, after the clinically relevant determinations are completed, the remnant material could be used for additional purposes, instead of being discarded. Since the secondary uses are not known at the time of obtaining the sample, the informed consent does not include project-specific information. Instead, the secondary use of leftover samples can be covered by informing the patient only about the possibility of a future use. Although still debated, broad consents should not be considered inferior to project-specific consents. In fact, severe criticism has been raised against the traditional project-specific written informed consent process. Evidence has accumulated indicating that traditional consent documents, which struggle to incorporate a series of required statements, are not understood, or even not read at all [17], [27]. In this sense, studies have found that readability of consent documents to be unacceptably low [28], [29]. For instance, recent reports showed that consent for COVID-19 vaccine trials were long and difficult [28], [30]. In addition, this situation will likely worsen, since legal requirements demand more disclaimer phrases to be incorporated into consents [28]. The experience of biobanks and other sample collections has promoted the development of alternative models in the search for IC [31], [32]. In biobanks, volunteer donors provide a broad consent open for a scope of secondary uses, they are asked what to do with potential incidental findings and they are informed that they can ask for withdrawal of samples and data at any point in the future. While it has been argued that broad consent fails to provide information to the participant, it actually represents a consent to sample governance, a consent to a set of rules on how the samples will be managed [33]. Broad consents ask the individual to decide if they would like to collaborate with research, provided that certain rules are met. Furthermore, evidence showed that broad consents have been perceived positively by study participants [34], [35]. To be sure, it should be noted that current or future projects making use of those samples would still have to be approved by an independent ethics committee. Clinical laboratories could use the experience of biobanks and find that asking broad consent at the time of sample collection could be a reasonable way to comply with ethical requirements [36], [37]. Whereas consent for clinical care tests is often implied (i.e., it is not standard procedure to request signed consent before diagnostic sample extraction), consent for secondary use of samples should be written and explicit. Admittedly, this implies a modification in the procedures that are familiar to anyone visiting a diagnostic laboratory. While it is uncertain how the public would react, surveys and studies indicate that people are usually prone to collaborate [26], [35], [37]. It is also debatable whether applying a broad consent system would require large changes in infrastructure and procedures, since clinical laboratories already have the capacity to store samples, safeguard specimens and data, and implement new protocols. Eventually, it could be expected that laboratories and the public become used to a new norm. 4 Final words Our most important mandate at the clinical lab is to provide a result of assured quality. But clinical laboratories also play a larger role in the medical and scientific community, a role in the production of knowledge. To this end, one of the assets of clinical laboratories is the expert handling of human samples, a practice that entails not only technical proficiency but also ethical responsibility. Regarding the latter, it is clear that interventional research requires the most stringent and specific consent from participants. On the other hand, observational studies which carry minimal risks might make use of a different set of procedures that still respect and protect the rights of the patients. While the use of leftover samples without consent can be ethically acceptable on occasions, there is an ongoing discussion regarding the relative weight of the participant’s right to know and consent. It is possible that participant’s involvement will gain more consideration, even if it means that the ability to carry out some research projects might be affected. In this scenario, participant’s involvement in the use of leftover samples and their associated data could take the form of broad consents obtained during clinical care. We believe it would be convenient that clinical laboratory management be aware of, and ideally transmit to their institutions, the existence of novel alternatives of ethical compliance which would enhance their capacity to fulfill their role in the production of new knowledge. Recommendations for investigators and management of medical laboratories (box text)● If the laboratory participates in research (including method validation and development), formalize the activity by having written protocols and training the personnel accordingly. Request a copy of the protocols that were approved by an ethics committee. ● Clearly define the scope of any project initiated by the lab or accorded with a third party. Anticipate if it will be possible to ask for an exemption of informed consent. ● Define the people involved and who has access to personal data of the participants. Establish a base protocol for the preparation and coding of samples to be given to a third party. ● Be aware of potential ethical issues to anticipate problems, especially when participating in projects involving genetic research, reproductive tissue samples, or vulnerable populations. ● It is a sound policy to always look for ethical review by an independent committee, even if it may seem unnecessary. ● Consider novel alternatives such as broad consent to comply with autonomy rights of research subjects, and establish the proper protocols to that end. ● If the scale of the sample gathering is too big, if it is sustained for too long, or across multiple projects, laboratory should consider the association with a biobank, or even the creation of one. The authors did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors for the preparation of this manuscript. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability No data was used for the research described in the article. Acknowledgements We are grateful to Dr. Richard Davey, Dr. Sudip Kumar Datta, Dr. Joseph Wiencek, and Dr. Julian Verona (members of the Task Force Ethics of the International Federation of Clinical Chemistry) for their constructive comments to improve the manuscript. ==== Refs References 1 Park J.J.H. Mogg R. Smith G.E. Nakimuli-Mpungu E. Jehan F. Rayner C.R. Condo J. Decloedt E.H. Nachega J.B. Reis G. Mills E.J. How COVID-19 has fundamentally changed clinical research in global health Lancet Glob Health. 9 2021 10.1016/S2214-109X(20)30542-8 2 Aviv-Reuven S. Rosenfeld A. Publication patterns’ changes due to the COVID-19 pandemic: A longitudinal and short-term scientometric analysis Scientometrics. 126 2021 10.1007/s11192-021-04059-x 3 Allen M.J. Powers M.L.E. Gronowski K.S. Gronowski A.M. Human tissue ownership and use in research: What laboratorians and researchers should know Clin Chem. 56 2010 10.1373/clinchem.2010.150672 4 FDA, Guidance on Informed Consent for In Vitro Diagnostic Device Studies Using Leftover Human Specimens that are Not Individually Identifiable, 2006. 5 Kottow M. The battering of informed consent J Med Ethics. 30 2004 10.1136/jme.2003.002949 6 Ells C. Thombs B.D. The ethics of how to manage incidental findings CMAJ. Canadian Medical Association Journal. 186 2014 10.1503/cmaj.140136 7 Beshir L. A framework to ethically approach incidental findings in genetic research Electronic Journal of the International Federation of Clinical Chemistry and Laboratory Medicine. 31 2020 8 Reverby S.M. Ethical Failures and history lessons: The U.S. public health service research studies in tuskegee and guatemala Public Health Rev. 34 2012 10.1007/bf03391665 9 Krugman S. The willowbrook hepatitis studies revisited: Ethical aspects Rev Infect Dis. 8 1986 10.1093/clinids/8.1.157 10 Mahomed S. Staunton C. Ethico-legal analysis of international sample and data sharing for genomic research during COVID-19: A South African perspective BioLaw Journal. 2021 10.15168/2284-4503-785 11 von Thenen N. Ayday E. Cicek A.E. Re-identification of individuals in genomic data-sharing beacons via allele inference Bioinformatics. 35 2019 365 371 10.1093/bioinformatics/bty643 30052749 12 M. Shabani, L. Marelli, Re‐identifiability of genomic data and the <scp>GDPR</scp>, EMBO Rep. 20 (2019). 10.15252/embr.201948316. 13 M. Gymrek, A.L. McGuire, D. Golan, E. Halperin, Y. Erlich, Identifying personal genomes by surname inference, Science (1979). 339 (2013). 10.1126/science.1229566. 14 el Emam K. Jonker E. Arbuckle L. Malin B. A systematic review of re-identification attacks on health data PLoS One. 6 2011 10.1371/journal.pone.0028071 15 Packhäuser K. Gündel S. Münster N. Syben C. Christlein V. Maier A. Deep learning-based patient re-identification is able to exploit the biometric nature of medical chest X-ray data Sci Rep. 12 2022 14851 10.1038/s41598-022-19045-3 36050406 16 Drabiak-Syed K. Lessons from havasupai tribe v. arizona state university board of regents: Recognizing group, cultural, and dignitary harms as legitimate risks warranting integration into research praCtice J Health Biomed Law. 6 2010 17 Nijhawan L. Janodia M. Muddukrishna B. Bhat K. Bairy K. Udupa N. Musmade P. Informed consent: Issues and challenges J Adv Pharm Technol Res. 4 2013 10.4103/2231-4040.116779 18 Borovecki A. Mlinaric A. Horvat M. Smolcic V.S. Informed consent and ethics committee approval in laboratory medicine Biochem Med (Zagreb) 28 2018 10.11613/BM.2018.030201 19 Hofmann B.M. Bypassing consent for research on biological material Nat Biotechnol. 26 2008 10.1038/nbt0908-979b 20 Lynch H.F. Wolf L.E. Barnes M. Implementing regulatory broad consent under the revised common rule: Clarifying key points and the need for evidence, journal of law Medicine and Ethics. 47 2019 10.1177/1073110519857277 21 Burton J.L. Wells M. The alder hey affair Arch Dis Child. 86 2002 10.1136/adc.86.1.4 22 Beskow L.M. Lessons from HeLa cells: The ethics and policy of biospecimens Annu Rev Genomics Hum Genet. 17 2016 10.1146/annurev-genom-083115-022536 23 Brothers K.B. Morrison D.R. Clayton E.W. Two large-scale surveys on community attitudes toward an opt-out biobank Am J Med Genet A. 155 2011 10.1002/ajmg.a.34304 24 Chen D.T. Rosenstein D.L. Muthappan P. Hilsenbeck S.G. Miller F.G. Emanuel E.J. Wendler D. Research with stored biological samples: What do research participants want? Arch Intern Med. 165 2005 10.1001/archinte.165.6.652 25 Mezuk B. Eaton W.W. Zandi P. Participant characteristics that influence consent for genetic research in a population-based survey: The baltimore epidemiologic catchment area follow-up Community Genet. 11 2008 10.1159/000113880 26 Grady C. Eckstein L. Berkman B. Brock D. Cook-Deegan R. Fullerton S.M. Greely H. Hansson M.G. Hull S. Kim S. Lo B. Pentz R. Rodriguez L. Weil C. Wilfond B.S. Wendler D. Broad consent for research with biological samples: Workshop conclusions American Journal of Bioethics. 15 2015 10.1080/15265161.2015.1062162 27 Jacquier E. Laurent-Puig P. Badoual C. Burgun A. Mamzer M.F. Facing new challenges to informed consent processes in the context of translational research: The case in CARPEM consortium BMC Med Ethics. 22 2021 10.1186/s12910-021-00592-9 28 Emanuel E.J. Boyle C.W. Assessment of length and readability of informed consent documents for COVID-19 vaccine trials JAMA Netw Open. 4 2021 10.1001/jamanetworkopen.2021.10843 29 Pietrzykowski T. Smilowska K. The reality of informed consent: empirical studies on patient comprehension—systematic review Trials. 22 2021 10.1186/s13063-020-04969-w 30 Bothun L.S. Feeder S.E. Poland G.A. Readability of participant informed consent forms and informational documents: From phase 3 COVID-19 vaccine clinical trials in the united states Mayo Clin Proc. 96 2021 10.1016/j.mayocp.2021.05.025 31 Annaratone L. de Palma G. Bonizzi G. Sapino A. Botti G. Berrino E. Mannelli C. Arcella P. di Martino S. Steffan A. Daidone M.G. Canzonieri V. Parodi B. Paradiso A.V. Barberis M. Marchiò C. Basic principles of biobanking: From biological samples to precision medicine for patients Virchows Archiv. 479 2021 10.1007/s00428-021-03151-0 32 Gefenas E. Lekstutiene J. Lukaseviciene V. Hartlev M. Mourby M. Cathaoir K. Controversies between regulations of research ethics and protection of personal data: Informed consent at a cross-road Med Health Care Philos. 25 2022 10.1007/s11019-021-10060-1 33 Boers S.N. van Delden J.J.M. Bredenoord A.L. Broad Consent Is Consent for Governance American Journal of Bioethics. 15 2015 10.1080/15265161.2015.1062165 34 Masiye F. Jaoko W. Rennie S. Stakeholder views on informed consent models for future use of biological samples in malawi and south africa BMC Med Ethics. 24 2023 4 10.1186/s12910-023-00882-4 36658544 35 Warner T.D. Weil C.J. Andry C. Degenholtz H.B. Parker L. Carithers L.J. Feige M. Wendler D. Pentz R.D. Broad consent for research on biospecimens: The views of actual donors at four U.S. medical centers, journal of empirical research on human research Ethics. 13 2018 10.1177/1556264617751204 36 Helgesson G. In defense of broad consent Cambridge Quarterly of Healthcare Ethics. 21 2012 10.1017/S096318011100048X 37 Garrison N.A. Sathe N.A. Antommaria A.H.M. Holm I.A. Sanderson S.C. Smith M.E. McPheeters M.L. Clayton E.W. A systematic literature review of individuals’ perspectives on broad consent and data sharing in the united states Genetics in Medicine. 18 2016 10.1038/gim.2015.138
PMC010xxxxxx/PMC10274864.txt	==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 10.1101/2023.06.08.544231 preprint 1 Article Astrocyte store-operated calcium entry is required for centrally mediated neuropathic pain Prokhorenko Mariya A. Smyth Jeremy T. http://orcid.org/0000-0001-7809-2193 09 6 2023 2023.06.08.544231http://biorxiv.org/lookup/doi/10.1101/2023.06.08.544231 nihpp-2023.06.08.544231.pdf Abstract Central sensitization is a critical step in chronic neuropathic pain formation following acute nerve injury. Central sensitization is defined by nociceptive and somatosensory circuitry changes in the spinal cord leading to dysfunction of antinociceptive gamma-aminobutyric acid (GABA)ergic cells (Li et al., 2019), amplification of ascending nociceptive signals, and hypersensitivity (Woolf, 2011). Astrocytes are key mediators of the neurocircuitry changes that underlie central sensitization and neuropathic pain, and astrocytes respond to and regulate neuronal function through complex Ca 2+ signaling mechanisms. Clear definition of the astrocyte Ca 2+ signaling mechanisms involved in central sensitization may lead to new therapeutic targets for treatment of chronic neuropathic pain, as well as enhance our understanding of the complex central nervous system (CNS) adaptions that occur following nerve injury. Ca 2+ release from astrocyte endoplasmic reticulum (ER) Ca 2+ stores via the inositol 1,4,5-trisphosphate receptor (IP 3 R) is required for centrally mediated neuropathic pain (Kim et al, 2016); however recent evidence suggests the involvement of additional astrocyte Ca 2+ signaling mechanisms. We therefore investigated the role of astrocyte store-operated Ca 2+ entry (SOCE), which mediates Ca 2+ influx in response to ER Ca 2+ store depletion. Using an adult Drosophila melanogaster model of central sensitization based on thermal allodynia in response to leg amputation nerve injury (Khuong et al., 2019), we show that astrocytes exhibit SOCE-dependent Ca 2+ signaling events three to four days following nerve injury. Astrocyte-specific suppression of Stim and Orai, the key mediators of SOCE Ca 2+ influx, completely inhibited the development of thermal allodynia seven days following injury, and also inhibited the loss of ventral nerve cord (VNC) GABAergic neurons that is required for central sensitization in flies. We lastly show that constitutive SOCE in astrocytes results in thermal allodynia even in the absence of nerve injury. Our results collectively demonstrate that astrocyte SOCE is necessary and sufficient for central sensitization and development of hypersensitivity in Drosophila , adding key new understanding to the astrocyte Ca 2+ signaling mechanisms involved in chronic pain. ==== Body pmc
PMC010xxxxxx/PMC10286518.txt	==== Front Nefrologia Nefrologia Nefrologia 0211-6995 1989-2284 Sociedad Española de Nefrología. Published by Elsevier España, S.L.U. S0211-6995(23)00099-1 10.1016/j.nefro.2023.06.007 Article Experiencia en vida real con terapias frente a COVID-19 leve-moderado en trasplantados renales: ¿Cómo tratar a partir de ahora a los pacientes con enfermedad renal crónica? Real-world experience with mild-moderate COVID-19 therapies in kidney transplant patients: How to treat patients with chronic kidney disease from now on?Alonso Marta 1 Villanego Florentino 1 Vigara Luis Alberto 1 Rodríguez María Eugenia 3 Eady Myriam 2 García Ana 1 Mínguez María Carmen 1 Montero María Elisa 1 Segurado Oscar 1 García Teresa 1 Mazuecos Auxiliadora 1⁎ 1 Servicio de Nefrología, Hospital Universitario Puerta del Mar, Cádiz, España 2 Servicio de Nefrología, Hospital Universitario de Jerez, Cádiz, España 3 Servicio de Farmacia Hospitalaria, Hospital Universitario Puerta del Mar, Cádiz, España ⁎ Autor de correspondencia: Servicio de Nefrología, Hospital Puerta del Mar, Av. Ana de Viya 21, 11009 Cádiz, España 22 6 2023 22 6 2023 12 6 2023 © 2023 Sociedad Española de Nefrología. Published by Elsevier España, S.L.U. 2023 Sociedad Española de Nefrología Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc
PMC010xxxxxx/PMC10286519.txt	==== Front Contemp Clin Trials Contemp Clin Trials Contemporary Clinical Trials 1551-7144 1559-2030 Published by Elsevier Inc. S1551-7144(23)00194-5 10.1016/j.cct.2023.107271 107271 Article Recruiting experiences of NIH-funded principal investigators for community-based health behavior interventions during the COVID-19 pandemic Seguin-Fowler Rebecca A. a⁎ Demment Margaret b Folta Sara C. c Graham Meredith b Hanson Karla d Maddock Jay E. e Patterson Megan S. f a Institute for Advancing Health through Agriculture (IHA), Department of Nutrition, College of Agriculture and Life Sciences, Texas A&M University System, 1500 Research Parkway, Centeq Building B, College Station, TX 77845, United States of America b Texas A&M AgriLife Research and Extension Center, 17360 Coit Rd, Dallas, TX 75252, United States of America c Friedman School of Nutrition Science and Policy, Tufts University, 150 Harrison Ave, Boston, MA 02111, United States of America d Department of Public & Ecosystem Health, Cornell University, Ithaca, NY 4853, United States of America e Department of Environmental and Occupational Health, Texas A&M University, College Station, TX 77843, United States of America f Department of Health Behavior, Texas A&M University, 1266 TAMU, College Station, TX 77843, United States of America ⁎ Corresponding author. 22 6 2023 22 6 2023 10727110 2 2023 8 6 2023 20 6 2023 © 2023 Published by Elsevier Inc. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Successful recruitment into randomized trials and interventions is essential to advance scientific knowledge to improve health. This rapid assessment study explored how the COVID-19 pandemic affected participant recruitment overall, identified how it exacerbated existing challenges to recruit hard-to-reach populations, and described how NIH-funded Principal Investigators (PIs) responded to COVID-era recruitment challenges. A cross-sectional survey of NIH-funded PIs conducting interventions and trials related to health behaviors was conducted in 2022. The survey was completed by 52 PIs, most of whom were highly experienced in this type of research. Eighteen PIs reported it was very difficult to recruit participants now (39.1%) compared to before COVID-19 when only one did (2.2%). PIs reported changing recruitment and data collection methods (29.4%), increasing staff dedicated to recruitment (29.4%), and increasing participant compensation (23.5%). Recruitment methods shifted from in-person activities to social media and other electronic communications. Barriers to recruitment included reluctance to participate in research, COVID-19 protocols and precautions, overwhelmed community partners, staff burnout and turnover, and limited access to technology for some populations that were already hard to reach. Facilitators to recruitment consisted of increased access and ability to use remote technologies, use of social media, strong community ties, and wanting to be part of something positive. PIs perceived recruitment as much more difficult after the onset of COVID-19, though research teams were able to pivot to more online and remote options. These tools may have a lasting impact in modernizing recruitment, data collection, and intervention techniques in future trials. Keywords Randomized trials COVID-19 Physical activity Nutrition Recruitment Abbreviations COVID Coronavirus disease NIH National Institutes of Health PI Principal investigator RQA Rapid Qualitative Analysis ==== Body pmcData availability Data will be made available on request.
PMC010xxxxxx/PMC10286520.txt	==== Front Pediatr Neonatol Pediatr Neonatol Pediatrics and Neonatology 1875-9572 2212-1692 Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. S1875-9572(23)00107-9 10.1016/j.pedneo.2023.05.002 Review Article Pathophysiological and clinical point of view on Kawasaki disease and MIS-C Vaňková Lenka ab∗ Bufka Jiří c Křížková Věra ab a Department of Histology and Embryology, Faculty of Medicine in Pilsen, Pilsen b Charles University in Prague, Czech Republic c Department of Paediatrics, Teaching Hospital in Pilsen, Pilsen, Czech Republic ∗ Corresponding author. , Department of Histology and Embryology, Faculty of Medicine in Pilsen, Charles University in Prague, , alej Svobody 76, Pilsen, 323 00, Czech Republic Tel.: +420702 063 861 22 6 2023 22 6 2023 22 11 2022 24 4 2023 19 5 2023 © 2023 Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. 2023 Taiwan Pediatric Association Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. This article compares two important pathophysiological states, Kawasaki disease, and multisystem inflammatory syndrome, in children associated with COVID-19 (MIS-C). Both occur predominantly in children, have a temporal association with an infectious agent, and are associated with immune-system alteration and systemic inflammation under certain circumstances. The two share common pathophysiology, including enhancement of interleukin-1 neutrophils, activation of the inflammasome, pyroptosis, or NETosis. Moreover, the clinical presentation of the diseases overlaps. However, they are indeed two separate diseases, proven by the differences in the epidemiological and etiological aspects and the pathophysiological processes involved in the development and frequency of some clinical signs. This article highlights potentially exciting areas that have not yet been studied in detail, which could help better understand the development of these diseases. Keywords COVID-19 Kawasaki disease MIS-C ==== Body pmc
PMC010xxxxxx/PMC10286521.txt	==== Front Acad Pediatr Acad Pediatr Academic Pediatrics 1876-2859 1876-2867 Published by Elsevier Inc. on behalf of Academic Pediatric Association S1876-2859(23)00226-7 10.1016/j.acap.2023.06.020 Article Changes in Positive Childhood Experiences during the COVID-19 pandemic Crouch Elizabeth PhD ⁎ Radcliff Elizabeth PhD Probst Janice PhD Rural and Minority Health Research Center, Arnold School of Public Health, University of South Carolina, 220 Stoneridge Drive, Suite 204, Columbia, SC, 29210, USA ⁎ Correspondence to: Rural and Minority Health Research Center, Arnold School of Public Health, University of South Carolina, 220 Stoneridge Drive, Suite 204, Columbia, SC, 29210 22 6 2023 22 6 2023 5 5 2023 16 6 2023 © 2023 Published by Elsevier Inc. on behalf of Academic Pediatric Association. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objectives Changes in family life associated with COVID-19 precautions may have reduced children’s access to positive childhood experiences (PCEs). The purpose of this study is to examine the prevalence of PCEs before and during the COVID-19 pandemic among school age children. Methods This cross-sectional study used data from the 2018-2019 National Survey of Children’s Health (NSCH, n=42,464) and the 2020-2021 NSCH (n=54,256) to examine the pre-pandemic period (June 2018 to January 2020) and compared results to information obtained during the early pandemic period (June 2020 through January 2022), using bivariate analyses and Z-tests. Results Positive childhood experiences declined in four of the seven PCEs measured, from 2018-2019 to 2020-2021: after school activities, community volunteerism, guiding mentor, and resilient family, with all differences significant by p<0.0001. After school activities decreased from 79.8% to 72.2%, community volunteering decreased from 43.9% to 35.1%, guiding mentor decreased from 88.8% to 86.3%, and resilient family decreased from 92.7% to 84.6%. PCEs increased for safe neighborhood (64.7% to 67.2%), supportive neighborhood (55.8% to 57.5%), and connected caregiver (65.3% to 94.7%). Conclusions As children have experienced higher levels of parental stress and disruption during their lives during the COVID-19 pandemic, policy makers and program makers must find ways to increase exposure to PCEs following the pandemic. The quantification of these PCEs is a great start, with further research needed to describe ways that schools and community organizations have found to expose children to PCEs in safe ways. Keywords positive childhood experiences child development COVID-19 pandemic Abbreviations PCE positive childhood experienceData Resource Center for Child and Adolescent Health NSCH National Survey of Children’s Health ==== Body pmc What’s New Exposure to many PCEs, including after school activities, community volunteerism, guiding mentor, and connected caregiver, declined during the COVID-19 pandemic. These findings further inform educators and policymakers on ways to increase exposure to PCEs during the ongoing pandemic. Introduction The COVID-19 pandemic disrupted children’s lives in the U.S. across multiple contexts. School and daycare closures were common to reduce the spread of the disease.1, 2, 3 Children residing in under resourced environments may have found it difficult to pivot to online schooling, given the challenges of inequitable access to broadband due to availability or cost.4 Social distancing, home schooling, and lockdowns may have been challenging for families, as caregivers worked from home, were laid off, or rotated childcare shifts with shift work.5 Changes in family life associated with COVID-19 precautions may have reduced children’s access to positive childhood experiences (PCEs). PCEs, which include nurturing safe, stable relationships, peer to peer social interaction, constructive social engagement, and social emotional competencies,6 have been shown to prevent, reduce, and mitigate the effects of adverse childhood experiences.7 PCEs help to build resiliency and improve social emotional development in children, which may have long lasting effects on children into adulthood.8 PCEs occur not just at the individual or household level, but also may be experienced in schools and community settings. To the extent that these experiences were curtailed during the COVID-19 pandemic, children’s social and emotional development may have been impaired. The analyses reported here examine the prevalence of PCEs before and during the COVID-19 pandemic among school age children, using nationally representative information from the National Survey of Children’s Health (NSCH). We examined the pre-pandemic period (June 2018 to January 2020) and compared results to information obtained during the pandemic period (June 2020 through January 2022). Methods Data Source Cross-sectional data for the National Survey of Children’s Health (NSCH) was used for this study. The NSCH is a nationally representative survey of parents or caregivers residing in households with at least one child between the ages of 0 and 17. One child from each household is randomly selected to be the subject of the survey, with the survey administered by the U.S. Census Bureau, using both mail-in and online collection methods. Further information on selection and sampling methodology can be found on the DRC website (childhealthdata.org). Analysis was restricted to children 6 years of age and older, since many PCEs are associated with after-school activities that would not be relevant to younger children. Population The 2018-2019 NSCH is a combined two-year data set, combining the 2018 NSCH, with data collection from June 2018 to January 2019, and the 2019 NSCH, which collected data from June 2019 to January 2020. There were 42,464 observations of children age six or older with complete PCE and demographic information. The 2020-2021 NSCH data collection occurred from June 2020 to January 2021 (2020 NSCH) and June 2021 to January 2022 (2021 NSCH), with 54,256 observations of children age six or older in with complete PCE and demographic information. Based on prior research, PCEs were measured using seven questions across four categories from the NSCH.9, 10, 11 The four PCE domains were 1) being in nurturing and supportive relationships; 2) living in safe, stable, and equitable environments; 3) having opportunities for constructive social engagement; and 4) learning social and emotional competencies. To measure being in nurturing and supportive relationships, NSCH questions about mentorship, family resilience, and family communication were used. For mentorship, caregivers of the child are asked “other than you or other adults in your home, is there at least one adult in this child’s school, neighborhood, or community who knows this child well and who he or she can rely on for advice or guidance?” If the response was yes, the child was deemed as being in nurturing, supportive relationships. Family resilience was measured using the NSCH composite measure by the question: “when your family faces problems, how often are you likely to do each of the following?” 1) “talk together about what to do,” 2) “work together to solve our problems,” 3) “know we have strengths to draw on,” and 4) “stay hopeful even in difficult times.” Response choices included none of the time, some of the time, most of the time, or all of the time. If the caregiver responded with all or most of the time to all four items, the child was categorized as living in a household with family resilience. For the category of living and developing in safe, stable, equitable environments, NSCH questions about the child’s neighborhood and community were used. Supportive environment was addressed by questions about social interactions in the neighborhood: “To what extent do you agree with these statements about your neighborhood or community…” 1) “people in this neighborhood help each other out,” 2) “we watch out for each other’s children in this neighborhood,” and 3) “when we encounter difficulties, we know where to go for help in our community.” Possible responses included definitely agree, somewhat agree, somewhat disagree, or definitely disagree. Children were categorized as living in a supportive neighborhood if caregivers reported “definitely agree” to at least one of the items above and “somewhat agree” or “definitely agree” to the other two items. Neighborhood safety was measured with a single question: “To what extent do you agree [that]… the child is safe in our neighborhood.” A response of “definitely agree” was then coded as a safe neighborhood. Opportunities for constructive social engagement were measured using NSCH questions about participation in after school activities and volunteerism during the past 12 months. Caregivers were asked whether the child participated in 1) “a sports team” or “take sports lessons after school or on weekends,” 2) “any clubs or organizations after school or on weekends,” or 3) “any other organized activities or lessons, such as music, dance, language, or arts?” If a child participated in one or more extracurricular activities, the child was deemed as participating in an after-school activity. For volunteerism, the following question was used: “During the past 12 months did the child participate in any type of community service or volunteer work at school, place of worship, or in the community?” If the answer to this was “yes”, the child was deemed as having volunteered in their community, school, or church. Learning social and emotional competencies were assessed using the NSCH question “how well can you and this child share ideas or talk about things that really matter?” Response choices include very well, somewhat well, not very well, or not very well at all. These responses were categorized into the following two categories: 1) very well or 2) somewhat well to not very well or not very well at all. Demographic characteristics of the child were included to describe the sample (age, gender, race/ethnicity, and if the child had special health care needs). Race/ethnicity had four groups: Non-Hispanic White, Non-Hispanic Black, Hispanic, and Non-Hispanic Other. Special healthcare needs was dichotomized into yes/no, using the NSCH special healthcare needs tool, which asks about use of prescription medication, functional limitations, elevated use of services, specialized therapy, and ongoing developmental, emotional, or behavioral conditions. Family and caregiver characteristics were also chosen to describe the sample: primary language spoken in the household, the highest level of educational attainment of a parent or guardian in the household, family structure, poverty/income status, health insurance status, and geographic residence. Primary language spoken in the home was classified as English or not English. Educational attainment was categorized into less than or equal to a high school degree/GED or at least some college education or more. Family structure had four categories: two parents, currently married: two parents not currently married: single parent: and other. Poverty/income was measured using the following four levels: 0-99% of the federal poverty level (FPL), 100%-199% FPL, 200%-399 FPL, and 400% FPL or above. Health insurance was sorted into the following: public, private, public and private, and not insured/unspecified. Analytic approach Survey sampling weights, cluster, and strata used by the NSCH were used in all analyses. Z-score tests were used to test the differences between proportions across years. A p-value <0.05 was deemed statistically significant. SAS statistical software was used for all analyses (SAS, version 9.3; SAS Institute, Cary, NC). This study was approved as exempt from the [university concealed for review] Institutional Review Board. Results Children represented in the 2018-2019 and 2020-2021 NSCH samples were demographically similar, with only a few statistically significant differences: primary language spoken in the home and poverty/income level (p<0.05). In both samples, the majority of children were male and six to twelve years of age. Over a quarter of the children in both samples were of Hispanic ethnicity, and just less than a quarter of children had special healthcare needs. Most caregivers had some college education or more, and were two parents, currently married. There was a lower percentage of children living below the poverty level in 2020-2021 (17.1%), compared to 18.5% in the 2018-2019 NSCH (p<0.05). The percentage of children living in a household where the primary language was not English increased from 13.7% (2018-2019) to 14.8% (2020-2021). Positive childhood experiences declined in four of the seven PCEs measured, from 2018-2019 to 2020-2021: after school activities, community volunteerism, guiding mentor, and resilient family, with all differences significant by p<0.0001. After school activities decreased from 79.8% to 72.2%, community volunteering decreased from 43.9% to 35.1%, guiding mentor decreased from 88.8% to 86.3%, and resilient family decreased from 92.7% to 84.6%. PCEs increased for safe neighborhood (64.7% to 67.2%), supportive neighborhood (55.8% to 57.5%), and connected caregiver (65.3% to 94.7%). Discussion This is the first study, to our knowledge, to estimate the national prevalence of PCEs during the COVID-19 pandemic among school age children. PCEs are important to examine, as they are critical for healthy social emotional development in children in both the short and long-term.6, 8 Yet, with the onset of the COVID-19 pandemic, and many churches, schools, and community organizations going virtual, many of these “outside” opportunities to experience PCEs declined. We found that children experiencing after school activities, community volunteerism, guiding mentor, and resilient family all declined during the start of the COVID-19 pandemic, while still remaining remarkably high overall, likely due to state variations in length of lockdowns and variations in schools reopening. Further research is needed to examine how these experiences varied by race/ethnicity, age, and income- level, as COVID-19 impacted different racial/ethnic groups in our country very differently.12 The PCEs that increased included higher proportions of living in a safe neighborhood, supportive neighborhood, or connected caregiver. This may reflect a “turning inward” to rely or connect with those closest to us during the pandemic. In the beginning, children were home in their neighborhoods or families, with many neighborhoods displaying teddy bears or rainbows in windows, having drive-by birthday parties for children, and other neighborhood cohesion activities.13 Residents’ perceptions of neighborhood cohesion have been found to buffer the impact of COVID-19 on the mental health of individuals, with residents who felt cohesiveness less likely to report symptoms of depression.13 Limitations of this study include the self-reporting of PCE data by the caregivers of the children, who may over-report PCEs as they are socially desirable experiences for children to have. Furthermore, the PCEs measured are not from a validated instrument, but have been quantified numerous times in prior literature.10 NSCH does not survey homeless or transient families, using an address-based sampling method. Thus, results from the NSCH will not represent those families. This study has many strengths, including being the first to examine PCEs before and during the COVID-19 pandemic. Table 1 Characteristics of Respondents to the 2018-2019 National Survey of Children’s Health, n=42,464, and to the 2020-2021 National Survey of Children’s Health, n=54,256. Table 1Characteristic 2018-2019 NSCH 2020-2021 NSCH (%) (%) Characteristics of Child Sex of Child Male 51.0 51.0 Female 49.0 49.0 Age of Child 6-12 years old 58.2 58.1 13-17 years old 41.8 41.9 Race/Ethnicity of Child Non-Hispanic White 50.9 51.6 Non-Hispanic Black 13.4 12.6 Hispanic 25.3 25.3 Non-Hispanic Other 10.5 10.0 Special Health Care Needs Yes 23.2 24.2 Characteristics of Caregiver/Household Primary Language⁎ Not English 13.7 14.8 Guardian Education Less than high school or high school 29.0 28.7 Some college or more 71.0 71.3 Family Structure Two parents, currently married 63.7 64.8 Two parents, not currently married 7.8 6.1 Single parent 22.3 24.2 Other 6.2 4.8 Poverty/ Income Level⁎ 0-99% Federal Poverty Level 18.5 17.1 100%-199% Federal Poverty Level 21.3 21.1 200%-399% Federal Poverty Level 28.4 29.9 400% Federal Poverty Level or above 31.9 31.8 Health Insurance Public 28.3 28.1 Private 59.1 58.9 Public and Private 4.2 4.3 Not Insured/Unspecified 8.4 8.7 ⁎ Indicates significantly different, p<0.05, between the 2018-2019 and 2020 NSCH Table 2 Positive Childhood Experiences Reported by Respondents to the 2018-2019 National Survey of Children’s Health, n=42,464, and to the 2020-2021 National Survey of Children’s Health, n=54,256. Table 2Characteristic Pre-COVID2018-2019(%) During COVID2020-2021(%) P value Positive Childhood Experiences After school activities 79.8 72.2 <0.0001 Community volunteer 43.9 35.1 <0.0001 Guiding mentor 88.8 86.3 <0.0001 Connected caregiver 65.3 94.7 <0.0001 Safe neighborhood 64.7 67.2 <0.0001 Supportive neighborhood 55.8 57.5 <0.0001 Resilient family 92.7 84.6 <0.0001 Conclusions Following the reopening of community spaces, we must recognize that children have experienced higher levels of parental stress and disruption and that policymakers and program makers must find ways to increase exposure to PCEs.14 The quantification of these PCEs is a great start, with further research needed to describe how these experiences varied by race/ethnicity, age, and income- level, as well as on school, community, and church support. Declaration of Competing Interest In the interest of transparency, we ask you to disclose all relationships/activities/interests listed below that are related to the content of your manuscript. “Related” means any relation with for-profit or not-for-profit third parties whose interests may be affected by the content of the manuscript. Disclosure represents a commitment to transparency and does not necessarily indicate a bias. If you are in doubt about whether to list a relationship/activity/interest, it is preferable that you do so. Acknowledgements The authors declare that they have no conflict of interest. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. ==== Refs References 1 Pan A. Liu L. Wang C. Association of public health interventions with the epidemiology of the COVID-19 outbreak in Wuhan, China JAMA 323 19 2020 1915 1923 10.1001/jama.2020.6130 32275295 2 Parks S.E. Zviedrite N. Budzyn S.E. COVID-19–related school closures and learning modality changes — United States, August 1–September 17, 2021 MMWR Morb Mortal Wkly Rep 70 39 2021 1374 1376 10.15585/mmwr.mm7039e2 34591828 3 Zviedrite N. Hodis J.D. Jahan F. COVID-19–associated school closures and related efforts to sustain education and subsidized meal programs, United States, February 18–June 30, 2020 PloS one 16 9 2021 e0248925 10.1371/journal.pone.0248925 4 Chandra S., Chang A., Day L., et al. Closing the K–12 digital divide in the age of distance learning. Common Sense and Boston Consulting Group; 2020. Accessed May 4, 2023. 〈https://www.commonsensemedia.org/kids-action/publications/closing-the-k-12-digital-divide-in-the-age-of-distance-learning〉 5 Cluver L. Lachman J.M. Sherr L. Parenting in a time of COVID-19 Lancet. 395 10231 2020 e64 10.1016/S0140-6736(20)30736-4 6 Sege R.D. Browne C.H. Responding to ACEs with HOPE: Health outcomes from positive experiences Acad Pediatr 17 7 2017 S79 S85 10.1016/j.acap.2017.03.007 28865664 7 Centers for Disease Control and Prevention Accessed May 4 Help youth at risk for ACEs 2023 Accessed May 4 〈https://www.cdc.gov/violenceprevention/aces/help-youth-at-risk.html〉 Accessed May 4 8 Banyard V. Hamby S. Grych J. Health effects of adverse childhood events: Identifying promising protective factors at the intersection of mental and physical well-being Child Abuse Negl 65 2017 88 98 10.1016/j.chiabu.2017.01.011 28131000 9 Crouch E. Radcliff E. Merrell M.A. Bennett K.J. Rural‐urban differences in positive childhood experiences across a national sample J Rural Health 37 3 2021 495 503 10.1111/jrh.12493 32639648 10 Crouch E. Radcliff E. Merrell M.A. Racial/ethnic differences in positive childhood experiences across a national sample Child Abuse Negl 115 2021 105012 10.1016/j.chiabu.2021.105012 11 Crouch E. Radcliff E. Merrell M.A. Positive childhood experiences promote school success Matern Child Health J 25 10 2021 1646 1654 10.1007/s10995-021-03206-3 34390426 12 Gillespie D.L. Meyers L.A. Lachmann M. The experience of 2 independent schools with in‐person learning during the COVID‐19 pandemic J Sch Health 91 5 2021 347 355 10.1111/josh.13008 33768529 13 Robinette J.W. Bostean G. Glynn L.M. Perceived neighborhood cohesion buffers COVID-19 impacts on mental health in a United States sample Soc Sci Med 285 2021 114269 10.1016/j.socscimed.2021.114269 14 Lucassen N. de Haan A.D. Helmerhorst K.O. Keizer R. Interrelated changes in parental stress, parenting, and coparenting across the onset of the COVID-19 pandemic J Fam Psychol 35 8 2021 1065 1076 10.1037/fam0000908 34398624
PMC010xxxxxx/PMC10286522.txt	==== Front Thrombosis Update 2666-5727 Published by Elsevier Ltd. S2666-5727(23)00013-5 10.1016/j.tru.2023.100142 100142 Article Pulmonary embolism in hospitalized COVID-19 patients: Short- and long-term clinical outcomes Luu Inge H.Y. a Buijs Jacqueline a Krdzalic Jasenko b de Kruif Martijn D. c Mostard Guy J.M. a ten Cate Hugo d Dormans Tom P.J. e Mostard Remy L.M. c Leers Math P.G. f van Twist Daan J.L. a∗ a Department of Internal Medicine, Zuyderland Medical Centre, PO-box 5500, 6130, MB, Sittard, the Netherlands b Department of Radiology, Zuyderland Medical Centre, PO-box 5500, 6130, MB, Sittard, the Netherlands c Department of Pulmonology, Zuyderland Medical Centre, PO-box 5500, 6130, MB, Sittard, the Netherlands d Department of Internal Medicine and Biochemistry, Cardiovascular Research Institute Maastricht (CARIM), Maastricht University Medical Centre, PO-box 616, 6200, MD, Maastricht, the Netherlands e Department of Intensive Care, Zuyderland Medical Centre, PO-box 5500, 6130, MB, Sittard, the Netherlands f Department of Clinical Chemistry and Haematology, Zuyderland Medical Centre, PO-box 5500, 6130, MB, Sittard, the Netherlands ∗ Corresponding author. 22 6 2023 22 6 2023 10014220 4 2023 15 6 2023 17 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction Pulmonary embolism (PE) is a frequent complication in COVID-19. However, the influence of PE on the prognosis of COVID-19 remains unclear as previous studies were affected by misclassification bias. Therefore, we evaluated a cohort of COVID-19 patients whom all underwent systematic screening for PE (thereby avoiding misclassification) and compared clinical outcomes between patients with and without PE. Materials and methods We included all COVID-19 patients who were admitted through the ED between April 2020 and February 2021. All patients underwent systematic work-up for PE in the ED using the YEARS-algorithm. The primary outcome was a composite of in-hospital mortality and ICU admission. We also evaluated long-term outcomes including PE occurrence within 90 days after discharge and one-year all-cause mortality. Results 637 ED patients were included in the analysis. PE was diagnosed in 46 of them (7.2%). The occurrence of the primary outcome did not differ between patients with PE and those without (28.3% vs. 26.9%, p = 0.68). The overall rate of PE diagnosed in-hospital (after an initial negative PE screening in the ED) and in the first 90 days after discharge was 3.9% and 1.2% respectively. One-year all-cause mortality was similar between patients with and without PE (26.1% vs. 24.4%, p = 0.83). Conclusions In a cohort of COVID-19 patients who underwent systematic PE screening in the ED, we found no differences in mortality rate and ICU admissions between patients with and without PE. This may indicate that proactive PE screening, and thus timely diagnosis and treatment of PE, may limit further clinical deterioration and associated mortality in COVID-19 patients. Keywords COVID-19 Pulmonary embolism Clinical outcomes Handling Editor: Dr P Emmanouil ==== Body pmcAbbreviations BMI body mass index bpm beats per minute CCI Charlson Comorbidity Index CI confidence interval COPD chronic obstructive pulmonary disease CORADS COVID-19 CT classification score CRP C-reactive protein CT computed tomography CTPA computed tomography-pulmonary angiography DOAC direct oral anticoagulant ED emergency department ICU intensive care unit OR Odds ratio PE pulmonary embolism RT-PCR reverse transcriptase-polymerase chain reaction 1 Introduction Pulmonary embolism (PE) is a frequent complication in patients with COVID-19 [[1], [2], [3]]. Several observational studies and systematic reviews addressed the outcomes of critically ill COVID-19 patients with PE [[4], [5], [6]]. Although most studies observed worse outcomes in patients with PE, there is still considerable uncertainty as the reported results are not conclusive. Moreover, available studies have been limited due to methodological constraints. First, it is likely that PE diagnoses have been missed since the vast majority of studies only performed computed tomography-pulmonary angiography (CTPA) in case of clinical suspicion for PE. As ruling out PE solely on clinical grounds is difficult (if not impossible) [7], these studies were susceptible to misclassification bias. Second, by including patients in whom PE was diagnosed throughout hospitalization, substantial confounders that influence the risk for PE (e.g., the use of thromboprophylaxis and immunomodulators) may have also affected the prognosis of patients in those studies. In the present study, we aimed to evaluate the influence of concomitant PE on the prognosis of COVID-19 patients, while tending to avoid aforementioned misclassification. Therefore, we studied a cohort of hospitalized COVID-19 patients whom all underwent systematic PE screening in the emergency department (ED) upon admission. We compared patients with and without PE at the moment of ED presentation with regard to clinical outcomes including in-hospital mortality and the need of intensive care unit (ICU) care. Additionally, we evaluated the occurrence of PE at 90 days after discharge, and one-year all-cause mortality. 2 Methods In this retrospective study, we included all ED patients aged 18 years and over who were admitted to Zuyderland Medical Centre, the Netherlands, between April 7, 2020 and February 28, 2021, with a definite diagnosis of COVID-19. A definite diagnosis of COVID-19 was defined as having clinical symptoms suspected for COVID-19 (according to the WHO case definition for suspected COVID-19) [8] and either a positive reverse transcriptase-polymerase chain reaction (RT-PCR) for SARS-CoV-2 upon ED presentation or a pending test result for SARS-CoV-2 upon ED presentation that was subsequently confirmed by a positive RT-PCR during hospitalization, and/or a COVID-19 CT classification score (CO-RADS) [9] 4 or 5 on chest computed tomography (CT) obtained at the ED. All patients were systematically screened for PE during ED assessment using the YEARS algorithm [10], which consists of three clinical items: clinical signs of deep vein thrombosis, hemoptysis, and PE as the most likely diagnosis. In case of ≥1 YEARS-items, a D-dimer cut-off level of ≥500 μg/L was applied. If none of the three YEARS-items were present, the cut-off was increased to 1000 μg/L. Patients who had D-dimer values below the cut-off were considered to have PE excluded and underwent a non-contrast-enhanced chest CT or a chest X-ray as part of the standard clinical work-up for COVID-19. In patients with D-dimer values above the cut-off value, CTPA was performed. The YEARS algorithm was previously proved to be efficient in ruling out PE in patients with COVID-19 [11]. D-dimer levels were measured using CS2500 automated blood coagulation analysers (Sysmex Corporation, Kobe, Japan; upper limit of measuring range >35000 μg/L). CTPA examinations were performed using 64-detector-row CT-scanners (Incisive, Philips Medical Systems, The Netherlands, and Definition Flash, Siemens Healthineers, Germany) and reviewed by board-certified radiologists. We divided our study population into a group of patients who were diagnosed with PE and a group in whom PE was excluded at the time of admission. The main goal of this study was to evaluate differences in clinical outcomes between COVID-19 patients with PE and those without PE. The primary outcome was a composite endpoint including need of ICU admission or in-hospital mortality of any cause. Secondary outcomes were 1) ICU admission, 2) in-hospital, 30-days and 1-year mortality, and 3) length of hospital stay. All patients in whom PE was ruled out at the ED were administered prophylactic-dose anticoagulation (subcutaneous dalteparin 5000 IU once daily) during hospitalization. Patients who were diagnosed with PE received therapeutic-dose anticoagulation, generally subcutaneous dalteparin 200 IU/kg body weight daily during hospitalization, which was converted to a Direct Oral Anticoagulant (DOAC) after discharge, for at least 3 months. Patients were excluded from the present analysis if they had a contraindication to CTPA (e.g., allergy to intravenous contrast agents, impaired renal function, or inability to cooperate) or to anticoagulant treatment (e.g., active major bleeding). Patients with acute respiratory failure who were not able to undergo CTPA during ED stay (because of immediate high-flow oxygen therapy or mechanical ventilation) were excluded as well. Patients in whom the diagnostic protocol was violated (e.g., no D-dimer testing during ED presentation or no CTPA despite indicated according to the protocol) were also excluded from the present analysis. During the pandemic, a nationwide system was developed to expedite interhospital patient transfers to facilitate an equitable distribution of COVID-19 patients across the country's hospitals. As a result, patients who were transferred to other hospitals were lost to follow up and therefore excluded from the analysis. Data collection included information on demographics, clinical manifestations, laboratory data, and date of first ICU admission. Detailed information on the use of antiplatelet or anticoagulant drugs was also collected. The Charlson Comorbidity Index (CCI) was calculated for each patient [12]. In patients with PE, the location of the thrombi was recorded. To assess long-term mortality, information on post-discharge deaths was obtained by referencing the mortality register, ensuring reliable data. For the occurrence of PE after discharge, we manually reviewed all CTPAs up to one year following the patients' discharge. During this review process, the indication for CTPA and the presence of PE were carefully documented. All data were registered in the local ELVIS registry. All patients received written information about the registry and an opt-out form in case they did not want to participate. The study was approved by the local ethical committee (METCZ20200076). Differences between groups were examined using the Chi-square test or Fisher's exact test for categorical variables and the Student's t-test or Mann-Whitney U-test for continuous data. Missing data were removed from the denominator when calculating relative frequencies. The Kaplan-Meier survival analysis was used to estimate the event-free survival for the primary and secondary endpoints. For the composite endpoint, the adverse event that occurred first was accounted for. The Log-Rank test was used to compare these survival curves. Univariate and multivariate regression analyses were performed to identify factors associated with the primary endpoint. Survival analysis was performed using GraphPad Prism (version 9.0 for Mac, GraphPad Software Inc., San Diego, CA, USA). All other analyses were performed using SPSS software (version 28.0; IBM Corp., Armonk, NY, USA). 3 Results During the study period, 778 patients with COVID-19 were admitted to the hospital from the ED. After applying the inclusion and exclusion criteria, a total of 637 ED patients with COVID-19 were included in the analysis (Fig. 1 ). 297 patients (46.6%) had a D-dimer below cut-off according to the YEARS algorithm and did not undergo CTPA in the ED. In 340 patients (53.4%) CTPA was performed as indicated by the YEARS algorithm. PE was diagnosed in 46 of them, resulting in an overall PE prevalence of 7.2% (46 out of 637 patients) upon ED admission. The location of PE is described in detail in Appendix A.Fig. 1 Flowchart of the study population. 1D-dimer cut-off ≥500 μg/L if ≥ 1 YEARS items; D-dimer cut-off is increased to 1000 μg/L if 0 YEARS items. ED, emergency department; CTPA; computed tomography pulmonary angiogram; PE, pulmonary embolism. Fig. 1 The characteristics of the included patients are summarized in Table 1 . Patients were predominantly men (64.4%), with a mean age of 68.9 ± 13.7 years and a mean body mass index (BMI) of 28.4 ± 5.8 kg/m2. Median time from onset of COVID-19-related symptoms to ED presentation was 7 [3–10] days, and the majority of patients (61.9%) fell into the 'severe' or 'critical' illness categories according to the National Institutes of Health (NIH) Clinical Spectrum of SARS-CoV-2 [13]. Infection Patients with PE upon ED admission had statistically significant higher levels of CRP, platelet count, and D-dimer (Table 1).Table 1 Patient characteristics. Table 1N PE No-PE p value 46 591 Demographics Age (year) 70.5 ± 11.4 68.8 ± 13.8 0.43 Male sex, n (%) 29 (63.0) 381 (64.5) 0.85 BMI (kg/m2) 27.8 ± 5.6 28.5 ± 5.9 0.51 > 30 kg/m2, n (%) 10 (23.3) 178 (31.9) 0.24 Use of anticoagulant drugs, n (%) 3 (6.5) 98 (16.6) 0.07 Comorbidities Diabetes mellitus, n (%) 17 (37.0) 154 (26.1) 0.11 COPD, n (%) 6 (13.0) 97 (16.4) 0.55 Active malignancy, n (%) 4 (8.7) 28 (4.7) 0.48 Charlson Comorbidity Index 4 [2–5] 3 [2–5] 0.11 Time symptom onset to presentation ED (days) 7 [4–11] 7 [3–10] 0.86 Vital signs Temperature (°C) 37.8 ± 1.1 37.9 ± 1.1 0.26 Heart rate (bpm) 97 ± 24 92 ± 19 0.17 Mean arterial pressure 97 ± 14 98 ± 14 0.69 COVID-19 disease severitya Moderate, n (%) 15 (32.6) 228 (38.6) 0.42 Severe, n (%) 13 (28.3) 277 (46.9) 0.02 Critical, n (%) 18 (39.1) 86 (14.6) <0.001 Laboratory findings Hemoglobin (mmol/L) 8.7 [7.6–9.4] 8.4 [7.6–9.1] 0.46 Leukocytes (×10E9/L) 8.4 [6.6–11.8] 7.2 [5.2–9.4] 0.13 Platelet count, (×10E9/L) 225 [178–341] 208 [165–264] 0.03 CRP (mg/L) 116 [49–185] 73 [34–125] 0.02 Ferritin (μg/L) 914 [384–1519] 685 [313–1339] 0.20 D-dimer (μg/L) 4510 [1937–12598] 918 [544–1923] <0.001 Data shown as mean ± standard deviation or median [interquartile range], unless otherwise stated. BMI, body mass index; bpm, beats per minute; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; ED, emergency department; PE, pulmonary embolism. a COVID-19 disease severity according to the NIH Clinical Spectrum of SARS-CoV-2. 3.1 Primary outcome Fig. 2 shows the Kaplan-Meier time to event curve for the primary outcome (composite of in-hospital mortality or ICU admission). The mean time to event for the PE group was significantly shorter (1.8 ± 2.8 days for PE vs. 4.7 ± 5.2 days for the non-PE group; p = 0.04). However, the occurrence of the primary outcome throughout the whole hospitalization period did not differ between patients with PE and those without PE (28.3% vs. 26.9%; HR (log-rank test) 1.13, 95%CI 0.62–2.04; p = 0.68). In an additional analysis in which patients who were on chronic anticoagulant drugs were excluded, comparable results were found (PE vs non-PE, 27.9% vs. 27.8%; HR (log-rank test) 1.07, 95%CI 0.58–1.95; p = 0.83).Fig. 2 Kaplan-Meier time-to-event curve for the composite endpoint of ICU admission or in-hospital death (from any cause) in patients with PE (red line) and without PE (blue line). Fig. 2 In a univariate analysis of the entire study population, significant predictors for the primary endpoint included an older age (OR 1.19 per 5 years, 95% CI 1.11–1.28, p < 0.001), CCI score ≥4 (OR 2.47, 95% CI 1.66–3.66, p < 0.001), C-reactive protein (CRP) > 100 mg/L (OR 1.94, 95% CI 1.36–2.77, p < 0.001), and lactate >1.8 mmol/L (OR 1.89, 95% CI 1.30–2.77, p < 0.001). The presence of PE, sex, and BMI were not associated. In multivariate analysis, age, CRP and lactate were still found to be independently associated with in-hospital death or ICU admission (Table 2 ).Table 2 Regression analysis for factors associated with the primary endpoint in ED patients with COVID-19. Table 2Variables Univariate analysis Multivariate analysis OR 95% CI P-value OR 95% CI P-value Age per 5 years 1.19 1.11–1.28 <0.001 1.14 1.02–1.28 0.02 Male sex 1.45 0.99–2.11 0.06 BMI (kg/m2) <25 Ref 25-30 1.04 0.67–1.61 0.86 ≥30 0.83 0.51–1.33 0.43 Chronic comorbidity COPD 1.20 0.76–1.91 0.44 Malignant neoplasms 1.06 0.48–2.34 0.88 CCI ≥4 2.47 1.66–3.66 <0.001 1.42 0.81–2.51 0.22 Vital signs Heart rate >100 bpm 1.00 0.68–1.46 0.98 Temperature ≥38.5 °C 0.97 0.67–1.42 0.89 Laboratory results CRP >100 mg/L 1.94 1.36–2.77 <0.001 1.62 1.07–2.46 0.02 Lactate >1.8 mmol/L 1.89 1.30–2.77 <0.001 1.57 1.02–2.40 0.04 Pulmonary Embolism 1.07 0.55–2.09 0.84 BMI, body mass index; bpm, beats per minute; CCI, Charlson comorbidity index; CI, confidence interval; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; ED, emergency department; OR, Odds ratio. 3.2 Secondary outcomes 3.2.1 Short-term outcomes Short-term outcomes are shown in Table 3 . Patients with PE were more likely to be admitted to ICU directly from ED in comparison to patients without PE (15.2% vs. 5.9%, p = 0.03). However, despite a trend towards an increased need of ICU care throughout the whole period of hospitalization in patients with PE (19.6%) compared to patients without PE (12.4%), the difference was not statistically significant (p = 0.16). Among those who were admitted to ICU, the same pattern was observed in terms of need of invasive mechanical ventilation (PE vs. non-PE, 10.9% vs. 6.6%, p = 0.24) and vasopressor use (10.9% vs. 5.8%, p = 0.19), but these differences were not statistically significant either. The median length of hospital stay was 6 days, with no significant difference between patients with and without PE (p = 0.44, Supplemental Fig. 1). In-hospital mortality was comparable between patients with PE and those without (19.6% vs. 19.1%, p = 0.94) with a mean time to event of 10.4 ± 10.1 and 8.7 ± 7.6 days respectively (p = 0.52).Table 3 Short-term outcomes. Table 3 PE (n = 46) No PE (n = 591) P-value Admission to ICU, n (%) 9 (19.6) 73 (12.4) 0.16 Directly from ED, n (%) 7 (15.2) 35 (5.9) 0.03 All-cause mortality In-hospital, n (%) 9 (19.6) 113 (19.1) 0.94 30-days, n (%) 9 (19.6) 120 (20.3) 0.90 One-year, n (%) 12 (26.1) 144 (24.4) 0.83 Length of hospital stay (days), median [IQR] 7 [4–13] 6 [4–11] 0.46 ED, emergency department; ICU, intensive care unit. 3.2.2 PE during hospitalization Among the 591 patients in whom PE was considered excluded at the moment of ED presentation, 131 patients (22.2%) underwent CTPA during hospitalization because of clinical suspicion of new PE. PE was found in 23 patients resulting in an overall in-hospital-diagnosed PE prevalence of 3.9%, with a median time of diagnosis of 6 [4–9] days after admission. Of these 23 patients, 15 previously underwent CTPA in the ED as indicated by the YEARS algorithm, which was negative in all of them. The other 8 patients (34.8%) had D-dimer values below the cut-off of the YEARS algorithm and were considered to have PE excluded without performing CTPA during ED evaluation. In an additional analysis comparing all patients with PE (diagnosed in the ED and/or during hospitalization) to those without PE, there were still no differences found in the composite outcome of in-hospital mortality or ICU admission (p = 0.66). 3.2.3 Long-term outcomes 30-day all-cause mortality was comparable between patients with PE (19.6%) and patients without PE (20.3%, p = 0.90; Fig. 3 A). Similarly, 1-year all-cause mortality rate was not significantly different between both groups (26.1% vs. 24.4%, p = 0.83; Fig. 3B). In the first 90 days after discharge, CTPA was performed in 41 of 515 patients (8.0%) who left the hospital alive, in 33 because of suspected PE and in 8 of the cases for a non-PE indication (e.g., surveillance CT for malignancies, or follow-up CT after severe COVID-19). PE was found in 6 patients, resulting in an overall PE occurrence of 1.2% in the first 90 days after discharge. In all of them, PE was newly developed, as these patients already underwent CTPA at ED presentation or during hospitalization, which was negative at that moment.Fig. 3 Kaplan-Meier time-to-event curve for (A) 30-days survival, and (B) one-year survival, in patients with PE (red line) and without PE (blue line). Fig. 3 4 Discussion In the present study, we compared the clinical outcomes between COVID-19 patients with PE and those without PE at the time of ED admission. All included patients underwent systematic work-up for PE in the ED using the YEARS algorithm to avoid misclassification. PE was diagnosed in 46 out of 637 patients (7.2%). Patients with PE had a significant shorter time to the primary endpoint (i.e., ICU admission or in-hospital death) than those without PE, but the total incidence of the primary endpoint throughout the entire hospitalization period was comparable between both groups. At one year follow-up, all-cause mortality was not significantly different between patients with and without PE. We observed that patients with PE were more likely to be admitted to the ICU directly from the ED compared to those without PE. However, the overall rates of ICU admissions and in-hospital mortality for the total duration of hospitalization did not differ between groups. We hypothesize that PE causes more respiratory distress in the early stage of hospitalization, resulting in ICU admission relatively early during hospitalization. In hospitalized COVID-19 patients without PE, the course of disease is presumably insidiously progressive, resulting in ICU admission at a later stage. To the best of our knowledge, other studies investigating the timing of PE in relation to ICU admission or death in patients with COVID-19 are lacking. The lack of a significant association between PE and in-hospital mortality and ICU admission in the present study is in contrast with several previous studies. A German retrospective study in hospitalized COVID-19 patients found a case-fatality rate of 28.7% in patients with PE compared to 17.7% in patients without PE [14]. A retrospective case-control study among ED patients with COVID-19 found a higher number of ICU admissions in COVID-19 patients with PE (OR 3.19, 95%CI 1.95–5.21) [15]. We hypothesize that the discrepancy between our findings and the results reported in those studies is caused by the lack of systematic PE screening in the latter. First, those studies may be biased by misclassification as CTPA was only performed in case of a clinical suspicion. This may have resulted in a substantial number of missed PEs [16]. In our cohort, systematic PE screening was performed, which substantially reduces the risk of misclassification. Second, it is likely that proactive and systematic screening for PE in our study contributed to early diagnosis and treatment, thereby limiting the progression of PE and the associated clinical deterioration and mortality. With this in mind, we advocate for evaluation of PE in an early stage of COVID-19 disease, allowing prompt diagnosis and timely initiation of treatment. Consistently to the short-term outcomes, no differences were found in mortality rate at one year follow-up between COVID-19 patients with and without PE. In fact, in both groups, the majority of casualties occurred in the first 30 days. Only few studies reported long-term outcomes with follow-up periods from 90 to 98 days in patients with COVID-19-related PE [17,18], showing a very low mortality rate after the first 30-days, consistent with our data. We found no other studies that evaluated mortality over a one-year period after a PE diagnosis in ED patients with COVID-19. The comparable one-year mortality rate in patients with and without PE obtained in our study probably indicates that PE does not have long standing effects that predispose patients to worse survival, even after the acute phase of the COVID-19 disease. In the majority of patients in whom PE was diagnosed during hospitalization, PE was initially excluded by CTPA upon admission. Hence, ruling out PE at the time of ED presentation and using prophylactic anticoagulation does not exclude the possibility of developing new thrombi during hospital stay. Screening for in-hospital PE was not systematically performed but only if considered indicated by the treating physician, most often because of respiratory deterioration and/or a progressive increase in D-dimer levels. With this approach, the overall rate of in-hospital-diagnosed PE (after an initial negative PE screening in the ED) was low (3.9%). However, among those patients who were referred for additional CTPA during hospitalization, the number of positive CTPAs for PE was substantial (23 out of 131 patients). Thus, although the risk for development of PE after initial negative ED-screening is low, assessment of hospital-acquired PE is still recommended in case of unexplained clinical deterioration during hospitalization. The current study is, to the best of our knowledge, the first to report short- and long-term PE related outcomes in patients with COVID-19 whom underwent systematic PE screening in the ED upon admission. In contrast to previous studies, this approach limits the risk for misclassification bias. Still, this study has a few limitations. First, since we did not study a control group in whom PE was evaluated based on clinical gestalt, we cannot draw conclusions with regard to the impact of systematic PE screening on the overall prognosis. Second, we cannot rule out that indication bias has occurred as raising awareness of COVID-19 associated PE throughout the pandemic may have lowered the threshold for ICU admission in patients with PE, and thus may have affected the results. Third, although PE was systematically screened for in all patients in the ED, the decision to perform CTPA during hospitalization was based on physician's clinical gestalt (in case of suspected PE). As such, we cannot exclude in-hospital-acquired PE with certainty in those who did not underwent CTPA. Last, our results relate to a cohort of hospitalized COVID-19 patients in the first year of the COVID-19 pandemic. Consequently, the findings may be less applicable to other populations and settings, such as non-hospitalized patients, and the impact of newer COVID-19 variants are unknown. 5 Conclusion In a cohort of hospitalized COVID-19 patients whom all underwent systematic PE screening in upon admission, the risk for ICU admission and mortality (in-hospital and post-discharge) did not differ between patients with and without PE. These findings are in sharp contrast to the worse outcomes of COVID-19 patients with PE reported in observational studies without systematic PE screening. This could be related to misclassification bias in those studies, but may also suggest that systematic and proactive PE screening and thus early PE treatment limits further clinical deterioration and mortality. Therefore, we advocate evaluating PE systematically in COVID-19 patients at an early stage of hospitalization, allowing prompt diagnosis and timely initiation of treatment of PE. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Supplementary data The following are the Supplementary data to this article.Multimedia component 1 Multimedia component 1 Multimedia component 2 Multimedia component 2 Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.tru.2023.100142. ==== Refs References 1 Avila J. Long B. Holladay D. Thrombotic complications of COVID-19 Am. J. Emerg. Med. 39 2021 213 218 33036855 2 Malas M.B. Naazie I.N. Elsayed N. Thromboembolism risk of COVID-19 is high and associated with a higher risk of mortality: a systematic review and meta-analysis EClinicalMedicine 29–30 2020 100639 3 Kollias A. Kyriakoulis K.G. Lagou S. Venous thromboembolism in COVID-19: a systematic review and meta-analysis Vasc. Med. 26 2021 415 425 33818197 4 Gong X. Yuan B. Yuan Y. Incidence and prognostic value of pulmonary embolism in COVID-19: a systematic review and meta-analysis PLoS One 17 2022 e0263580 5 Mir T. Attique H.B. Sattar Y. Does pulmonary embolism in critically ill COVID-19 patients worsen the in-hospital mortality: a meta-analysis Cardiovasc. Revascularization Med. 31 2021 34 40 6 Nasrullah A. Gangu K. Shumway N.B. COVID-19 and pulmonary embolism outcomes among hospitalized patients in the United States: a propensity-matched analysis of national inpatient sample Vaccines 10 2022 2104 36560514 7 van Twist D.J.L. Luu I.H.Y. Kroon F.P.B. Pulmonary embolism in COVID-19: the actual prevalence remains unclear Radiology 299 2021 E254 33724068 8 World Health Organization Clinical management of severe acute respiratory infection when COVID-19 is suspected, world Health organisation, interim guidance V 1.2 https://www.who.int/publications-detail/clinical-management-of-severe-acute-respiratory-infection-when-novel-coronavirus-(ncov)-infection-is-suspected 2020 9 Prokop M. van Everdingen W. van Rees Vellinga T. CO-RADS: a categorical CT assessment scheme for patients suspected of having COVID-19-definition and evaluation Radiology 296 2020 E97 E104 32339082 10 van der Hulle T. Cheung W.Y. Kooij S. Simplified diagnostic management of suspected pulmonary embolism (the YEARS study): a prospective, multicentre, cohort study Lancet 390 2017 289 297 28549662 11 Stals M.A.M. Kaptein F.H.J. Bemelmans R.H.H. Ruling out pulmonary embolism in patients with (suspected) COVID-19—a prospective cohort study TH Open 5 2021 e387 e399 34541450 12 Charlson M.E. Pompei P. Ales K.L. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation J. Chron. Dis. 40 1987 373 383 3558716 13 COVID-19 Treatment Guidelines Panel Coronavirus disease 2019 (COVID-19) treatment guidelines https://www.covid19treatmentguidelines.nih.gov 14 Hobohm L. Sagoschen I. Barco S. COVID-19 infection and its impact on case-fatality in patients with pulmonary embolism Eur. Respir. J. 2022 2200619 15 Miró Ò. Jiménez S. Mebazaa A. Pulmonary embolism in patients with COVID-19: incidence, risk factors, clinical characteristics, and outcome Eur. Heart J. 42 2021 3127 3142 34164664 16 Luu I.H.Y. Frijns T. Buijs J. Systematic screening versus clinical gestalt in the diagnosis of pulmonary embolism in COVID-19 patients in the emergency department PLoS One 18 2023 e0283459 17 Buso G. Mazzolai L. Rueda-Camino J.A. Pulmonary embolism in patients with COVID-19: comparison between different care settings Semin. Thromb. Hemost. 2021 s-0041-1740152 18 Demelo‐Rodríguez P. Ordieres‐Ortega L. Ji Z. Long‐term follow‐up of patients with venous thromboembolism and COVID‐19: analysis of risk factors for death and major bleeding Eur. J. Haematol. 106 2021 716 723 33608914
PMC010xxxxxx/PMC10286525.txt	==== Front Glob Pediatr Glob Pediatr Global Pediatrics 2667-0097 The Author(s). Published by Elsevier Inc. S2667-0097(23)00032-5 10.1016/j.gpeds.2023.100066 100066 Article Understanding Time-to-Recovery among Guatemalan Children before and during COVID-19 Braxton Morgan E. PhD, RN, PED-BC 1⁎ Larson Kim L. PhD, RN, MPH, FNAP 2 Melendez Carlos R. PhD, MPH, MS 2 1 Arizona State University, 550 North 3rd Street, Phoenix, AZ 85004-0698 2 East Carolina University, 2205 W 5th St, Greenville, NC 27834 ⁎ Corresponding Author: 22 6 2023 22 6 2023 100066© 2023 The Author(s). Published by Elsevier Inc. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Purpose To understand malnutrition recovery at a Guatemalan Nutrition Rehabilitation Center (NRC) before and during the COVID-19 pandemic. Design and Methods A retrospective chart review was conducted on-site in November 2022. The NRC is located on the outskirts of Antigua, Guatemala. They manage the care of 15-20 children at a time, providing food, medicine, and health assessments. A total of 156 records were included (126 prior to the onset of COVID; 30 after the onset of COVID). Descriptive variables collected were age, gender, severity of malnutrition, height, weight, amoxicillin, multivitamins, nebulizer/bronchodilator, and zinc. Principal Results There was no significant difference in time-to-recovery between COVID cohorts. Mean time-to-recovery was 5.65 weeks, or 39.57 days (SD = 25.62, 95% CI [35.5, 43.7]) among all recovered cases (n =149). The cohort admitted after the onset of COVID-19 (March 1, 2020) had a significantly higher weight gain and discharge weight. In the total sample, amoxicillin was the only significant predictor variable for recovery time; with children receiving it being more likely to recover in >6 weeks. The few differences between cohorts was possibly attributed to the sample after the onset of COVID-19. These records had minimal sociocultural data. Major Conclusions Conducting a family needs assessment on admission could identify sociocultural factors that may facilitate nutritional recovery, such as housing conditions and potable water access. Further research is needed to more fully understand the complexities that the COVID-19 pandemic has had on childhood malnutrition recovery. Key words Malnutrition time-to-recovery Guatemala Nutrition Rehabilitation Center ==== Body pmc1 Introduction Nearly half of all deaths of children under age five are linked to malnutrition1. Severe Acute Malnutrition (SAM) puts children at greater risk of communicable diseases, delayed recovery and increased risk of complications2. Children in low- and middle-income countries (LMIC) are disproportionately affected by SAM3. Guatemala, a low-income country, has the highest rate of childhood malnutrition in Latin America, and fifth highest worldwide4. 1.1 Malnutrition in Guatemala In a recent cross-sectional study of nine Latin American and Caribbean countries, Guatemala had the highest prevalence of growth stunting and number of households in rural areas5. Socioeconomic and ethnic disparities of SAM within Guatemala are constant; children who identified as indigenous Maya, and living in poorer, rural areas are at greatest risk for SAM6 , 7. The home environment contributes largely to child health; many Guatemalan homes do not have reliable access to potable water and sanitation8 , 9. A controversial concept, maternal employment has had both positive and negative impacts on child health; there are potential financial benefits, but there is also uncertainty in the management of childcare and food preparation by someone other than the mother10 , 11 , 12. Investigators have found community food assistance programs13 , 14 and home gardens15, offered a solution to childhood malnutrition in Guatemala. However, there are concerns about sustainability of these types of interventions, since they require complementary strategies to achieve sustainability15. Rohloff16 highlighted the need to reduce persistent health disparities, by recognizing social context and prioritizing families’ needs. Investigators must consider the impact of the sociocultural context on child health outcomes to eliminate malnutrition in Guatemala. 1.2 Impact of COVID-19 The World Health Organization (WHO) announced a global pandemic of the novel SARS-CoV2 virus in the Spring of 202017. The pandemic disproportionately impacted LMICs, placing an estimated 1.2 million Guatemalans in need of emergency food-aid; many faced loss of jobs and food insecurity from the closure of public transportation systems and open food markets18 , 19. As the pandemic approaches its third year, young children from LMICs are at greatest risk of food insecurity and subsequent malnutrition2. Overall, the onset of the pandemic threatens long-term sequela for undernourished children, particularly those who are marginalized, and has the potential to undo previous efforts to improve global nutrition20. 1.3 Nutritional Standards and Management Nutritional standards for effective management of SAM are (a) a recovery rate of at least 75%, (b) a mortality rate less than 10%, (c) recovery within 6 weeks, and (d) children gaining ≥8 g/kg/day21 , 22. In managing SAM, supplementation of vitamins and minerals, including zinc is recommended to improve outcomes3 , 23. Oral amoxicillin is also recommended; yet, research on its use has had inconsistent findings, suggesting benefits21 , 24 and risks25 , 26. Malnutrition severity is determined by the on-site nutritionist using height-for-weight. It is classified using the WHO child growth standards, defined as the number of standard deviations below the growth curve, with -1 SD as mild, -2 SD as moderate and -3 SD as severe27. Children are considered ‘recovered’ when they are within the standard range for healthy weight based on weight-for-height27. Malnutrition treatment interventions include Outpatient Treatment Programs (OTPs) and Nutrition Rehabilitation Centers (NRCs). In OTPs, children are assessed in a community-based setting, provided food and medications, in particular Ready-to-Use Therapeutic Food (RUTF), and managed at home with regular evaluation28. In previous studies, time-to-recovery in OTPs ranged from 38.5-73 days; rates of weight gain were 4.2 -10.5 g/kg/day; and predictors of recovery included antibiotics29 , 30, vitamins, lack of comorbidities when enrolled in outpatient treatment31, and admission weight greater than 7 kg32. Community health workers (CHWs) often collaborate with OTPs to identify and refer at-risk children, as well as to independently manage acute cases within the community33. The OTPs provide a means of local access for early intervention; however, these programs have not consistently achieved nutritional standards for recovery28. In NRCs, malnourished children recover in a residential care setting that is monitored by healthcare professionals34. The majority of studies on NRCs have been set in India. Rates of weight gain in these studies ranged from 3.8 – 9.92 g/kg/day34 , 35 , 36. Despite the prevalence of NRCs in other countries, few studies have examined predictors of malnutrition recovery and none have been conducted in Guatemala. This is a unique setting in a country that could contribute to malnutrition recovery, making it an ideal setting to advance the science. Thus, the purpose of the study was to understand malnutrition recovery at a Guatemalan NRC before and during the COVID-19 pandemic. 1.4 Research Questions The research questions were: What impact has COVID-19 had on malnutrition recovery at an NRC in Guatemala? What relationships exist between individual clinical variables and time-to-recovery? 2 Material and Methods 2.1 Design A retrospective chart review was conducted in November 2021 to examine cases before and during the COVID-19 pandemic in order to evaluate the effects on malnutrition recovery. The research team consisted of a PhD nursing student (Principal Investigator) with primary proficiency in Spanish, biostatistician, and nursing professor with expertise in Latino population health; NRC leadership served as consultants. This study builds on a long-term community-university partnership and was approved by the university institutional review board (IRB # 21-001884). 2.2 Setting The NRC is located on the outskirts of Antigua, Guatemala. Children are referred to the NRC by hospitals, health clinics, social services, or self-referral. The NRC staff are composed of nursing assistants, a nutritionist, and pediatrician. They manage the care of 15-20 children at a time, providing food, medicine, and health assessments. Family-centered care is encouraged, where mothers reside at the center to help care for their child. 2.3 Sample The sample consisted of the medical records (cases) of eligible children treated at the NRC from January 1, 2019 to December 31, 2020. Inclusion criteria were: ≤ 5 years of age upon admission; a diagnosis of acute, moderate, or severe malnutrition; and a discharge from the NRC within the study period. Cases were excluded if parents refused to complete treatment. Cases were classified by date of admission as either pre-COVID-19 (prior to March 1, 2020) or post-COVID-19 (after March 1, 2020). Data was verified with the NRC leadership as needed. There were 205 cases available; 49 were excluded, leaving a total sample of 156 cases (see Figure 1 ).Figure 1 NRC cases included and excluded in the study Figure 1 2.4 Measures A structured audit tool, based on a pilot study37, was developed for data entry. No records were removed or photographed; all data were deidentified. The primary outcome variable was time-to-recovery. Primary predictor variables were pre- and post-COVID cohort, age, gender, severity of malnutrition, and admission weight. Secondary predictor variables were use of amoxicillin, multivitamins, nebulizer/bronchodilator, and zinc (see Figure 2 ). Secondary diagnoses were not available in the records.Figure 2 Conceptual and operational definitions Figure 2 2.5 Data Management and Analytic Strategy Data were collected on-site at the NRC by the PI and records were given an identification number using year (19 or 20) and case (01, 02, 03, etc.) in ascending order; for example, 1901, 1902, or 2001, 2002. Data were uploaded and analyzed using SPSS v. 28 and SAS v. 9.4. Variables were coded, outliers and missing data were identified, discussed by the research team and managed appropriately. Coded variables were validated by two members of the research team to ensure accuracy. Excluding sociocultural factors, there was less than 1% missing data. Descriptive statistics, Chi-Square tests, Student's t-tests, and multiple logistic regression were performed to answer the research questions (α = .05). Chi-square tests were done to analyze relationships between categorical variables. In some cases, Mantel-Haenszel Chi-square test was reported instead of the regular Chi-square when a categorical variable was not dichotomous. In cases where minimum expected counts were not met (≥5 in each cell), Fisher's or Mantel-Haenszel exact test was reported. 3 Theory This study was guided by the Social Ecological Model38, which informed the examination of the multilevel factors influencing childhood malnutrition. The model considers how health outcomes are shaped through the social context. Health outcomes are achieved from the interactions within and between the five levels (individual, interpersonal, organizational, community, and public policy), which continuously interact with one another38. Data collected in this study were analyzed within the framework of the model to collectively inform the researcher's holistic understanding of the influence of external factors on individual health outcomes. 4 Results The major finding of this study was that there was no significant difference in the primary outcome variable, mean time-to-recovery, between pre-COVID (n = 126) and post-COVID (n = 30) cohorts. The mean time-to-recovery for the pre-COVID cases was 40.46 days (SD = 26.65, 95% CI [35.8, 45.1]), or 5.78 weeks, and for the post-COVID cases was 35.90 days (SD = 20.80, 95% CI [28.5, 43.3]), or 5.13 weeks; (t (147) = .860, p = .391, two-tailed). Additionally, the post-COVID cohort had significantly greater discharge weight (p = .034) and weight gain (p = .010) (see Table 1 ). Both cohorts’ rates of weight gain were below the national standard (8 g/kg/day)22. There was no difference in height between cohorts (see Table 1).Table 1 Mean Growth among Recovered Cases in Pre- and Post- COVID Cohorts Table 1Variables Pre-COVID (n = 118) Post-COVID (n = 29) M SD M SD z p Admission Weight 6.1 2.0 6.6 2.2 1.65 .099 Height 66.2 9.5 69.1 11.2 1.61 .108 Discharge Weight 7.1 2.0 7.9 2.1 2.12 .034* Height 67.0 9.2 70.2 10.3 1.77 .077 Weight Gained 1.0 0.5 1.3 0.5 2.56 .011* Height Gained 0.9 1.2 1.2 1.4 1.27 .204 Grams/KG/Day 6.6 6.9 7.7 5.5 1.78 .076 Note. Weight reported in kilograms, height in centimeters. Reported growth on recovered cases; the six pre-COVID cases and one post-COVID case did not complete recovery were excluded. Missing data points led to the pre-COVID n = 118, and post-COVID n = 29. Mann-Whitney U/Wilcoxon Two-Sample test reported. Asterisks () indicate significant results. Six of the 17 categorical variables examining the relationship between COVID cohorts were significant: type of SAM (p < .001), admission weight <7 kg (p = .036), referral type (p = .025), multivitamins (p < .001), nebulizers/bronchodilators (p = .004), and mother's occupation (p = .013) (see Table 2 ). The majority of cases in both cohorts were admitted from the Southern region of Guatemala. Pre-COVID cases were primarily referred by the hospital, followed by self-referrals, while the majority of post-COVID cases were referred from health clinics, followed by the hospital. Of note, after the onset of COVID-19, the NRC required a 14-day isolation period upon admission, where mothers and children were isolated together and monitored for symptoms of COVID-19 (personal communication, November 8, 2021, NRC Director). Caution should be used in interpreting mother's occupation due to a majority of missing data (n = 92) (See Table 2). There was no statistically significant difference in the gender or age distribution of the two cohorts.Table 2 Frequencies and Chi-Square Results Between Pre-COVID (n=126) and Post-COVID (n= 30) Table 2 Pre-COVID Post-COVID Variable n % n % Total ꭓ2.‡ df p Age ≤24 months 115 82.7 24 17.3 139 – – .100 >24 months 11 64.7 6 35.3 17 Gender Male 69 83.1 14 16.9 83 .64 1 .425 Female 57 78.1 16 21.9 73 Severity of SAM Mild 5 100.0 0 0.0 5 3.52 1 .061 Moderate 59 85.5 10 14.5 69 Severe 62 75.6 20 24.4 82 Type of SAM Marasmus 22 66.7 11 33.3 33 11.83 1 <.001 Kwashiorkor 4 50.0 4 50.0 8 None 100 87.0 15 13.0 115 Admit Weight <7 kg 92 85.2 16 14.8 108 4.41 1 .036* ≥7 kg 34 70.8 14 29.2 48 Referral Type Walk-in 31 93.9 2 6.1 33 6.62 1 .010* Hospital 61 84.7 11 15.3 72 Health Clinic/Dept. 25 61.0 16 39.0 41 Social Services 9 90.0 1 10.0 10 Amoxicillin Yes 49 87.5 7 12.5 56 2.64 1 .104 No 76 76.8 23 23.2 99 Multivitamin Yes 28 56.0 22 44.0 50 28.72 1 <.001* No 97 92.4 8 7.6 105 Variable n % n % Total ꭓ2.‡ df p Nebulizer/ Bronchodilator Yes 70 89.7 8 10.3 78 8.33 1 .004* No 55 71.4 22 28.6 77 Zinc Yes 106 80.9 25 19.1 131 – – .785 No 19 79.2 5 20.8 24 Region Northern 3 75.0 1 25.0 4 1.01 1 .356 Central 50 86.2 8 13.8 58 Southern 73 77.7 21 22.3 94 Mother's Occupation Agriculture 1 50.0 1 50.0 2 6.30 1 .013* Homemaker 39 84.8 7 15.2 46 Sales 6 85.7 1 14.3 7 Service 0 0.0 7 100.0 7 No Contribution 1 50.0 1 50.0 2 Unknown 79 85.9 13 14.1 92 Father's Occupation Agriculture 18 78.3 5 21.7 23 .07 1 .807 Construction 4 66.7 2 33.3 6 Food Service 4 80.0 1 20.0 5 Homemaker 0 0.0 1 100.0 1 Security/Govern. 2 66.7 1 33.3 3 Service/Industry 3 60.0 2 40.0 5 Technology 5 100.0 0 0.0 5 No Contribution 37 77.1 11 22.9 48 Unknown 53 88.3 7 11.7 60 Note. N = 155 for the following variables, Amoxicillin, Nebulizer/bronchodilator, Zinc, and Multivitamin. Asterisks () indicate significant results. ‡ Refers to Mantel-Haenszel for all non-binary (nominal or ordinal) variables, and Fisher's Exact (–) when minimum expected counts (≥5) were not met. 4.1 Recovery and Discharge Since there was no significant difference in time-to-recovery, the primary outcome variable, cohorts were combined in further analyses. In the total sample (N = 156), 59% (n = 92) of the cases recovered within six weeks, 37% (n = 57) recovered in greater than six weeks, and 4% (n = 7) did not recover within the study timeframe. Overall, mean time-to-recovery, or the mean of the total number of days from admission to recovery (n = 149), was 39.57 days (SD = 25.62, 95% CI [35.5, 43.7]), or 5.65 weeks, which meets the international standard of recovery (<6 weeks). Time-to-discharge, or the total number of days from admission to discharge, for the total sample (N = 156) was an average of 43.97 days (SD = 25.59, 95% CI [40.0, 48.0]), or 6.28 weeks. The discrepancy between time-to-recovery and time-to-discharge existed because 21% (n = 36) recovered nutritionally but had a delay in discharge, ranging from 2-50 days, with a mean of 20.1 days (SD = 14.24, 95%CI [15.5, 24.8]), or 2.87 weeks. The majority of all cases were severely malnourished on admission (n=82, 53%), followed by moderate (n = 69, 44%) and mild (i.e. acute) (n = 5, 3%). There was no significant relationship between malnutrition severity on admission and delayed discharge. 4.2 Sociocultural Variables Data regarding sociocultural characteristics was more limited than projected. Occupation was available for 41% of mothers (n = 64) and 62% of fathers (n = 96). Among mothers, 30% (n = 46) were homemakers, 10% (n = 16) were employed in the service sector (i.e. laundry or food sales), and 1% (n = 2) were not financially contributing to their child's care. There were 30% of fathers (n = 47) employed in professional jobs, such as journalism or law enforcement, and 31% (n = 48) fathers not financially contributing (see Table 2). Mothers’ (n = 141) mean age was 26 (SD = 6.59), ranging from 15-41 years; fathers’ (n = 99) mean age was 30 (SD = 8.85), ranging from 18-66 years. Additionally, 17 cases had no access to running water and 42 cases indicated grandparents were involved in their living situation or providing for their daily needs. Overall, investigators are cautious when interpreting findings related to sociocultural data, due to the large amount of unknown data (see Table 2). Also, while not formally reported in case records, investigators were informed that few families return for follow-up assessments due to lack of transportation, particularly with increased bus fares after the onset of the pandemic (personal communication, November 12, 2021, NRC Director). 4.3 Indicators of Recovery Several predictor variables were analyzed to determine impact on recovery time (see Table 3 ). Univariate analysis indicated a significant association between recovery time and only two predictors, use of amoxicillin (p = .014) and nebulizers/bronchodilators (p = .039). Of note, there was no significant relationship between severity of malnutrition on admission with use of either amoxicillin ꭓ2 (1, N = 155) = .210, p = .659, or nebulizers/bronchodilators, ꭓ2 (1, N = 155) = .619, p = .476 (Data not shown).Table 3 Frequencies and Chi-Square Results Evaluating Categorical Variables and Recovery (N = 156) Table 3 ≥6 weeks or no recovery <6 weeks Univariate Analysis Variables n % n % Total ꭓ2. df OR [95% CI] p COVID Cohorts Post-COVID 10 33.3 20 66.7 30 .91 1 0.67 [0.29,1.54] .341 Pre-COVID 54 42.9 72 57.1 126 Age ≤24 months 56 40.3 83 59.7 139 .29 1 0.76 [0.28,2.09] .592 >24 months 8 47.1 9 52.9 17 Gender Female 29 39.7 44 60.3 73 .10 1 0.90 [0.48,1.71] .757 Male 35 42.2 48 57.8 83 Severity of Malnutrition Mild 1 20.0 4 80.0 5 3.44 1 0.28 [0.03,2.57] .064 Moderate 24 34.8 45 65.2 69 0.47 [0.05, 4.43] Severe 39 47.6 43 52.4 82 0.59 [0.30, 1.14] Admit Weight <7 kg 49 45.4 59 54.6 108 2.74 1 1.83 [0.89,3.75] .098 ≥7 kg 15 31.3 33 68.8 48 Amoxicillin Yes 30 53.6 26 46.4 56 6.07 1 2.31 [1.18,4.52] .014* No 33 33.3 66 66.7 99 Multivitamin Yes 17 34.0 33 66.0 50 1.35 1 0.66 [0.33,1.33] .245 No 46 43.8 59 56.2 105 Nebulizer/ Bronchodilator Yes 38 48.7 40 51.3 78 4.24 1 1.98 [1.03,3.79] .039* No 25 32.5 52 67.5 77 Variables n % n % Total ꭓ2. df OR [95% CI] p Zinc Yes 57 43.5 74 56.5 131 2.88 1 2.31 [0.86,6.20] .090 No 6 25.0 18 75.0 24 Note. N = 155 for the following variables, Amoxicillin, Nebulizer/bronchodilator, Zinc, and Multivitamin. Mantel-Haenszel was reported for non-binary variables (Severity of Malnutrition). Asterisks () indicate significant results. Double asterisks (*) indicate significance in both univariate and multivariable analyses (Data for multivariate not shown). Multiple logistic regression was conducted to assess the impact of the nine predictor variables on the outcome variable (recovery in ≤6 weeks). Only one predictor variable of the model, amoxicillin, was significant, Wald ꭓ2 = 4.14, p = .042. The adjusted odds ratio for amoxicillin was aOR = 2.14 [1.03,4.47], indicating children receiving amoxicillin were 2.14 times more likely to recover in >6 weeks, after controlling for other predictor variables (see Table 3). 5 Discussion This retrospective chart review provides insight on time-to-recovery among children admitted to an NRC in Guatemala before and during COVID-19. The lack of difference in the primary outcome variable, time-to-recovery, between the pre- and post-COVID cohorts may be due to the considerably small sample size in the post-COVID. The small sample was likely impacted by both fear of the pandemic, and new isolation policies at the NRC. There was also a difference in referral trends between cohorts; with fewer self-referrals in the post-COVID cohort (n = 2) than the pre-COVID cohort (n = 31). This would suggest that families were less inclined to seek out treatment or experienced other challenges post-pandemic, preventing them from seeking treatment. The significantly higher weight gain and discharge weight post-COVID compared to pre-COVID was unanticipated. This could be explained by the small post-COVID cohort receiving closer attention by NRC staff, due to the lower staff to child ratio. Overall, average time-to-recovery, the rate of weight gain, and lack of fatalities in this sample are evidence that the NRC has effectiveness levels comparable to other NRCs and OTPs21 , 35 , 36. One concern is the delay many children faced between recovery and discharge. Due to the nature of the medical records (paper charts) used to collect data, there is limited insight as to why children experienced these delays. Often this is due to treatment of secondary illnesses, such as respiratory or gastrointestinal infections, or adverse social conditions that delay a safe discharge plan (personal communication, November 9, 2021, NRC Director). However, this could not be verified in the data. Although the Social Ecological Model38 informed the study, findings were primarily at the individual level and macro-level data was sparse. Despite limited data, the sociocultural findings in this study provide further environmental concern for child health in this setting. There was some evidence that these families relied on extended family for housing, financial needs, and childcare. Additionally, when evaluating parents’ employment, mothers were most often homemakers, and fathers were not contributors to family financial well-being. Compounding this, working mothers’ occupations were lower-paying jobs (i.e. food sales or laundry) compared to fathers (i.e. law enforcement, banking). Employment opportunities and family obligations are ongoing challenges for Guatemalan women10 , 11. Additionally, in some cases, families were without access to transportation or potable water. For these reasons, collecting sociocultural data upon admission, it is essential to consider the impact of limited income and transportation barriers. Amoxicillin was the only significant predictor of time-to-recovery in this study. Previous studies evaluating amoxicillin as treatment for malnutrition have had inconsistent findings; some indicate it contributes to recovery, others found no benefit24 , 25. Interestingly, in this sample, the relationship between amoxicillin and recovery was the inverse of what was expected; it was found to be associated with slower recovery rates. Further, there was no significant relationship between amoxicillin and severity of malnutrition. A known side effect of amoxicillin is increased risk of diarrhea39, which could be a possible explanation of the slower recovery rates in these children. Of note, amoxicillin was not widely used in this sample (n = 56, 36%). Predictor variables in other studies, i.e., gender, age, vitamin supplementation, and admission weight less than 7 kg29 , 30 , 32 had no influence on recovery in this study. This could be related to the homogeneity of the cases; for example, there was a nearly even distribution between males and females, and most children (n = 139, 89%) were ≤2 years. Another unanticipated finding was the widespread use nebulizers/bronchodilators (n = 78, 50%), suggesting respiratory illnesses. An absence of secondary diagnoses in the case records prevents analysis of this in the data. While medications were not the focus of this study, numerous medications were prescribed to children during their recuperation, such as, acyclovir, acetaminophen, folic acid, and reflux prescriptions. Future investigation of medications might provide insight to secondary diagnoses and prescription practices for malnutrition. 5.1 Strengths and Limitations This study had several strengths. The findings add to our understanding of malnutrition in a vulnerable population and provides timely data on a population with limited representation in the literature within the context of a pandemic. Also, this study built on a community-university partnership and NRC staff provided the local perspective; such partnerships are critical to advance the science of global health. Two limitations were in the sample and setting. A smaller sample in the post-COVID cohort could have negatively affected the statistical power of the tests used in the analysis. Also, the post-COVID sample came from only 10 months, as opposed to 14 months pre-COVID. Still, the post-COVID sample suggests the impact of the pandemic on child health. It was anticipated that the effects of the pandemic would have resulted in an intense need for nutrition services like the NRC; however, the services were less utilized than in previous years. Additionally, data was only collected from one NRC in Guatemala (albeit the largest), thus, findings have limited generalizability. Collecting data from case records handwritten in Spanish was a challenge; in cases where notes were not discernible, the PI consulted the NRC director. 5.2 Implications for Policy, Research, and Practice A health policy concern at the time of this study was the need for increased availability of COVID-19 testing in Guatemala. Access to testing would decrease the need for a lengthy isolation period for newly admitted children, improving parent's experience and receptiveness toward receiving needed care, as well as promoting staff safety. Another policy issue is management of acute malnutrition, which is generally not considered a major health concern by the Guatemalan Ministry of Health (personal communication, November 23, 2021, NRC Nutritionist). However, close monitoring of acute cases is critical to mitigate worsened conditions. Five cases (1 pre-COVID, 4 post-COVID) in this study were admitted with acute malnutrition; these cases could have been managed within the community through collaboration with CHWs, supplemental nutrition, and ongoing evaluation33. Development of CHW outreach with rural and indigenous communities7 in the Southern region may be an area to focus on in future intervention studies. When provided proper training and support from public health nurses, CHWs can monitor and track at-risk children, in line with the USAID4 current work in supporting rural indigenous populations. Future research could further explore the long-term effects the COVID-19 pandemic on childhood malnutrition. Investigation of secondary diagnoses and detailed medication uses could provide understanding of the additional medical complexities commonly experienced by malnourished children and the impacts of these complexities on recovery. A prospective cohort design might decrease the likelihood of missing data. A multi-site study evaluating the impact of amoxicillin on recovery would solidify understanding of its effectiveness and risks. A number of features in this dataset could benefit from a qualitative descriptive case study for a greater depth of understanding from the perspectives of families and staff. A qualitative study investigating the sociocultural context could provide needed data to increase the effectiveness of SAM recovery interventions. In clinical practice, understanding the sociocultural characteristics that impact child health is key, as these factors inform all aspects of care, including health education and discharge planning. A strategy to increase access to sociocultural data among malnourished children is to implement an admission family needs assessment, including a home evaluation. This will allow clinicians to quickly identify upstream factors contributing to food insecurity, such as potable water and housing conditions. The home evaluation piece is essential to establish childcare and home safety needs on admission, in order to prevent children staying at the center longer than necessary, increasing risk of contracting communicable diseases. Once needs are identified, families can be connected with appropriate governmental resources, such as the health department, as well as local non-profit resources dedicated to mothers and children, such as Corazon de Los Niños (http://www.corazondelosninos.org/) or other university partnerships. A family needs assessment could also provide further evidence of population health necessities; this information could serve in procurement of additional resources and funding. 6 Conclusion This study informs our understanding of malnutrition recovery at a Guatemalan NRC before and during the COVID-19 pandemic. The discovery of relevant information on clinical variables in a vulnerable population informs potential program development strategies within the NRC. Further research is needed to more fully understand the long-term complexities that the pandemic has had on childhood malnutrition recovery. Findings may be utilized in further investigation and intervention development to reduce malnutrition in LMICs amid the ongoing COVID-19 pandemic. Financial Support This study was supported by East Carolina University College of Nursing's doctoral student research grant. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We would like to sincerely thank the leadership of Casa Jackson, a program of La Asociación Nuestros Ahijados. In particular, we want to express our appreciation to Vecany Sanchez and Robbie Middleton, for their partnership with the research team. ==== Refs References 1 World Health Organization [WHO]. Social determinants of health. https://www.who.int/health-topics/social-determinants-of-health#tab=tab_1, 2021(accessed 16 May 2023). 2 United Nations International Children's Emergency Fund [UNICEF]. Malnutrition. https://data.unicef.org/topic/nutrition/malnutrition/, 2021 (accessed 16 May 2023). 3 United Nations International Children's Emergency Fund [UNICEF]. Children, food and nutrition. https://www.unicef.org/media/106506/file/The%20State%20of%20the%20World%E2%80%99s%20Children%202019.pdf, 2019 (accessed 16 May 2023). 4 United States Agency for International Development [USAID]. Nutrition profile: Guatemala. https://2017-2020.usaid.gov/sites/default/files/documents/1864/Guatemala-Nutrition-Profile-Mar2018-508.pdf, 2021 (accessed 16 May 2023). 5 Flores-Quispe M. Restrepo-Méndez M.C. Maia M. Ferreira L.Z. Wehrmeister F.C. Trends in socioeconomic inequalities in stunting prevalence in Latin America and the Caribbean countries: Differences between quintiles and deciles Int. J. Equity Health 18 2019 156 10.1186/s12939-019-1046-7 31615530 6 Batis C. Mazariegos M. Martorell R. Gil A. Rivera J.A. Malnutrition in all its forms by wealth, education and ethnicity in Latin America: who are more affected? Public Health Nutr 23 2020 s1 s12 10.1017/S136898001900466X 32900396 7 Gatica-Domínguez G. Victora C. Barros A. Ethnic inequalities and trends in stunting prevalence among Guatemalan children: an analysis using national health surveys 1995-2014 Int. J. Equity Health 18 2019 110 10.1186/s12939-019-1016-0 31319862 8 Kragel E.A. Merz A. Flood D. Haven K.E. Risk factors for stunting in children under the age of 5 in rural Guatemalan highlands Ann. of Glob. Health 86 2020 8 10.5334/aogh.2433 32064226 9 Voth-Gaeddert L.E. Oerther & D.B. Acute to chronic malnutrition: How significant water, sanitation, and hygiene factors change with health outcomes and geographies in the western highlands of Guatemala J. of Enviro. Health 82 2019 20 28 10 Burroway R. Are all jobs created equal? A cross-national analysis of women's employment and child malnutrition in developing countries Soc. Sci. Res. 67 2017 1 13 10.1016/j.ssresearch.2017.07.003 28888278 11 Ekbrand H. Halleröd B. The more gender equity, the less child poverty? A multilevel analysis of malnutrition and health deprivation in 49 low- and middle-income countries World Dev 108 2018 221 230 10.1016/j.worlddev.2018.01.028 12 Oddo V.M. Surkan P.J. Hurley K.M. Lowery C. Ponce S. Jones-Smith J.C. Pathways of the association between maternal employment and weight status among women and children: Qualitative findings from Guatemala Matern & Child Nutr 14 2018 e12455 10.1111/mcn.12455 13 Juarez M. Dionicio C. Sacuj N. Lopez W. Miller A.C. Rohloff P. Community-based interventions to reduce child stunting in rural Guatemala: A quality improvement model Int. J. Environ. Res. Public Health 18 2021 773 10.3390/ijerph18020773 33477580 14 Olney D.K. Leroy J. Bliznashka L. Ruel M.T. PROCOMIDA, a food-assisted maternal and child health and nutrition program, reduces child stunting in Guatemala: A cluster-randomized controlled intervention trial J Nutr 148 2018 1493 10.1093/jn/nxy138 30184223 15 Guzmán-Abril A. Alajajian S. Rohloff P. Proaño G.V. Brewer J. Jimenez E.Y. A home garden intervention improves child length-for-age z-score and household-level crop count and nutritional functional diversity in rural Guatemala J Acad Nutr Diet 122 2021 640 649 10.1016/j.jand.2021.04.002 34020932 16 Rohloff P. On the frontlines of chronic paediatric undernutrition in Guatemala EBioMedicine 64 2021 103223 10.1016/j.ebiom.2021.103223 17 World Health Organization [WHO]. Archived: WHO timeline- COVID-19. https://www.who.int/news/item/27-04-2020-who-timeline-covid-19, 2020 (accessed 16 May 2023). 18 Action Against Hunger. Number of people facing hunger doubled due to COVID-19. https://www.actionagainsthunger.org/story/number-people-facing-hunger-guatemala-doubled-due-covid-19, 2020 (accessed 12 November 2021). 19 People for Guatemala. Coronavirus crisis in Guatemala: A timeline. https://www.peopleforguatemala.org/covid-19/coronavirus-crisis-in-guatemala-a-timeline/, 2020 (accessed 12 November 2021). 20 Victora C.G. Christian P. Vidaletti L.P. Gatica-Domínguez G. Menon P. Black R.E. Revisiting maternal and child undernutrition in low-income and middle-income countries: Variable progress towards an unfinished agenda Lancet 397 2021 1388 1399 10.1016/S0140-6736(21)00394-9 33691094 21 Kabalo M.Y. Seifu C.N. Treatment outcomes of severe acute malnutrition in children treated within Outpatient Therapeutic Program (OTP) at Wolaita Zone, Southern Ethiopia: retrospective cross-sectional study J Health Popul Nutr 36 2017 7 10.1186/s41043-017-0083-3 28279227 22 Sphere Project. The Sphere handbook: Humanitarian charter and minimum standards in humanitarian response. http://www.sphereproject.org/handbook, 2011 (accessed 12 November 2021). 23 M.Palacios A. Hurley K.M. De-Ponce S. Alfonso V. Tilton N. Lambden K.B. Reinhart G.A. Freeland-Graves J.H. Villanueva L.M. Black M.M. Zinc deficiency associated with anaemia among young children in rural Guatemala Matern & Child Nutr 16 2020 e12885 10.1111/mcn.12885 24 World Health Organization [WHO]. Severe acute malnutrition update: Current WHO guidelines and the WHO essential medicine list for children. https://www.who.int/selection_medicines/committees/expert/21/applications/s6_paed_antibiotics_appendix7_sam.pdf, 2016 (accessed 12 November 2021). 25 Isanaka S. Grantz K.H. Berthé F. Schaefer M. Adehossi E. Grais R.F. Extended follow-up from a randomized clinical trial of routine amoxicillin in the treatment of uncomplicated severe acute malnutrition in Niger JAMA Pediatr 174 2020 295 297 10.1001/jamapediatrics.2019.5189 31930369 26 Nel E. Severe acute malnutrition Curr Opin Clin Nutr Metab Care 21 2018 195 199 10.1097/MCO.0000000000000465 29461265 27 World Health Organization [WHO]. WHO child growth standards and the identification of severe acute malnutrition in infants and children. https://apps.who.int/iris/bitstream/handle/10665/44129/9789241598163_eng.pdf;jsessionid=C0453EB7C82A0AABCDD77015F4F73DAB?sequence=1 (accessed 14 June 2023). 28 Liben M.L. Wuneh A.G. Shamie R. G/Her K. Factors associated with child survival in children admitted to outpatient therapeutic program at public health institutions in Afar Regional State, Ethiopia: a prospective cohort study J Health Popul Nutr 38 2019 35 10.1186/s41043-019-0193-1 31775913 29 Mamo W.N. Derso T. Gelaye K.A. Akalu T.Y. Time to recovery and determinants of severe acute malnutrition among 6–59 months children treated at outpatient therapeutic programme in North Gondar zone, Northwest Ethiopia: A prospective follow up study Ital. J. Pediatr. 45 2019 136 10.1186/s13052-019-0732-93 31684989 30 Teshome G. Bosha T. Gebremedhin S. Time-to-recovery from severe acute malnutrition in children 6–59 months of age enrolled in the outpatient treatment program in Shebedino, southern Ethiopia: A prospective cohort study BMC Pediatr 19 2019 10.1186/s12887-019-1407-9 31 Doocy S. Tappis H. Villeminot N. Suk A. Kumar D. Fazal S. Grant A. Pietzsch S. Point-of-use water treatment improves recovery rates among children with severe acute malnutrition in Pakistan: results from a site-randomized trial Public Health Nutr 21 2018 3080 3090 10.1017/S1368980018001647 30132426 32 Gebremedhin K. Ayele G. Boti N. Andarge E. Fikadu T. Predictors of time-to-recovery from severe acute malnutrition treated in an outpatient treatment program in health posts of Arba Minch Zuria Woreda, Gamo zone, Southern Ethiopia: A retrospective cohort study PloS One 15 2020 e0234793 10.1371/journal.pone.0234793 33 López-Ejeda N. Charle Cuellar P. Vargas A. Guerrero S. Can community health workers manage uncomplicated severe acute malnutrition? A review of operational experiences in delivering severe acute malnutrition treatment through community health platforms Matern Child Nutr 15 2019 e12719 10.1111/mcn.12719 30315743 34 Rastogi S. Maheshwari C. Raghav S.K. Muzammil K. A prospective observational study to evaluate the efficacy of facility-based management in malnourished children at NRC, district Meerut J Family Med Prim Care 7 2018 1341 1345 10.4103/jfmpc.jfmpc_157_18 30613522 35 Chaturvedi A. Patwari A.K. Soni D. Pandey S. Prost A. Gope R.K. Sharma J. Tripathy P. Progress of children with severe acute malnutrition in the malnutrition treatment centre rehabilitation program: evidence from a prospective study in Jharkhand, India Nutr. J. 17 2018 69 10.1186/s12937-018-0378-2 30021572 36 Golandaj J.A. Vatavati S.R. Kallihal K.G. Effect of nutritional interventional measures on admitted children in selected nutritional rehabilitation centres of Karnataka India. Nutr. Food Sci. 47 2017 53 66 10.1108/NFS-06-2015-0078 37 Authors A feasibility study to examine clinical variables of childhood malnutrition in Guatemala Hisp. Health Care Int. 2023 1 8 10.1177/15404153221150452 38 U. Bronfenbrenner, The ecology of human development. Harvard University Press, 1979. 39 National Library of Medicine. Amoxicillin. https://medlineplus.gov/druginfo/meds/a685001.html (accessed 14 June 2023).
PMC010xxxxxx/PMC10286526.txt	==== Front Applied Corpus Linguistics 2666-7991 2666-7991 The Authors. Published by Elsevier Ltd. S2666-7991(23)00023-0 10.1016/j.acorp.2023.100063 100063 Article Directives in Covid-19 government guidance: an international comparison Vincent Benet 1⁎ Power Kate 2 Crosthwaite Peter 2 Gardner Sheena 1 1 Coventry University 2 University of Queensland ⁎ Corresponding author, Dr Benet Vincent, Coventry University 22 6 2023 22 6 2023 100063© 2023 The Authors. Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The importance of language to changing public behaviours is acknowledged in crisis situations such as the COVID-19 pandemic. A key means of achieving these changes is through the use of directive speech acts, yet this area is currently under-researched. This study investigates the use of directives in the 2020 COVID-19 briefings of four leaders of English-speaking nations, Jacinda Adern, Boris Johnson, Scott Morrison, and Nicola Sturgeon. We developed a classification system including 13 directive types and used this to compare directive use across these four leaders, examining directness and forcefulness of directive use. The analysis finds Sturgeon to be the most prolific directive user and also to have the highest reliance on imperatives. Johnson, meanwhile, has a preference for directives involving modal verbs, particularly with first- and second-person pronouns. In contrast, Ardern and Morrison show a higher use of indirect directives, normally thought to be a less effective strategy. While Ardern often combines this strategy with judicious use of imperatives, this is not seen in Morrison's COVID-19 briefings. These findings tend to confirm earlier, more impressionistic evaluations of the communication styles of these leaders but also suggest other avenues for research on directive use. We conclude with implications for political crisis communication and analysis of directives in crisis communication. Key terms speech acts directives political discourse Covid-19 corpus pragmatics ==== Body pmc1 Introduction The sudden outbreak of COVID-19 in early 2020 presented governments around the world with the challenge not just of deciding the best course of action but also of communicating this effectively with the public (Finset et al., 2020). In such situations “the public expect the government to be a fast and reliable source of information” (UK Government Communication Service, 2021) and leaders need to “communicate clear consistent messages in an empathetic manner” (McGuire et al., 2020, p. 361), “[a]cknowledging team effort… while also taking personal responsibility where appropriate” (Marsen & Ali-Chand, 2022, p. 23). Numerous studies have shown that communication strategies can have a significant effect on public adoption of the behaviours governments wish to promote (e.g., Dada et al., 2021; Hansson, 2017; Lunn et al., 2020; McGuire et al., 2020). Language plays a vital part in “transforming political will into social action” (Taylor & Partington, 2018, p. 1) and in managing crisis situations (Lunn et al., 2020; Marsen & Ali-Chand, 2022; Nielsen et al., 2020; Sanders, 2020). Of particular importance in this respect are directives, that is, “utterances designed to get someone to do something” (Goodwin, 2006, p. 517). As Searle (1979) points out, directives may be framed in a range of different ways to show different levels of directness and forcefulness, from inviting to insisting (see Section 2.1). Our interest is in how they have been used by political leaders in the context of instructions and recommendations during the COVID-19 pandemic. An important way in which instructions and recommendations have been delivered during the pandemic is in press conferences and other public briefings. The application of corpus approaches to understanding how politicians communicate in such contexts is well-established (Ädel, 2010). In this study, we focus on four leaders of island nations on opposite sides of the world: Boris Johnson and Nicola Sturgeon in the UK, Jacinda Ardern in New Zealand and Scott Morrison in Australia. Reactions to crisis communication by these leaders have widely differed. Ardern has generally been praised (e.g., Dada et al., 2021; McGuire et al., 2020; Menon, 2020; Reyes Bernard et al., 2021). Johnson and his government, in contrast, have faced extensive criticism (e.g., Jones, 2021; Oliver, 2020; Sodha, 2020). The Australian Prime Minister (PM), Scott Morrison has also faced criticism for confusing messaging (Davey, 2020) and lack of empathy (Reyes Bernard et al., 2021). Less has been written about Sturgeon, but her people-oriented, empathetic approach is noted by Dada et al. (2021). These evaluations are particularly interesting when viewed against the greatly contrasting outcomes in these leaders’ management of the pandemic. New Zealand and Australia were ranked the first and eighth most effective nations at managing the pandemic in 2020, while the UK came in 66th (Leng et al., 2021). It is interesting to consider the extent to which these outcomes might be associated with communication strategies. While there is a long tradition of research on both political discourse and directives, the two have seldom been considered together, particularly in the context of crisis communication, as noted by Marsen & Ali-Chand (2022). This study compares the use of directives by Ardern, Johnson, Morrison, and Sturgeon using corpora of transcripts of COVID-19 briefings throughout 2020, to discover how they attempted to ensure compliance from the public with the measures they introduced. As Finset et al. (2020) point out, the way such recommendations are delivered is vital in ensuring compliance with measures designed to protect public health. Therefore, studying them is of interest to the communications teams supporting national and other leaders, who face the challenge of conveying the seriousness of crisis situations and motivating public action without unduly threatening their audience's face and thus risking noncompliance. 2 Literature review 2.1 Categorising directives: directness and forcefulness As suggested in Section 1, directives encompass a range of more specific speech acts (e.g. request, invite, encourage, command) which can be realised using various linguistic devices (Searle 1979). The choices made are generally agreed to depend on contextual factors such as relationship between speaker and hearer (Brown and Levinson, 1987). Understanding how realisations differ and what this means can be achieved by considering directives in terms of directness and forcefulness. Directness is the extent to which “people literally say what they mean” (Ervin-Tripp 1976, p. 26; Searle 1979). A number of frameworks have been proposed that recognise different levels of directness, typically from Hint (entirely indirect) to the most direct form, the Imperative. One of the most influential of these is presented in Ervin-Tripp (1976); this framework was developed by House and Kasper (1981), who distinguish eight levels of directness (see Table 1 ) relating to realisations of requests. This framework is readily applicable to the wider category of directives. In the example provided by House and Kasper, someone asking (or directing) another person to open a window might vary the directness with which this request (or direction) is expressed from highly indirect (e.g., “it's very cold in here” makes no reference to what should be done in response) to highly direct (e.g., “close the window”, which leaves the hearer in no doubt as to what they are being asked to do). Considering the level of directness is important because one can be clear by being direct, but being direct may impinge on matters of politeness (Ervin-Tripp, 1976; Searle, 1979; Weigel & Weigel, 1985) and threat to negative face, although in cases of urgency such considerations may be overridden (Brown & Levinson, 1987; Goodwin, 2006; Vine, 2009).Table 1 Levels of directness in requests (adapted from House & Kasper, 1981, p. 163) Table 1:1. Mild Hint It's very cold in here 2. Strong Hint Why is the window open? 3. Query-Preparatory Can you close the window? 4. State-Preparatory You can close the window 5. Scope-Stating I would prefer it if you closed the window 6. Locution-derivable You should close the window 7. (a) Hedged-Performative I must ask you to close the window (b) Explicit-Performative I ask you to close the window 8. Mood-derivable Close the window! A dimension additional to directness that helps distinguish between realisations of directives relates to their forcefulness, “degree of intensity” (Searle, 1979, p. 28) or “strength” (Sbisà, 2001). As well as making choices over how direct to be, a speaker can vary the intensity of a directive even while retaining the same level of directness. It is generally considered more forceful, for example, to say you must than to utter you should. This dimension is well documented in the literature on modality, where it is related to “value” (e.g. Halliday and Matthiessen, 2014, p. 149) or “strength” (Huddleston & Pullum, 2002, p. 179) but it is not so commonly discussed in pragmatics. Vine (2009, p. 1395), for example, refers to the imperative as “the most forceful form” without distinguishing between types of imperative which could be used to alter this forcefulness (compare “close the window” and “let's close the window”). Politicians giving instructions can be seen as selecting from these options in relation to contextual factors, including the perceived urgency of the situation and the relationship they want to present as obtaining between themselves and their audience. An important question, therefore, is the extent to which choices of this sort are handled in leaders’ COVID-19 crisis communication. 2.2 COVID-19 and research on strategies in crisis communication The COVID-19 pandemic has sparked great interest in crisis and risk communication, particularly around how behavioural change can be managed in similar crisis contexts. The realisation in early 2020 that the COVID-19 pandemic presented serious unprecedented challenges led to a number of position papers proposing how communication should be handled and suggesting areas of research that might inform more effective communication by leaders. Jaspal and Nerlich (2020) predicted some of the difficulties involved in governments seeking compliance to measures that potentially threaten individuals’ identities by changing their routines. They warned against emphasising negative emotions such as fear, instead proposing communication strategies that engage diverse groups and frame public health measures in more positive terms, such as team spirit. Lunn et al. (2020, p. 4) likewise emphasised the importance of using inclusive language, making “clear statements of a desired collective behaviour”, and presenting such behaviour as benefiting all of society, a viewpoint shared by Finset et al. (2020). One important area of research has looked at the crisis communication strategies of national leaders during the pandemic in speech events involving communication with the public (press conferences/briefings, speeches to parliament, statements to the nation, media releases). Research in this vein highlights aspects of this communication such as the extent to which leaders or groups of leaders (e.g. men vs. women – Dada et al., 2021) used particular strategies or mentioned specific topics. One vital aspect of crisis communication mentioned in Section 1 is the clarity and consistency of messaging, and this focus is apparent in several studies of national leaders’ COVID-19 communication. In their study of Ardern's interactions with the NZ public, McGuire et al. (2020) argue that she communicates “clearly and formally” (p. 368). Reyes Bernard et al. (2021) also describe Morrison as a clear communicator, although this claim seems inconsistent with the observation of Marsen and Ali-Chand (2022) that he is the “most indirect” (p.10) of the three leaders they study (Ardern, Morrison, and Bainimarama, the leader of Fiji). The UK government, and by implication Boris Johnson were found by Jones (2021), Nielsen et al. (2020) and Sanders (2020) to have created confusion through inconsistent messaging and thereby to have lost public trust. These findings are based mostly on content analysis not involving close linguistic analysis of speeches, such that a complete picture of the effectiveness of UK government messaging is does not emerge from this area of research. This shortcoming is also notable in the work of McGuire et al. (2020) and Reyes Bernard et al. (2021), both of whom claim that the leaders they analyse communicate clearly without defining what clarity might be, for example by equating it with simple language or a lack of jargon as McClaughlin et al. (2021) and Wolf (2011) do. A further important finding from this group of studies concerns the discursive strategies used to encourage behaviour change. Leaders commonly invoke ideas of social solidarity (Dada et al., 2021; McGuire et al., 2020), although not necessarily in the same way. Based on the analysis of 122 speeches given by 20 world leaders (half male, half female), Dada et al. (2021) found that male leaders focus more on war rhetoric, while women – including Jacinda Ardern and Nicola Sturgeon – favour a more compassionate approach based on empathy. As Dada et al. (2021) point out, “[w]hile war rhetoric plays to a collectivism based on fear and division, empathy appeals to a collectivism based on compassionate social cohesion” (p. 10). Ardern, in particular, is consistently praised for being sympathetic and approachable (McGuire et al., 2020; Marsen & Ali-Chand 2022), while Reyes Bernard et al. (2021) point out that Morrison largely avoids expressing empathy. Studies on strategies of crisis communication employed by national leaders in the pandemic have also noted the prominence of appeals to the public to follow health guidelines through an emphasis on what Dada et al. (2021) term “responsibility” and “paternalism”1 . By “responsibility” Dada et al. (2021) mean that leaders “encourage individuals to act independently” while “paternalism” refers to “employ[ing] tactics such as shame, guilt, or punishment to influence the desired behaviour” (p. 7). They find that both of these rhetorical strategies were employed by most of the leaders of either gender in their study (including Ardern, Sturgeon and Johnson). Without using the same terms, similar concepts are referred to or exemplified in McGuire et al. (2020) and Marsen and Ali-Chand (2022). These strategies are of particular interest for the present study because they relate to instructions and how they are delivered, drawing attention to the importance of directives. However, the studies examined in this section do not generally identify which linguistic features are used to create the rhetorical effects they observe, even though the use of language seems key to the intended effect. 2.3 Studies on the linguistic aspects of COVID-19 crisis communication As noted in the previous section, a series of studies on strategies of crisis communication in the COVID-19 pandemic has produced interesting results without generally focusing on specific linguistic features. The exception is Marsen and Ali-Chand (2022), whose study compares the use of speech acts in six key speeches given by the leaders of Australia, New Zealand and Fiji in the period of March-June 2020. Their focus on speech acts entails closer attention to language use and how it differs from leader to leader. As they point out, the “ways in which directives are framed indicate the relationship between interactants, matters of status and authority, and possible expectations of addressees” (Marsen & Ali-Chand, 2022, p. 24). Directives were the most frequently occurring speech act in all three leaders’ communications, but they were used in different ways. Morrison is found to be the most indirect due in part to his use of hedging, while Ardern combines expressions of sympathy with directives which, Marsen and Ali-Chand argue, has the effect of reducing the force of the directives. While these findings from Marsen and Ali-Chand (2022) are revealing, there are some aspects of their approach in need of adjustment for a study focusing on the linguistic realisations of directives. The first of these concerns how to deal with indirect speech acts. In seeking to avoid overlap between different speech acts, Marsen and Ali-Chand did not focus on indirect directives but classed utterances according to their face-value speech act. Taking this approach removes from the scope of investigation some instances of declaratives (e.g. reference to a rule) and commissives (e.g. promises, warnings and threats)2 which we would want to include. This approach also narrowed the set of forms counted as realising directives to imperatives, modals of obligation3 , and what Marsen and Ali-Chand refer to as “‘I want’ and ‘I ask’ statements”4 , although other forms are known to conventionally realise directive speech acts (see Section 3.2). A further aspect of directives not investigated by Marsen and Ali-Chand which seems important in the context of directives is the contrast between directness and forcefulness. The final aspect of Marsen and Ali-Chand (2022) – and indeed all of the studies mentioned in section 2 so far – that we feel could be built on is that they do not make use of corpus methods to arrive at, support or complement their findings. There are, however, a number of studies which have adopted corpus approaches to study crisis communication during the pandemic. The crisis communication of Ardern and Morrison has already been examined in a corpus study by AUTHOR2 and AUTHOR3 (2022), who contrasted the keywords in each PM's 2020 COVID briefings, to probe their discursive styles and examine the association between the styles of each PM and public perception of how well they managed the pandemic. Their findings give empirical linguistic support to earlier claims about Ardern's interpersonal, empathetic approach, indicating the importance of the use of personal pronouns in clear communication, and in combination with if-clauses providing clear procedural instructions. Similarly, these findings support observations from others (e.g. Marsen & Ali-Chand, 2022) that Morrison uses language in ways that avoid taking responsibility for unpopular moves while simultaneously claiming credit for government decision-making. Another study of interest is that reported by Williams and Wright (2020, 2022), who contrast the use of inclusive we - also termed “patriotic” we (Wales, 1996) - and exclusive we in Downing Street briefings from March-June 2020. Inclusive we refers to the speaker and their audience/interlocutors, while exclusive we refers to the speaker and other parties not present but excludes the audience. Williams and Wright found that government spokespeople tended to use exclusive we (i.e. where we did not include the general public) in constructions which acted to distance them from responsibility for key actions, something that did not apply in instances of inclusive we. This pattern included a number of instances where government spokespeople were uttering directives such as extract 1 (marked by have to).(1) we have to take special steps to protect the particularly vulnerable (Johnson, 22 March 2020) In this example, Williams and Wright (2020) argue, the key action referred to is to protect the vulnerable but the responsibility for doing so is subtly distanced from the government (we) to the steps. This study thus provides an interesting counterpoint to studies such as Marsen and Ali-Chand (2022) that assume we is straightforwardly a marker of unity/togetherness. Another study benefiting from corpus techniques and focusing on Johnson's COVID-19 communication strategies is McClaughlin et al. (2021), which examines speeches given between March 2020 and April 2021. This study also identifies the salience of we in the context of bringing people together, observing that the actions thereby referred to are commonly quite vague. McClaughlin et al. (2021) note the co-occurrence of we with must, arguing that “Johnson's instructions are presented as a collective obligation” (p. 4; distinction is not made between inclusive and exclusive we). They also point out the strategies of expressing gratitude to the public and showing empathy to support instruction giving, but criticise Johnson's communication for its over-reliance on reference to personal responsibility, for the counterproductive use of war metaphors, for contradictory messaging, and for vagueness and for lack of concision. Several key themes emerge from this review of research into national leaders’ COVID-19 crisis communication concerning the strategies they use to persuade the public to follow instructions. We have seen that much has been written about both the clarity and consistency of communication, identifying strategies such as appealing to social solidarity, showing empathy and taking or avoiding responsibility. However, there has been no systematic focus on one of the key means of persuasion, the ways directives are formulated. Yet greater understanding of this instruction giving aspect of crisis communication seems particularly important for understanding and potentially drawing lessons from different leaders’ individual styles and strategies. 3 Methodology 3.1 Data Collection Notwithstanding some notable differences in both their preferred channels of communication, all four of the leaders included in this study spoke publicly about COVID-19 throughout the first year of the pandemic, hosting press conferences and other official briefings, and making formal speeches designed to guide popular sense-making and behaviour. So as to compare only like genres, we retrieved from each leader's official media relations website transcripts of every press conference and speech focusing on COVID-19 published between January and December 2020. In selecting this time frame, our aim was to map these leaders’ respective discursive approaches to containing the pandemic by guiding public behaviour before widespread vaccination became possible. In collecting our corpus, we focused on the monologic segments of the briefings rather than including dialogic question-and-answer sessions which often followed them. One reason for focusing on monologic contexts is that most previous work on directives has focused on dialogic contexts (e.g. Bax, 1986; Curl & Drew, 2008; Ervin-Tripp, 1976; Weigel & Weigel, 1985). This focus was also partly determined on the grounds of consistency, since not all of the announcements were followed by dialogic question-and-answer sessions. A further factor in our decision was availability, since in most cases only the monologic parts of briefings were transcribed. We extracted from press conference transcripts only the official speeches or opening remarks made by each national leader. We also included COVID-19 focused speeches delivered outside press conference settings, such as Johnson's lockdown announcement on 23 March 2020. Together, these press conference excerpts together with additional speeches (hereafter ‘briefings’) comprise our total corpus. As shown in Table 2 below, the four leaders’ respective sub-corpora vary both in size and publication dates. Ardern and Morrison both commenced COVID-19 briefings more than a month before either Johnson or Sturgeon. Ardern also ceased publishing COVID-19 briefings around two months earlier than the other leaders, although she continued Facebook Live briefings until 21 December 2020. Sturgeon was by far the most prolific of the four and also averaged the highest number of words per briefing, which is why her sub-corpus is much larger than the others. Ardern had relatively few briefings because she varied her output, also running 115 Facebook Live sessions not included here as they were largely dialogic and not of a type with official briefings, being more informal in nature. Johnson's low number of briefings, meanwhile, can be attributed partly to the decision by the UK government to have a range of different speakers hold the Downing Street briefings (his contribution accounts for only 37 of the 119 briefings held in this period), a practice which resulted in some inconsistency in messaging (Oliver, 2020).Table 2 Corpus composition Table 2: Ardern (JA) Johnson (BJ) Morrison (SM) Sturgeon (NS) Role Prime Minister, New Zealand Prime Minister, United Kingdom Prime Minister, Australia First Minister, Scotland Media relations website https://www.beehive.govt.nz/minister/rt-hon-jacinda-ardern https://www.gov.uk/government/collections/slides-and-datasets-to-accompany-coronavirus-press-conferences https://www.pm.gov.au/media https://www.gov.scot/collections/first-ministers-speeches/ Total briefings 33 37 76 152 Size (tokens) 43,577 46,635 136,928 324,539 Mean word count per file 1320.5 1260.4 1801.7 2135.1 Start/end date 28 January - 5 October 3 March - 30 December 29 January - 24 December 17 March - 21 December 3.2 Data analysis and development of categorisation scheme Once all of the briefing transcripts had been collected, they were uploaded to NVivo (QSR International, 2022) for annotation by the first two authors. Our initial coding framework drew on previous work on realisations of directives (Bax, 1986; Ervin-Tripp, 1976; House & Kasper, 1981; Searle, 1979; Weigel & Weigel, 1985). These frameworks are based on dialogic rather than monologic contexts. This meant that we did not find some realisation types in our corpus (e.g. House & Kasper's “Scope-Stating” level in Table 1) and therefore omitted them. We also added other directive types which are rare in dialogic contexts but present in our corpus (e.g., Impersonal constructions; see Table 3 ).Table 3 Coding of directives in the study Table 3: CODE Type of Directive Linguistic Realisations Examples 1 Imperative Command Imperative verb or let(s) + Verb stay at home, protect the NHS and save lives (Johnson, 11 May) 2 Performative Direct Personal Performative Directive [I / we] + Verb + [you / national identity category] + to-infinitive To older people - we are asking you to stay away from your grandkids (Sturgeon, 20 March) 3 Performative Indirect Personal Performative Directive [I / we] + Verb + [people/everyone] + to-infinitive We continue to ask everyone who is on public transport and planes to wear a mask or face covering (Ardern, 24 August) 4 Modal Declarative Direct Modalised statement Declarative modal statement with first-or second-person subjects you should still be spending the majority of your time at home (Sturgeon, 1 June) 5 Modal Declarative Indirect Modalised statement Declarative modal statements with unspecific third-person addressee (including semi-modals need to/have to) People should Stay Alert (Johnson 11 May) Gatherings at home need to be capped at 10 (Ardern, 11 May) 6 Modal Int Modalised question/ Request Interrogatives with modal auxiliaries and first or second person subjects Can you build in contact tracing tools or mechanisms to keep track of your supply train and customers? (Ardern, 9 April) 7 BE Verbed to be Verbed to directive / Prohibition/ Requirement be + Verbed (+ to-infinitive) with verb of obligation or prohibition mass gatherings are still prohibited (Johnson, 3 July) all licensed bars and restaurants will be required to close indoors and outdoors from 6pm this evening (Sturgeon, 9 October) 8 Impersonal direct Impersonal directive / Assessment Impersonal construction (it + is + Adjective + that/to-infinitive), addressed to first or second person, or to national identity categories it is absolutely essential that we guard against future outbreaks (Sturgeon, 14 August) 9 Impersonal indirect Impersonal directive / Assessment Impersonal construction (it + is + Adjective + that/to-infinitive), no obvious addressee. it's vital that everyone exercises the greatest possible personal responsibility (Johnson, 16 December) 10 Nominal Directive with performative Noun Nouns of obligation referring to duties, prohibitions There is now a travel ban on visitors from Denmark into any part of the UK (Sturgeon 9 November) 11 Verbing Directive realized by Verbing (this/that means) Verb+ing indicating action (not) to be performed That means, in particular, not visiting other people's houses (Sturgeon, 13 April) 12 Infinitive Directive realized by infinitive to-infinitive referring to action (not) to be performed And on top of that, to restrict the amount of time a patron is in the premises to no more than 30 minutes and preferably less (Morrison, 24 March) 13 Condition Directive realized by conditional clause Directive as a condition under which an action is permitted using e.g. provided / if / as long as As long as physical distancing between different households is maintained, this can include overnight stays (Sturgeon, 10 July) 14 Noun Phrase Directive realized by nominalization or noun Nominalizations/nouns signalling the desired action and use of isolation and quarantine for those exposed to COVID (Ardern, 15 July) let me just close by reminding everybody again of FACTS - the five rules that we all must follow to stay as safe as possible: Face coverings in enclosed spaces, … (Sturgeon, 10 August) 15 Hint Hint, i.e. indirect directive Indirect instruction not associated with a specific realization It will still be possible for the police to break up large and irresponsible gatherings (Johnson 23, June) the more we do the right thing together as Australians, the more lives we will save (Morrison, 22 March) 16 Incomplete Incomplete Incomplete / abandoned directive or false start we now no longer will be allowing anyone unless they're a citizen or a resident or a direct family member in those cases (Morrison 19 March) In creating the framework and applying it to the annotation process, AUTHOR1 was responsible for coding Johnson and Sturgeon's briefings and AUTHOR2 coded those delivered by Ardern and Morrison. We took several steps to ensure that this annotation would be as reliable as possible. First, the coding framework was co-developed by AUTHOR1 and AUTHOR2, who worked closely to ensure a shared understanding of both the focus and boundaries (inclusion / exclusion criteria) for each directive type. Second, we frequently discussed and reflected on the coding framework, iteratively and collaboratively updating it in response to the directives found in our corpus. Third, AUTHOR1 and AUTHOR2 consulted regularly about any ambiguous or problematic instances in order to clarify interpretations and maximise consistency of coding across the entire corpus. Despite taking these steps, some minor inconsistencies emerged in the process of comparing results for this paper; these inconsistencies chiefly concern judgements about the directness of a small number of statements. An example of this is seen in (2), which appears to be self-directed yet lacks a specific addressee.(1) I think it is important to reiterate New Zealand has no officials in Wuhan, whereas the United States does. (Ardern, 28 January) We also expanded our initial coding framework to account for finer distinctions of directness and forcefulness of directives than are usually included in work on directives. Directness is determined by whether directives are addressed to the TV audience or to (often unspecified) third parties: first- and second-person pronouns and national identity categories (e.g., “Australians”) are counted as direct, while instructions aimed at people in general or lacking obvious addressees are indirect. This distinction does not apply in all categories; Hints, for example, are by definition indirect and have no direct alternative. Forcefulness is determined based on the grammatical and lexical means used to express the directive. For most categories, there are options which can strengthen or weaken the force of the directive. A well-established example is the distinction between you must (more forceful) and you should (weaker). Even within Hints it is possible to distinguish forceful from less forceful utterances; explicit mention of rules or bans and/or the consequences of inaction was seen as marking forcefulness. The distinction between more and less forceful is one which has wide support in the modality literature, although it is less commonly noted in work on speech acts (Sbisà, 2001). As with other categorizations, decisions relating to relative forcefulness of instances arose from constant consultation between team members relating to instances found in the corpus. Our aim in making these distinctions was to better understand the strategies and styles these leaders used in directing their populations to follow, or not to follow, specific behaviours. The use of a more direct, more forceful means of delivering a directive suggests to the audience a higher level of urgency and a greater need to adapt their behaviour accordingly compared to less direct, lower intensity formulations. 3.3 Statistical analysis Given the disparities in corpus sizes across leaders (see Table 2), we transformed the raw frequencies of directive use to a value of n per 1,000 tokens for statistical analysis of the coded data. We then utilised the partykit package in R5 (version 4.2.1, R Core Team, 2022) to create conditional inference trees (CITs) (Tagliamonte & Baayen, 2012). CITs, like other multivariate tree-based methods, recursively partition the data into two sections to maximise prediction accuracy, making them a versatile multivariate method with easily interpretable outputs (Baayen et al., 2013). CITs, unlike traditional classification and regression trees (CARTs), use significance tests to establish whether a particular split is warranted (Gries, 2020). This technique decreases the need for pruning (Hothorn et al., 2006), while variables with more potential splitting points are not artificially preferred (Boulesteix et al., 2015). CITs are thus used in our study to discover correlations between the predictors and the dependent variable. The lm command was used to perform subsequent linear regression analysis in R, in order to inferentially confirm the significance and effect size of the latent constructs underlying the observed variables (Norouzian & Plonsky, 2018). Standardized parameters were obtained by fitting the model on a standardized, scaled version of the dataset. 95% Confidence Intervals (CIs) and p-values were computed using the Wald approximation, which calculates if explanatory variables in a model are significant. To ensure the regression models met required assumptions of normality, the means and SDs of the numeric data (i.e., the normalised frequencies of metadiscourse items) were converted to a normalised dataset through a procedure known as scaling in R. This involves converting each original value into a z-score by dividing the values of each column by the corresponding scale value from the input, thus ensuring the data meet normality criteria. Secondly, a Durbin-Watson test was conducted on each regression analysis to determine the potential for autocorrelation (also called serial correlation) in residuals. Each test statistic was approximately 2.0, with test statistic values in the range of 1.5 to 2.5 considered relatively normal while values under 1 or more than 3 are a cause for concern (Field, 2009). 4 Findings / discussion 4.1 Use of directives across leaders Table 4 shows the combined raw/normalised use of all coded directives for each leader. While Sturgeon has by far the highest average of directives per briefing and Johnson the lowest, Sturgeon's briefings were also the longest and Johnson's the shortest, meaning that the difference is lower when considered in terms of directives per 1,000 tokens.Table 4 Overview of quantitative data on directives for each leader Table 4: Directives Ardern Johnson Morrison Sturgeon Total directives (raw) 602 544 1153 4558 Avg directives per briefing 18.2 14.7 15.1 30 Mean (n per 1,000 tokens) 11.47 11.67 8.27 14.17 Std. Deviation (n per 1,000 tokens) 9.24 5.36 7.36 5.54 Figure 1 shows a scaled comparison of the four PMs in terms of directive use. This makes clear Sturgeon's heavy use of directives and also Morrison's much lower relative use. It has been noted elsewhere (Reyes-Bernard et al., 2021) that Morrison's briefings generally focused on economic recovery, which may go part way to explaining his lower overall use of COVID-related instructions. We can also note that, at least in terms of density of directives, Ardern and Johnson are quite similar, which is an interesting result given the very different perceptions of these leaders’ effectiveness in dealing with COVID-19 and communications with the public; density of directives alone cannot reveal a great deal about communication strategies.Figure 1 Scaled comparison of the overall use of directives by the four leaders Figure 1: A CIT analysis (Figure 2 ) appears to confirm the findings shown in Figure 1 inferentially. This presents a tree-based comparison of speakers’ scaled normalised frequencies of directive use shown in Figure 1. Variance in these scaled frequencies by speaker is represented by critical splits in the scaled data, with the first such split suggesting Sturgeon's directive use is likely to be higher than that of the other speakers, while the next split suggests Morrison's directive use is likely to be lower than that of the other speakers.Figure 2 Conditional Inference Tree of Directive use across speakers (scaled) Figure 2: To confirm these results, a linear regression model was run in R (estimated using OLS, a method for estimating the unknown parameters in a linear regression model) to predict the use of directives (n per 1,000) across the four leaders. The model explains a statistically significant albeit weak proportion of variance (R2 = 0.12, F(3, 294) = 13.97, p < .001, adj. R2 = 0.12). The model's intercept (Sturgeon), is at 0.30 (95% CI [0.15, 0.45], t(294) = 3.99, p < .001). Within this model, the effect of Speaker [Ardern] is statistically significant and negative (beta = -0.39, 95% CI [-0.74, -0.03], t(294) = -2.15, p = 0.032; Std. beta = -0.39, 95% CI [-0.74, -0.03]), the effect of Speaker [Johnson] is statistically significant and negative (beta = -0.36, 95% CI [-0.70, -0.02], t(294) = -2.09, p = 0.038; Std. beta = -0.36, 95% CI [-0.70, -0.02]), and the effect of Speaker [Morrison] is statistically significant and negative (beta = -0.85, 95% CI [-1.11, -0.59], t(294) = -6.43, p < .001; Std. beta = -0.85, 95% CI [-1.11, -0.59]). These results confirm Sturgeon's use of directives in general was heavier in her speeches than those of the other PMs. This finding also seems to confirm that Sturgeon viewed the briefings as a vehicle for communicating COVID-related instructions to the public, which is not always true for the other leaders, who delivered briefings that included no directives (Ardern and Morrison) or very few (Johnson). 4.2 Directives by type A clearer idea of the styles of the leaders can be obtained by examining the types of directives that they used. The normalised frequencies of directives for each type are shown in Figure 3 . As we can see, the types of directive that occur most frequently are Imperatives, followed by Direct Modal Declaratives (MDD), Hints, and Indirect Modal Declaratives. It is interesting to note the predominance of both very direct (Imperatives, MDD) and very indirect (Hints) options within the context of monologic COVID-19 briefings. This tendency towards extremes of directness may be associated with the need by these leaders to make both high- and low-level impositions on their respective populations, leading to an adjustment in their directive choices in response to “the assumed degree of face-threat” involved in specific utterances (Decock & Depraetere, 2018, p. 35). It is important to recall, however, that the meaning and interpersonal effects of (in)directness are highly nuanced and often culturally specific (Haugh, 2015).Figure 3 Distributions of normalised frequencies (per 1000 tokens) for each category and each leader Figure 3: As Figure 3 also indicates, a number of directive categories were very rarely found, including some that were only in the communication of specific leaders, such as Incomplete directives, which are only attributed to Morrison. To determine the statistical likelihood of the use of particular directive types across these leaders, a CIT was conducted that included all 13 directive types coded within our dataset (alpha = 0.0033 for significance). The CIT results shown in Figure 4 suggest that the use of imperatives is key to the variation across the leaders (p<.001). While this finding might seem unsurprising, bearing in mind the close association between imperatives and directives, the material point, shown by Figure 3 and even more clearly in Figure 5 , is that the distribution of imperatives is not even across the four leaders: high use of imperatives is particularly associated with Sturgeon's briefings. In contrast, the use of Hints (associated with high imperative use) is particularly prominent in Ardern's speeches (p=.001). We also found that the use of modal declarative directives with first- and second-person pronouns is a key feature of Johnson's speeches. Of these leaders Morrison has the lowest prevalence of the most direct types of directives and a relatively higher frequency of the most indirect, Hints, providing support for Marsen and Ali-Chand's (2022) observation that he has a tendency for indirectness. We now examine these findings in more detail.Figure 4 CIT for use of directive subcategories across the four leaders (scaled) Figure 4: Figure 5 Scaled comparison of use of Imperatives across the four leaders Figure 5: 4.2.1 Imperatives Figure 5 indicates Sturgeon's preference for imperative forms and the other leaders’ relatively lower use of this type of directive; Johnson and Morrison in particular seem less inclined towards using this direct option. A linear regression model was used to confirm the variance in the CIT for use of imperatives across the leaders, and this explained a statistically significant and substantial proportion of variance (R2 = 0.35, F(3, 294) = 52.97, p < .001, adj. R2 = 0.34). The model's intercept, corresponding to Speaker = Sturgeon, is at 0.56 (95% CI [0.43, 0.69], t(294) = 8.57, p < .001). Within this model, the effect of Speaker [Ardern] is statistically significant and negative (beta = -0.84, 95% CI [-1.15, -0.54], t(294) = -5.42, p < .001; Std. beta = -0.84, 95% CI [-1.15, -0.54]), the effect of Speaker [Johnson] is statistically significant and negative (beta = -1.05, 95% CI [-1.34, -0.76], t(294) = -7.06, p < .001; Std. beta = -1.05, 95% CI [-1.34, -0.76]), while the effect of Speaker [Morrison] is statistically significant and negative (beta = -1.33, 95% CI [-1.55, -1.11], t(294) = -11.69, p < .001; Std. beta = -1.33, 95% CI [-1.55, -1.11]). Sturgeon's heavy use of imperatives compared with the three other PMs is therefore confirmed. This finding reflects the care Sturgeon took with wording instructions directly and in repeating them at key points in her briefings - usually to conclude the message. This communicative strategy emerges particularly clearly after the introduction on 19 June of the FACTS acronym referring to five key instructions which were repeated in every subsequent briefing with very similar wording. Figure 6 below shows the relevant extract from the 19 June briefing with imperatives underlined. Note that both “ordinary imperatives” (avoid, clean) and “let-imperatives” (let me run through) (Huddleston & Pullum, 2002, p. 924) are included in this category; ordinary imperatives are seen as more forceful than let-imperatives. Sturgeon's use of imperatives aligns with recommendations in the literature to keep messages simple and concise (Lunn et al., 2020; McClaughlin et al., 2021). We can also note here that two of the FACTS instructions are not imperatives, but Noun Phrase type directives, where a noun phrase is presented without a finite verb. The use of Noun Phrase directives seems to be a strategy for referring to rules that the audience is expected to know about in a simple and straightforward way, although it is less direct and forceful than the use of imperatives.Figure 6 extract from Sturgeon's 19 June briefing introducing FACTS Figure 6: Another feature of Sturgeon's style seen in Figure 6 which also contributes to the higher concentration of imperatives in her briefings is her fondness for expressions using let-imperatives of the form let me + communicative verb, as if asking for permission to speak (e.g., let me stress, let me be clear, let me begin/end, let me thank). Sturgeon uses this expression on average almost twice in every briefing and almost five times as frequently as Johnson. The use illustrated in Figure 6, which signals what is coming next (Carter & McCarthy, 2006), is the most common one, with just under half of all instances. A similar use of let me is to introduce explanations, as in example 3 where it is used to draw attention to the importance of the following explanation.(1) Let mebe clear, because I know it is a question that has been asked, that doesn't limit you to seeing just one specific household In contrast to the forcefulness and urgency of ordinary imperatives, this use of let-imperatives can be seen as Sturgeon acknowledging and including her audience in the briefings, a characteristic noted by Dada et al. (2021). At the same time, instances involving let me + communicative verb introducing other directives (around 11%), such as examples 4 and 5, suggest a different, more forceful interpretation.(1) let me state very clearly how I expect people to be behaving. People should be staying at home (2) let me stress how important it is that people in those households do self-isolate for the entire period advised Let me in these examples seems to draw attention to the fact that it is the First Minister who is delivering the directive by emphasising her expectations and her evaluation of the importance of following the instructions. This contrast in use of the same directive structure points to the importance of follow-up research into how directives combine in discourse in ways that we have not investigated in this study. 4.2.2 Modal Declaratives Direct Johnson's preference for the use of modal declaratives with first- and second-person pronouns in the briefings was noted in reference to Figure 4. This is more obvious in Figure 7 , which shows that this preference is pronounced compared to the other leaders, although Sturgeon also has a slightly higher use of this type of directive than Ardern and Morrison.Figure 7 Relative preference for modal declaratives with first- and second-person pronouns (scaled) Figure 7: Johnson tends to use this directive type at key moments such as his 23 March announcement of lockdown shown in example 6. We can see three instances of MDD in this extract, each using a different pronoun and all involving the more forceful must. The self-imposed directive I must give, used here to preface the announcement, makes an interesting contrast with Sturgeon's preference for let me give in a similar situation in framing it as an obligation rather than a request for permission to make an announcement. The use of we must do is an example of inclusive/patriotic we - an interpretation that is suggested by the use of the conjunction because indicating that staying at home is integral to stopping the spread. As also noted by McClaughlin et al. (2021), Johnson in general prefers to combine forceful forms with we, and the third instance in example 6 is typical of this feature; it seems easier to use forceful must if one is including oneself amongst those being directed to act. However, the use of stronger forms with second-person as seen here is relatively unusual in our corpus, perhaps reflecting an awareness that must and have to are more forceful and should therefore be used carefully, perhaps only in the most urgent situations.(1) From this evening I must give the British people a very simple instruction - you must stay at home. Because the critical thing we must do is stop the disease spreading between households. (Johnson, 23 March) A second linear regression model was used to confirm the variance in the CIT for use of MDD across the leaders, which explained a statistically significant and weak proportion of variance (R2 = 0.10, F(3, 294) = 11.22, p < .001, adj. R2 = 0.09). The model's intercept, corresponding to Speaker [Johnson] is at 0.66 (95% CI [0.35, 0.96], t(294) = 4.19, p < .001). Within this model, the effect of Speaker [Sturgeon] is statistically significant and negative (beta = -0.57, 95% CI [-0.91, -0.23], t(294) = -3.26, p = 0.001; Std. beta = -0.57, 95% CI [-0.91, -0.23]), the effect of Speaker [Ardern] is statistically significant and negative (beta = -0.90, 95% CI [-1.35, -0.45], t(294) = -3.93, p < .001; Std. beta = -0.90, 95% CI [-1.35, -0.45]) and the effect of Speaker [Morrison] is statistically significant and negative (beta = -1.05, 95% CI [-1.42, -0.67], t(294) = -5.48, p < .001; Std. beta = -1.05, 95% CI [-1.42, -0.67]). This confirms Johnson was statistically more likely to make use of MDD with first- and second-person pronouns than the other leaders, at least when the use of imperatives was relatively low (see also Figure 4). A breakdown of the pronouns used by each leader when they employ MDD directives is seen in Figure 8 . We can see here that Johnson makes more frequent use of first-person pronouns, both singular and plural, than the other leaders, while Sturgeon makes the most frequent use of you; this makes up a far higher proportion of Sturgeon's overall MDD directives than for other leaders. It is also notable how much more frequently Ardern uses we than you, which is in line with previous observations that her communication emphasises social solidarity (McGuire et al., 2021). As for Johnson's preference for using we, this seems to be part of a general strategy in the Downing Street briefings, as it is observed across speakers (Williams & Wright, 2020).Figure 8 Relative frequencies of first/second person pronouns used with MDD directives in briefings of each leader (normed frequencies within the columns) Figure 8: The use of second-person pronouns with MDD directives shown in example 7, which is taken from Sturgeon's briefings but is representative of the pattern for all four leaders. It is interesting that Sturgeon switches from a more indirect formulation (people should not be going out) to directly addressing the audience as she moves into the list of directives.(1) Other than for a few very specific reasons, people should not be going out. You should not be meeting up with people from other households. You should observe social distancing measures when it is essential for you to go out. And if you or other people in your household show signs of COVID 19, remember, a fever or a persistent cough, you should be isolating completely. (Sturgeon, 10 April) We see in example 7 the general preference for weaker, less forceful forms. The choice of should seems to indicate an awareness that formulations such as you must/have to are somehow to be avoided. At the same time, for the audience, using this less forceful item might suggest that compliance is somehow less pressing. 4.2.3 Hints Our third main finding is Ardern's relative preference for the use of Hints compared to the other leaders (see Figure 9 ). Given that she has been widely praised for clear communication, this is surprising, as Hints by definition are inexplicit, relying on the audience to understand what is meant from context (Ervin-Tripp, 1976). This is not entirely in line with Marsen and Ali-Chand's (2022) claim that Morrison is the more indirect in that his use of Hints is less pronounced than Ardern's. However, it should be noted that this is not the only way indirectness is shown.Figure 9 Relative use of Hints across the four leaders (scaled) Figure 9: A linear regression model was used to confirm the variance in the CIT for the use of Hints across PMs, explaining a statistically significant and weak proportion of variance (R2 = 0.10, F(3, 294) = 11.22, p < .001, adj. R2 = 0.09). The model's intercept, corresponding to Speaker = Johnson, is at 0.66 (95% CI [0.35, 0.96], t(294) = 4.19, p < .001). Within this model, the effect of Speaker [Sturgeon] is statistically significant and negative (beta = -0.57, 95% CI [-0.91, -0.23], t(294) = -3.26, p = 0.001; Std. beta = -0.57, 95% CI [-0.91, -0.23]), the effect of Speaker [Ardern] is statistically significant and negative (beta = -0.90, 95% CI [-1.35, -0.45], t(294) = -3.93, p < .001; Std. beta = -0.90, 95% CI [-1.35, -0.45]), and the effect of Speaker [Morrison] is statistically significant and negative (beta = -1.05, 95% CI [-1.42, -0.67], t(294) = -5.48, p < .001; Std. beta = -1.05, 95% CI [-1.42, -0.67]). This confirms that Ardern and Morrison used Hints more often in their briefings than either Johnson or Sturgeon. It is worth noting, however, that Ardern often uses Hints in combination with other more direct and more forceful options to encourage the public to work together in a way that Morrison does not. Examples 8 and 9 demonstrate this strategy. In both examples, Hints encouraging the audience to stick together and to avoid irresponsible behaviour are used to back up imperatives. Ardern shows some skill here in combining direct and forceful directives and then switching to we to emphasise the importance of social solidarity. In example 10 the implication is that the audience should access this advice to overcome the issues of misinformation. Again here we can note the switch of addressee in the directive, in this case to you, which seems to make it more relevant to the audience.(1) Go home tonight and check on your neighbours, start a phone tree with your street, plan how you'll keep in touch with one another. We will get through this together, but only if we stick together. (Ardern, 23 March) (2) so please be vigilant at level 2. Irresponsible behaviour will take us backwards. (Ardern, 5 November) (3) Finally, this is a time when I know people will want as much information as possible. It's also a time when there is plenty of mis-information. All the advice from the government aboutCOVID-19and how it affects you is availableat www.covid19.govt.nz (Ardern, 21 March) In short, Hints appear to be a supplementary resource for Ardern, adding weight to other directives which she issued more directly and forcefully. Conclusion This study is the first to our knowledge to attempt a rigorous analysis of the use of directives in crisis communication in general and more specifically in COVID-19 briefings. We have presented an overall picture of how Ardern, Johnson, Morrison, and Sturgeon used directives through a focus on overall use and on the major types of directive. Collectively, their COVID-19 communication featured frequent use of both the most direct types of directives (Imperatives, MDD) and the most indirect (Hints). We have also uncovered key differences between these leaders: Sturgeon relied on Imperatives more heavily than did the other leaders; Johnson made more frequent use of MDDs to deliver instructions; and both Ardern and Morrison often used Hints but, unlike Morrison, Ardern did so alongside Imperatives. Morrison is revealed to be the least enthusiastic user of directives, particularly of more direct and forceful types. Analysis of this variation provides some evidence to support previous claims that Ardern and Sturgeon were relatively effective compared to their male counterparts, in particular Morrison. The finding that Imperatives and MDDs are the most prevalent is expected based on previous research. The prevalence of Hints, meanwhile, is less expected based on the need for clarity and urgency in briefings under these circumstances, although in the hands of an effective communicator like Ardern, they seem to have a place. Hints can be an effective means of reducing face-threat, but in contexts where much is riding on people understanding and doing what needs to be done, overreliance on Hints alone may provide insufficient clarity to promote compliance. On the basis of our findings and previous literature on crisis communication, clarity appears to be of paramount importance in crisis situations, particularly when supported by empathy and social solidarity. In analogous situations, therefore, leaders might consider making judicious use of Imperatives to mark directness and urgency, combined with other formulations that suggest social solidarity, such as those involving inclusive we (we must/should), and Hints might be best reserved for supporting and summarising statements. The discussion of our findings has suggested that the combination of different types of directives is a feature of effective crisis communication and that directives are not evenly distributed across the briefings but tend to be used together in strategically important places. However, we have not been able to investigate these aspects of directive use in depth in this paper. Another area that we have not explored but that seems of importance is the investigation of choice of directive types in relation to the level of imposition. For example, it may be easier to require one's audience to wash their hands than to stay at home, and it seems likely that this would be reflected in the formulation of the relevant directives. It would also be interesting to expand the investigation to a wider range of leaders, including the use of directives in different languages. We hope that research that explores questions of this sort will find the framework of directives developed for this study useful. Our framework is informed by earlier studies but has been expanded to account for directives in monological political briefings. While this analysis focuses on the most salient parameters for our corpus, the framework is generic and thus has the potential to be applied to other data sets. It may also have relevance to contexts such as education and healthcare, in which people are routinely instructed to act in particular ways for their own benefit. Understanding the full range of options for how to communicate those instructions can promote critical self-reflection on the part of educators and healthcare practitioners, which may in turn improve the uptake of their recommendations. The new and expanded typology of directives generated through this research should also prove helpful to linguistic researchers exploring directive usage, particularly in high-stakes monological contexts. Author vitae Benet Vincent is Assistant Professor in Applied Linguistics at Coventry University. He is interested in a range of practical applications of corpus linguistics including in English for Academic Purposes, Translation and Pragmatics. Kate Power is a Lecturer at The University of Queensland. She uses critical discourse analysis to explore various social texts, including news reporting on women, academic writing, religious talk, and COVID-19 crisis communication. Peter Crosthwaite is Senior Lecturer in Applied Linguistics at the School of Languages and Cultures, University of Queensland, Australia. His research and supervision explore the applied uses of corpus linguistics for learning about language through language across numerous domains. Sheena Gardner is Professor of Applied Linguistics at Coventry University. Her teaching, research and supervision centre on academic genres and registers in the BAWE corpus of university student writing and the political discourses of Covid-19 briefings in the UK. Uncited References Partington and Taylor, 2018 Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. 1 See McClaughlin et al. (2023) in this issue for an interesting discussion of the impact of strategies such as these. 2 We would class instances of these speech acts that realise directives as Hints; a good example from Marsen & Ali-Chand (2022, p. 9) is the utterance ‘If people follow the government's directives, we will lock this virus down’ which their categorisation labels a promise (‘commissive’) but which is additionally an indirect directive asking the public to follow the directives 3 Marsen & Ali-Chand's term is “‘must’ and ‘need’ modals”; we have used here Biber et al.’s (1999) term as it has wider use. The modals in question are must, should, need (to), ought and have to 4 These seem to be what House & Kasper (1981) refer to as “performatives”, e.g. I ask you to close the window (see category 7 in Table 1). 5 The R notebook for our analysis can be downloaded here: https://tinyurl.com/4225ymy ==== Refs References Ädel, A. (2010). How to use corpus linguistics in the study of political discourse. In A. O'Keeffe & M. McCarthy (Eds), The RoutledgeHandbook of Corpus Linguistics (pp.591–604). Routledge. Baayen H.R. Janda L. Nesset T. Endresen A. Makarova A. Making choices in Russian: Pros and cons of statistical methods for rival forms Russian Linguistics 37 3 2013 253 291 Bax I.P. How to assign work in an office: A comparison of spoken and written directives in American English Journal of Pragmatics 10 6 1986 673 692 Boulesteix A.-L. Janitza S. Hapfelmeier A. Van Steen K. Strobl C. Letter to the editor: On the term ‘interaction’ and related phrases in the literature on random forests Briefings in Bioinformatics 16 2 2015 338 345 24723569 Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language usage. Cambridge University Press. Carter, R. & McCarthy, M. (2006). Cambridge Grammar of English. Cambridge University Press. Curl T.S. Drew P. Contingency and action: A comparison of two forms of requesting Research on Language and Social Interaction 41 2 2008 129 153 Dada S. Ashworth H.C. Bewa M.J. Dhatt R. Words matter: political and gender analysis of speeches made by heads of government during the COVID-19 pandemic BMJ Global Health 6 1 2021 p.e003910 10.1136/bmjgh-2020-003910 Davey M. March 24). Australia is crying out for clearer messaging on coronavirus, 'rambling' politicians told The Guardian 2020 https://www.theguardian.com/australia-news/2020/mar/24/australia-is-crying-out-for-clearer-messaging-on-coronavirus-rambling-politicians-told Accessed March 17, 2023 Decock S Depraetere I. (In)directness and complaints: A reassessment Journal of Pragmatics 132 2018 33 46 10.1016/j.pragma.2018.04.010 Ervin-Tripp S. Is Sybil there? The structure of some American English directives Language in Society 5 1 1976 25 66 10.1017/S0047404500006849 Field, A. (2009). Discovering Statistics using SPSS: Third Edition. Sage Publications. Finset A. Bosworth H. Butow P. Gulbrandsen P. Hulsman R.L. Pieterse A.H. van Weert J. Effective health communication–a key factor in fighting the COVID-19 pandemic Patient Education and Counseling 103 5 2020 873 32336348 Goodwin M.H. Participation, affect, and trajectory in family directive/response sequences Text & Talk 26 4-5 2006 515 543 Gries S.Th On classification trees and random forests in corpus linguistics: Some words of caution and suggestions for improvement Corpus Linguistics and Linguistic Theory 16 3 2020 617 647 Halliday, M.A.K. & Matthiessen, C.M.I.M. (2014). Halliday's Introduction to Functional Grammar. (4th ed.). Routledge. Hansson, S. (2017). Government communication. In R. Wodak & B. Forchtner (Eds.), The Routledge Handbook of Language and Politics (pp. 326-341). Routledge. Haugh, M. (2015). Im/politeness implicatures. De Gruyter. Hothorn T. Hornik K. Zeileis A. Unbiased recursive partitioning: A conditional inference framework Journal of Computational and Graphical Statistics 15 3 2006 651 674 House, J., & Kasper, G. (1981). Politeness markers in English and German. In Volume 2 Conversational Routine (pp. 157-186). De Gruyter Mouton. Huddleston, R., & Pullum, G. (2002). The Cambridge grammar of the English language. Cambridge University Press. Jaspal R. Nerlich B. Social representations, identity threat, and coping amid COVID-19 Psychological Trauma: Theory, Research, Practice, and Policy 12 S1 2020 S249 32463288 Jones, R. (2021). Order out of chaos: Coronavirus communication and the construction of competence. In R. Jones (Ed.), Viral Discourse (pp. 69-78). Cambridge University Press. Leng, A., Lemahieu, H., & Smith, B. (2021). COVID Performance Index. Lowy Institute. https://interactives.lowyinstitute.org/features/covid-performance/#region Accessed May 18, 2022 Lunn P. Belton C. Lavin C. McGowan F. Timmons S. Robertson D. Using behavioural science to help fight the coronavirus (No. 656) ESRI working paper 2020 McClaughlin E. Vilar-Lluch S. Parnell T. Knight D. Nichele E. Adolphs S. Clos J. Schiazza G. The reception of public health messages during the COVID-19 pandemic Applied Corpus Linguistics 3 1 2023 10.1016/j.acorp.2022.100037 McClaughlin, E., Nichele, E., Adolphs, S., Barnard, P., Clos, J., Knight, D., McAuley, D. & Lang, A. (2021). Public health messaging by political leaders: a corpus linguistic analysis of COVID-19 speeches delivered by Boris Johnson. Nottingham: UKRI/AHRC. https://doi.org/10.17639/3fgb-fn44 McGuire D. Cunningham J.E. Reynolds K. Matthews-Smith G. Beating the virus: an examination of the crisis communication approach taken by New Zealand Prime Minister Jacinda Ardern during the COVID-19 pandemic Human Resource Development International 23 4 2020 361 379 Marsen S. Ali-Chand Z. We all have a role to play’: A comparative analysis of political speech acts on the COVID-19 crisis in the South Pacific Communication Research and Practice 8 1 2022 19 35 10.1080/22041451.2022.2040701 Menon P. Ardern's online messages keep spirits up in New Zealand's coronavirus lockdown Reuters 2020 https://www.reuters.com/article/us-health-coronavirus-newzealand/arderns-online-messages-keep-spirits-up-in-new-zealands-coronavirus-lockdown-idUSKBN21H0HN Nielsen, R.K., Fletcher, R., Kalogeropoulos, A. & Simon, F. (2020). Communications in the coronavirus crisis: lessons for the second wave. Reuters Institute for the Study of Journalism. Norouzian, R., & Plonsky, L. (2018). Correlation and simple linear regression in applied linguistics. In A.E.K. Phakiti, P. De Costa, L. Plonsky & S. Starfield (Eds.), The Palgrave Handbook of Applied Linguistics Research Methodology (pp. 395-421). Palgrave Macmillan. Oliver D. COVID-19 highlights the need for effective government communications British Medical Journal 369 2020 m1863 10.1136/bmj.m1863 32404354 Partington, A. & Taylor, C. (2018). The Language of Persuasion in Politics. Routledge. QSR International Pty Ltd. (2020). NVivo (released in January 2022), https://www.qsrinternational.com/nvivo-qualitative-data-analysis-software/home R Core Team R: A language and environment for statistical computing 2022 R Foundation for Statistical Computing Vienna, Austria https://www.R-project.org/ Reyes Bernard N. Basit A. Sofija E. Phung H. Lee J. Rutherford S. Sebar B. Harris N. Phung D. Wiseman N Analysis of crisis communication by the Prime Minister of Australia during the COVID-19 pandemic International Journal of Disaster Risk Reduction 62 2021 p.102375 Sanders K. British government communication during the 2020 COVID-19 pandemic: learning from high reliability organizations Church, Communication and Culture 5 3 2020 356 377 10.1080/23753234.2020.1824582 Sbisà M. Illocutionary force and degrees of strength in language use Journal of Pragmatics 33 12 2001 1791 1814 Searle, J.R. (1979). Expression and meaning, Cambridge University Press Sodha S. Media experts despair at Boris Johnson's coronavirus campaign The Guardian 2020 https://www.theguardian.com/commentisfree/2020/mar/26/media-experts-despair-boris-johnson-coronavirus-campaign Tagliamonte S. Baayen R.H. Models, forests, and trees of York English: Was/were variation as a case study for statistical practice Language Variation and Change 24 2 2012 135 178 UK Government Communication Service (2021). A Guide for New Starters. https://gcs.civilservice.gov.uk/publications/a-guide-for-new-starters/ Accessed December 15, 2021 Vine B. Directives at work: Exploring the contextual complexity of workplace directives Journal of Pragmatics 41 2009 1395 1405 Wales, K. (1996). Personal Pronouns in Present-day English. Cambridge University Press. Weigel M.M. Weigel R.M. Directive use in a migrant agricultural community: A test of Ervin-Tripp's hypotheses Language in Society 14 1 1985 63 79 Williams, J., & Wright, D. (2022). Ambiguity, responsibility and political action in the UK daily COVID-19 briefings. Critical Discourse Studies. DOI: 10.1080/17405904.2022.2110132 Williams J Wright D Pronominal ambiguity and ascriptions of responsibility in the UK daily coronavirus briefings 2020 [Conference session]. 46th Conference of the Japan Association for English Corpus Studies (JAECS), 3-4 October 2020 Wolf, M. (2011). Health literacy. In B. Fischhoff, N. Brewer & J. Downs (Eds.), Communicating Risks and Benefits: An evidence-based user's guide (pp. 77-88). US Department of Health and Human Services.
PMC010xxxxxx/PMC10286559.txt	==== Front CJC Open CJC Open CJC Open 2589-790X Published by Elsevier Inc. on behalf of the Canadian Cardiovascular Society. S2589-790X(23)00147-6 10.1016/j.cjco.2023.06.004 Original Article Trends in major adverse cardiac events and healthcare utilization during the COVID-19 pandemic in Alberta, Canada Mackinnon Erin S. PhD 1 Anderson Todd MD 2 Raggi Paulo MD 3 Gregoire Jean MD 4 Wani Rajvi J. PhD 1 Packalen Millicent S. 1 Graves Erin MSc 5 Ekwaru Paul PhD 5 McMullen Suzanne MHA 5# Goodman Shaun G. MD 6 1 Amgen Canada Inc., Mississauga, Ontario, Canada 2 University of Calgary, Calgary, Alberta, Canada 3 University of Alberta, Edmonton, Alberta, Canada 4 Montreal Heart Institute, Montreal, Quebec, Canada 5 Medlior Health Outcomes Research Ltd., Calgary, Alberta, Canada 6 St. Michaels Hospital, Toronto, Ontario, and Canadian VIGOUR Centre, University of Alberta, Edmonton, Alberta, Canada # Corresponding Author: Suzanne McMullen, Suite 300 - 160 Quarry Park Blvd SE, Calgary, AB, T2C 3G3, Phone: 1.403.612.0086 22 6 2023 22 6 2023 22 2 2023 31 5 2023 19 6 2023 © 2023 Published by Elsevier Inc. on behalf of the Canadian Cardiovascular Society. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Atherosclerotic cardiovascular disease (ASCVD) is a leading cause of morbidity and mortality in Canada. The COVID-19 pandemic altered the usual care of ambulatory and acute cardiac patients. This study aimed to describe ASCVD-related clinical outcomes and healthcare resource utilization (HCRU) patterns during the coronavirus disease 2019 (COVID-19) pandemic in Alberta, Canada, relative to the three preceding years. Methods A repeated cross-sectional study design was conducted over three-month periods using administrative health data between March 15, 2017, and March 14, 2021. ASCVD-related clinical outcomes included major adverse cardiovascular events (MACE) endpoints. HCRU was assessed through general practitioner and other healthcare professional visits (including telehealth claims) for ASCVD events, emergency department visits, ASCVD diagnostic imaging tests, laboratory tests, and hospital length of stay. Results Relative to the control year period (March to June 2019) ASCVD-related events (i.e., hospitalizations, emergency department (ED) visits and physician office visits) declined by 23% during the three-months COVID-19 period (March to June 2020). Acute declines were not sustained following June 2020. In contrast, in-patient mortality rates involving a primary MACE endpoint increased from March to June 2020 COVID-19 period. Conclusions This study demonstrates the COVID-19 pandemic and corresponding public health restrictions impacted ASCVD-related care. While many clinical outcomes returned to pre-pandemic levels at the end of the observation period, our results suggest that patients’ HCRU declined, which could lead to further CV events and mortality. Understanding the impact of COVID-19 restrictions on ASCVD-related care may help improve healthcare resiliency. ==== Body pmcINTRODUCTION Atherosclerotic cardiovascular disease (ASCVD) is a leading cause of morbidity and mortality in Canada and worldwide.(1, 2, 3, 4) In the Canadian context, 37% and 53% of individuals age 20 years and older are estimated to have poor or intermediate cardiovascular (CV) health, respectively.(5) In addition to a considerable impact on patient health, the global cost of ASCVD has been estimated at USD 863 billion (2010), posing an enormous economic burden to health systems worldwide.(4) In the province of Alberta, Canada, the five-year prevalence of ASCVD was recently estimated at 8.99% (or 89.9 per 1,000 persons).(6) Moreover, between 2004 and 2013, acute myocardial infarction (AMI) costs the healthcare system in Alberta $1.03 billion, with each hospitalization costing an estimated $12,000.(7) The global epidemic of coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), was declared a pandemic by the World Health Organization (WHO) on March 11th, 2020.(8, 9, 10, 11) The COVID-19 pandemic has had a devastating impact on healthcare systems around the world. As of August 12, 2022, there have been 585,950,085 confirmed cases and 6,425,422 deaths reported to the WHO.(12) In Alberta, the first state of public health emergency due to COVID-19 was declared under the provincial Public Health Act on March 15, 2020.(13) Subsequent responses included public health restrictions such as stay-at-home orders and a shift in healthcare resource priorities.(14) Reopening from this first lockdown began on May 25, 2020. On December 8th 2020 strict lockdowns were again implemented with staggered reopening in January 2021.(13) During the height of the pandemic, elective surgeries were postponed, and non-urgent in-person healthcare visits were highly discouraged. Although these measures were critical for reducing the transmission of COVID-19, they created significant and unprecedented challenges for managing chronic medical conditions such as ASCVD.(15, 16) Emerging evidence is indicative of this shift in healthcare priorities and the reallocation of resources across Canada. According to the Canadian Institute for Health Information (CIHI), from March 2020 to June 2021, there were approximately 9,300 fewer emergency department (ED) visits per day, 560,000 fewer surgeries performed, and 14,000 additional respiratory admissions in intensive care units (ICUs), compared to the pre-pandemic period (January to December 2019).(14) Patients with CV disease are at higher risk of requiring ICU and developing severe COVID-19 symptoms such as acute respiratory distress syndrome, arrhythmia, or shock.(17, 18) Due to the COVID-19 pandemic, the usual care of both ambulatory and acute cardiac patients was reshaped to accommodate cancelled elective procedures which reduced the efficiency of existing urgent care pathways.(9) In April 2020, hospital case volumes for adult cardiac surgery declined by 51% -61% of pre-COVID levels across Canada.(19) Furthermore, several studies have suggested that patients endured serious cardiovascular events at home, such as acute coronary syndrome (ACS), due to public health messages to avoid the ED for non-COVID-19-related conditions, limited access to emergency medical services, and to avoid the healthcare system for fear of contracting COVID-19. There is limited literature looking into the impact of COVID-19 and lockdown measures on ASCVD-related clinical outcomes and healthcare resource utilization (HCRU) in Canada. An understanding of this impact is an important first step to improving healthcare resiliency in Canada, managing downstream consequences of gaps in HCRU during the pandemic, and planning for future continuity of care. The objectives of this study were to assess ASCVD-related clinical outcomes and changes to HCRU during the COVID-19 pandemic in Alberta, Canada, relative to the three years before the pandemic. MATERIALS AND METHODS Study Design This study used a repeated cross-sectional design of three-month periods of population-level administrative health data from the entire province of Alberta, Canada, between March 15, 2017, and March 14, 2021 (Figure 1 ).Figure 1 Study Design. Abbreviations: COVID-19: coronavirus-19. Data Sources The administrative health datasets used in this study are outlined in Supplemental Table S1. Research ethics board approval was obtained from the Health Research Ethics Board of Alberta – Community Health Committee. The authors confirm that patient consent is not applicable to this article. A waiver of consent was granted due to the population-based and retrospective nature of the study. Missing values were not expected as the province mandates reporting for administrative purposes. Study Population The study population includes all Alberta residents aged ≥18 years presenting to a physician’s office, hospital, or ED with an ASCVD-related event/procedure during one or more cross-sectional periods. Patients accessing care for ASCVD-related outcomes were identified based on the presence of the following: International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic codes; International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Canada (ICD-10-CA) diagnostic codes; and Canadian Classification of Health Interventions (CCI) intervention codes (Supplemental Table S2). These codes were extracted from the Discharge Abstract Database (DAD), National Ambulatory Care Reporting System (NACRS), or Practitioner Claims datasets. Outcomes of Interest Outcomes of interest were the number of ED visits or hospitalizations for any ASCVD event (in any position in the database), as well as primary and secondary composite major adverse cardiovascular event (MACE) endpoints, and the individual MACE endpoints, and the number of deaths on arrival to ED, in ED, and hospital after presenting with a primary MACE endpoint or ASCVD event. To qualify as an ASCVD event an individual required one diagnostic code for CV death, AMI, IS, cerebrovascular disease, transient ischemic attack, coronary atherosclerosis/old myocardial infarction, peripheral arterial disease, angina, or coronary revascularization (percutaneous coronary intervention (PCI), coronary artery bypass graft surgery (CABG)), or acute abdominal aneurysm. Primary and secondary MACE endpoints were defined as both a 5- and 3-point composite respectively, with primary MACE endpoint including CV death, AMI, Ischemic stroke (IS), hospitalization for unstable angina or coronary revascularization, and secondary MACE endpoint including CV death, AMI, or IS. This study implemented a 30-day rule stating that for each specific event (excluding coronary revascularization procedures), multiple occurrences of a diagnostic code for the same event type occurring within 30 days will be counted as a single event, indexed on the first occurrence of the diagnosis code. To define HCRU among patients presenting with ASCVD-related events, we assessed practitioner visits for any ASCVD event, telehealth vs. in-person visits, ASCVD diagnostic imaging tests (of note, diagnostic catheterization procedures was not available for analysis), laboratory tests, number of ED visits and hospitalizations for ASCVD-related outcomes, and median hospital length of stay (LoS, in days) for any ASCVD event and primary MACE endpoints. Physician claims visits were stratified by general practitioners [GP] and other healthcare professionals [HCP] (which included non-cardiac related specialties), based on the physicians’ registered specialty in the physician claims billing data. Billing codes for in-person and telehealth (new codes introduced in March 2020 nationwide) physician visits were captured within the administrative data (Supplemental Table S3). Low-density lipoprotein- cholesterol (LDL-C) and troponin test volumes were captured at the population level, not specific to those with ASCVD, from the Alberta Precision Laboratory (APL) dataset. CV-related diagnostic imaging tests, including computerized tomography (CT), magnetic resonance imaging (MRI), echocardiogram, nuclear studies, and stress tests, were identified using the CCI intervention codes from the DAD and NACRS datasets. The Population Registry dataset was used to restrict the HCRU outcomes to individuals ≥18 years of age. Data Analysis Outcomes were assessed in three-month periods, to capture seasonal variation in ASCVD-related care patterns and the varying public health measures during the first 12 months of the COVID-19 pandemic in Alberta (March 15, 2020, to March 14, 2021; Figure 1). In this study, three-month periods were defined as, Period 1 (March 15 – June 14), Period 2 (June 15 – September 14), Period 3 (September 15 – December 14), and Period 4 (December 15 – March 14) for the following years: 2017/18, 2018/19, 2019/20, and 2020/21. In the first three-month COVID-19 period (March 15, 2020, to June 14, 2020), the Public State of Emergency was declared in Alberta, Canada resulting in temporary healthcare restrictions defined as limited laboratory services, closure of community-based imaging services, cancellation/postponement of elective surgeries, and the introduction of telehealth visits (and associated ICD codes; Supplemental Table S3). Descriptive statistics were calculated to summarize the frequency of ASCVD diagnostic codes and resource use and compare the percent change in outcome during the four three-month COVID-19 periods (2020/21) to the weighted average of the four three-month control periods (2019/20). Comparisons were only made to the 2019/20 three-month control periods because many of the outcomes of interest demonstrated a declining trend over the three-year control period. Clinical outcomes were also calculated as rates per 100,000 population.(20) In addition, outcomes were depicted in line graphs to show increasing or decreasing trends over the study period. A best fit line based on only data from the control years was included in the graphs to depict deviations from the trend line during COVID-19 period. Results with counts <5 were not reported to align with Alberta Health privacy standards, to ensure the deidentification of patients included in the analysis. All analyses were conducted in SAS® version 9.4. Figures were produced using Tableau online version 2021.3. The research reported has adhered to the STrengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines. RESULTS Patient outcomes relevant to ASCVD The rate (per 100,000 population) of any documented ASCVD event remained stable throughout most of the study period. However, during Period 1 of the COVID-19 pandemic, an acute decline was observed (Figure 2 ). When comparing the Period 1 of the COVID-19 pandemic to Period 1 of the control year 2019/20, there were 23% fewer visits for ASCVD-related events. The acute decline in rate was not sustained through the remaining COVID-19 periods. Similar trends were seen for rates of ED visits or hospitalizations involving a primary or secondary MACE endpoints outcomes (Supplemental Figure S1). An analysis of outcome-specific rates for ED visits or hospitalizations for the 5 MACE endpoints (CV death, AMI, IS, unstable angina (UA), and coronary revascularization procedure) showed that the acute declines observed in the MACE composite ED visits or hospitalizations rates during Period 1 of the COVID-19 pandemic were driven by decreases in the number of patients presenting for all individual MACE endpoints, with the greatest decreases observed for coronary revascularization procedures and CV death (Supplemental Figure S2). An interesting trend was observed when stratifying by type of coronary revascularization procedures. The rate of ED visits of hospitalizations for elective (scheduled) procedures was stable throughout the control and COVID periods, whereas urgent procedures revealed a declining trend in the control period with steeper declines starting in P4 of the 2019/20 control year (Figure 3 ). The rates of ED visits or hospitalizations for AMI and IS remained relatively stable throughout the study period, while rates for UA displayed a small but steady decline (Supplemental Figure S2). The total number of ASCVD events occurring during an ED visit or hospitalization (Figure 4 ) followed a declining trend throughout the study period. Like the overall rate of visits for ASCVD events, there was an acute decline in the total number of ASCVD events occurring during an ED visit or hospitalization during Period 1 of the COVID-19 pandemic. More specifically, when comparing Period 1 of the control year 2019/20 to Period 1 of the COVID-19 pandemic, there was a 22% decrease in the number of ASCVD events recorded during an ED visit or hospitalization; the projected trend returned for the remainder of the COVID-19 periods. In the first period of COVID, the number of patients presenting with ASCVD events decreased and the number of CV deaths decreased (Supplemental Figure S2), however, the proportion of patients with in-hospital primary MACE endpoints that resulted in death was higher compared to previous rates (Figure 5 ). In the subsequent periods of COVID, the number of patients presenting with ASCVD events resumed at anticipated volumes but the number of CV deaths was higher than anticipated and the proportion of primary MACE hospitalizations that resulted in death spiked (Figure 5).Figure 2 Rate of any ASCVD event (per 100,000 population) in Alberta, Canada from 2017-2021. Abbreviations: COVID-19: coronavirus-19; P: period. Notes: The best fit line only considers control-years data and predicts trends for the COVID-19 period. Rate is based upon yearly population data from Statistics Canada (Table 17-10-0005-01 Population (18+ years) estimates on July 1st, by age and sex). Figure 3 Rate of ED visits or hospitalizations for coronary revascularization procedures (per 100,000 population) stratified by urgent (emergent) and elective (scheduled) procedures in Alberta, Canada from 2017-2021. Abbreviations: COVID-19: coronavirus-19; P: period. Notes: The best fit line only considers control years’ data and predicts trends for the COVID-19 period. Rate is based upon yearly population data from: Statistics Canada (Table 17-10-0005-01 Population (18+ years) estimates on July 1st, by age and sex). Figure 4 ASCVD events coded at ED visits or hospitalizations in Alberta, Canada from 2017-2021. Abbreviations: COVID-19: coronavirus-19; P: period. Notes: The best fit line only considers control years’ data and predicts trends for the COVID-19 period. Figure 5 Proportion of patients with in-hospital primary MACE endpoints that resulted in death in Alberta, Canada between 2017 and 2021. Abbreviations: COVID-19: coronavirus-19; P: period. Notes: The best fit line only considers control years’ data and predicts trends for the COVID-19 period. Rate is based upon quarterly MACE data. HCRU in patients with ASCVD Reductions in HCRU for physician visits, diagnostic imaging, and laboratory tests were observed during Period 1 of the COVID-19 pandemic, relative to Period 1 of all three control years. The total number of GP and other HCP visits for any ASCVD condition followed a pattern of seasonal variability throughout the control period. Both GP and other HCP visits exhibited an acute decline during Period 1 of the COVID-19 pandemic, later returning to seasonal trends (Supplemental Figure S3). Comparing Period 1 of the control year 2019/20 to Period 1 of the COVID-19 pandemic, there were 13% and 27% fewer ASCVD-related GP and other HCP visits, respectively, during the COVID-19 period. Telehealth visits accounted for 38% and 35% of all GP and other HCP practitioner visits, respectively, in Period 1 of the COVID-19 pandemic. Telehealth visits decreased by 14% in Period 2 of the COVID-19 pandemic; however, the distributions of GP and other HCP visits remained similar, with a greater proportion of telehealth visits during the COVID-19 period being conducted by GPs versus other HCPs. There was a slight decrease in the median hospital LOS for any ASCVD event and any MACE endpoints (both 17%) comparing Period 1 of the control year 2019/20 to Period 1 of the COVID-19 pandemic (Supplemental Figure S4). Diagnostic imaging volumes declined throughout the study period; however, a sharp decrease was observed for patients during Period 1 of the COVID-19 pandemic due to a large reduction in the number of stress tests (Figure 6 ). Between Period 1 of the COVID-19 pandemic and Period 1 of the control year 2019/20, there were 83% fewer diagnostic imaging tests performed. This acute decline was not sustained throughout the remaining three-month COVID-19 periods but remained below the numbers reported during the control periods. Diagnostic imaging volumes stratified by test type are displayed in Supplemental Figure S5. Similar trends were seen for laboratory testing, including LDL-C and troponin testing (data not shown).Figure 6 Number of patients with inpatient or ambulatory diagnostic imaging tests related to ASCVD in Alberta, Canada between 2017 and 2021. Abbreviations: COVID-19: coronavirus-19; P: period. Notes: The best fit line only considers control years’ data and predicts trends for the COVID-19 period. Includes diagnostic imaging tests recorded in the DAD or NACRS databases only. DISCUSSION While the rate of patients seeking medical care for ASCVD-related events remained stable in the three years before the COVID-19 pandemic, this study showed notable reductions in both the number and rates of ASCVD-related hospitalizations and HCP visits during the first three-month period (March to June 2020) of the pandemic, when public health lockdowns were first implemented. Similarly, a previous study conducted in Alberta observed reductions in hospitalizations for any medical or surgical cause.(21) Our study suggests that the COVID-19 pandemic and associated public health measures adversely impacted clinical management of ASCVD patients given the increase in the proportion of in-hospital primary MACE that resulted in death as well as reductions in HCRU in terms of physician visits, median length of hospital stays, diagnostic imaging, and laboratory test volumes. One could hypothesize that the increase in the proportion of MACE resulting in death may be due to a change in the severity of patients presenting to the hospital, with milder patients not presenting (i.e., a smaller denominator). In the subsequent periods after the initial lockdown measures were lifted, the proportion remained higher than the predicted trendline, while the number of ASCVD events was in line with the trendline. Patients may have delayed care-seeking behavior or had interruptions to routine care that altered the severity of the ASCVD events presented and thus the outcomes; however, this hypothesis cannot be proven as the severity of cases can not be verified based on administrative data. The acute decline in visits for composite MACE endpoints observed in this study was in part due to reduced coronary revascularization procedures. While this trend coincided with both global and local studies reporting declines due to cancellation of elective cardiovascular procedures,(9, 22) this study observed a decline in the urgent coronary revascularization procedures. In Alberta, elective procedures were not slowed down during lockdowns as an administrative decision. An Ontario-based study observed significant decreases in the number of referrals and procedures completed for cardiac patients during the pandemic period.(22) The decreases in the Ontario study were associated with an increase in all-cause mortality while awaiting appropriate care. An increase in all-cause mortality was also seen during the COVID-19 pandemic period among Alberta’s population.(23) Evidently in maximizing the hospital sector’s capacity and resources to manage the pandemic, surgical procedures and diagnostic imaging were disrupted, possibly compromising timely access to urgent care pathways and overall clinical outcomes.(14) The current study reveals a declining trend in the total number of ASCVD events occurring during ED visits and hospitalizations throughout the control years. While a low number of patients experiencing an ASCVD-related ED visit or hospitalization was observed during the March to June 2020 period of the COVID-19 pandemic, numbers appear to rebound back to the previous declining trend once lockdown measures were removed in June to September 2020. This pattern supports previous evidence suggesting an association between an increase in statin and antihypertensive drug use and a reduction in ASCVD-related events over time, a measure of good preventative care improving over time.(24) As anticipated, large reductions were observed in the number of physician visits during the first three months of the COVID-19 pandemic period. This reduction aligns with the changes in health care measures implemented in Alberta in March 2020.(25, 26, 27, 28) In addition to the healthcare restrictions, these reductions may also be due to lack of personal protective equipment (PPE) measures in place to enable in-person visits, hesitancy for patients to visit healthcare clinics, and the resulting implementation of telehealth visits. Laboratory and diagnostic test utilization also saw major reductions during the March to June 2020 COVID-19 period. Reductions can be attributed to public health guidance recommending cessation of routine testing for stable community patients.(28) The reduction in HCRU observed in this study may be due in part to patients avoiding interaction with the health system in response to the stay-at-home orders, as seen in other regions.(29) Delaying care may also have contributed to the observed increase in the proportion of inpatient hospital deaths involving primary MACE, as less severe patients stay home and the delay in seeking treatment results in worse outcomes. The reduction in HCRU measures observed in this study aligns with reports from other regions. A survey of healthcare providers from 108 countries revealed numerous institutions/clinics performed significantly less cardiac diagnostic procedures.(30) While many modalities of healthcare delivery decreased, a virtual platform (Telehealth) was introduced as a billing option in Alberta in March 2020, creating this option for physician visits,(25) resulting in a shift in care in patient management. In the March to June 2020 three months COVID-19 period, our study revealed that telehealth accounted for 38% of general practitioner and 35% of other healthcare professional visits. The use of telemedicine may vary by physician and patients based on physician and patient comfort with technology, patient needs, infrastructure, accessibility, and institutional mandates.(31) The impact of COVID-19 observed in the current study were largely due to the public health restrictions implemented across the province. Throughout 2020, hospital bed occupancy in Alberta was under capacity,(32) with the first major wave of cases occurring in the province in December 2020, reaching just under 2,000 daily reported cases.(33) Future research could extend this study to assess the impact of COVID during subsequent waves, including the largest outbreak in early 2022, peaking at close to 7,000 daily report cases.(33) Hospitals in urban centers neared capacity in 2021 and this may have resulted in further impacts to patient care.(32) Additionally, a retrospective longitudinal study design could assess the effects of health system disruptions due to the COVID-19 pandemic on long-term patient outcomes.(34, 35) Along with future research directions, there are strengths to this study that should be noted. To our knowledge, this is the first population-based Canadian study to evaluate the impact of the early COVID-19 pandemic lockdown on ASCVD-related clinical outcomes and HCRU within an Alberta population. We used comprehensive population-based datasets, to provide a thorough, descriptive analysis of the ASCVD population whose care may have been impacted by the first COVID lockdown. The repeated cross-sectional design three-year control period allowed for the capture of trends in the outcomes of interest, overall variability, as well as seasonal fluctuations in ASCVD event rates. Although this study provided a unique opportunity to examine ASCVD-related clinical outcomes and HCRU before and during the COVID-19 pandemic, there are limitations to consider when interpreting these results. First, administrative data are not collected for research purposes but for hospital administration purposes, which impacts the type of health information collected; therefore, the data may not capture certain confounding variables. Second, the implemented 30-day rule for ASCVD events may have deflated estimates for specific MACE endpoints including, AMI, IS, and UA. Therefore, estimates may not be comparable to previously reported literature. This study reported in-hospital mortality only; thus out-of-hospital deaths, including CV-related, were not captured and may therefore be underestimated. While death from COVID-19 may have been a competing risk to CV death among patients in the hospital, the impact of this on the observed CV-mortality rates is anticipated to be minimal as the number of COVID-19 cases during the lock-down period were low in Alberta. Finally, data from this study is from the province of Alberta, therefore, may not be generalizable to the rest of Canada. Many regions within Canada did impose similar restrictive measures to which these results may be generalizable, however, this study represents a time period when COVID infections were low in Alberta and may not reflect the situation or lockdown measures of other regions. CONCLUSION These study findings suggest acute and sustained impacts on ASCVD-related clinical outcomes and HCRU in Alberta during the first 12 months of the COVID-19 pandemic. It is important to highlight the increase in inpatient mortality due to a primary MACE that had not reverted back to pre-pandemic levels at the end of the study period. Understanding the impact of the COVID-19 pandemic and associated public health measures on this patient population is fundamental to improving the care and management of patients with ASCVD in the future. Further investigations regarding the long-term impact of system disruptions and treatment patterns throughout the course of the COVID-19 pandemic are warranted to draw awareness to the significant downstream impact placed on the health care system and cardiovascular patients. Supplementary Material ACKNOWLEDGEMENTS This study is based on data provided by Alberta Health. The interpretation and conclusions contained herein are those of the researchers and do not necessarily represent the views of the Government of Alberta. Neither the Government of Alberta nor Alberta Health expresses any opinion about this study. We would also like to thank Wayne Khuu and Guanmin Chen for their support on the data analysis, in addition to Patrick Berrigan, Claire Sharp, Hunter Graves, and Michelle Leong for their support with medical writing and general project support. FUNDING SOURCES This work was supported by Amgen Canada Inc. (Amgen), Mississauga, Ontario, Canada. Amgen collaborated with Medlior Health Outcomes Research Ltd. (Medlior) in the study design of the project. Medlior was responsible for requesting and analyzing the data, as well as reporting the results. DISCLOSURES This study was sponsored by Amgen Canada Inc (Amgen). Graves E, Ekwaru P, and McMullen S are employed by Medlior Health Outcomes Research Ltd., which received funding for the study from Amgen. Mackinnon ES, Wani RJ, and Packalen M are employed by Amgen Canada Inc. which funded this study and hold Amgen stock. Goodman SG has received research grant support (e.g., steering committee or data and safety monitoring committee) and/or speaker/consulting honoraria (e.g., advisory boards) from: Amgen, Anthos Therapeutics, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, CSL Behring, Daiichi-Sankyo/American Regent, Eli Lilly, Esperion, Ferring Pharmaceuticals, HLS Therapeutics, JAMP Pharma, Merck, Novartis, Novo Nordisk A/C, Pendopharm/Pharmascience, Pfizer, Regeneron, Sanofi, Servier, Tolmar Pharmaceuticals, Valeo Pharma; and salary support/honoraria from the Heart and Stroke Foundation of Ontario/University of Toronto (Polo) Chair, Canadian Heart Failure Society, Canadian Heart Research Centre and MD Primer, Canadian VIGOUR Centre, Cleveland Clinic Coordinating Centre for Clinical Research, Duke Clinical Research Institute, New York University Clinical Coordinating Centre, PERFUSE Research Institute, TIMI Study Group (Brigham Health). Grégoire JC has received consulting/Speaker’s Bureau honoraria from Amgen, Novartis, and Sanofi. Anderson TJ and Raggi P have no conflicts to disclose. ==== Refs REFERENCES 1 Tu JV, Khan AM, Ng K, Chu A. Recent Temporal Changes in Atherosclerotic Cardiovascular Diseases in Ontario: Clinical and Health Systems Impact. Can J Cardiol. 2017;33(3):378-384. Epub 2017/01/29. doi: 10.1016/j.cjca.2016.11.009 . Epub 2016 Nov 15. PubMed PMID: 28129964. 2 Benjamin EJ, Blaha MJ, Chiuve SE, Cushman M, Das SR, Deo R, et al. Heart Disease and Stroke Statistics-2017 Update: A Report From the American Heart Association. Circulation. 2017;135(10):e146-e603. Epub 2017/01/27. doi: 10.1161/cir.0000000000000485. PubMed PMID: 28122885; PubMed Central PMCID: PMC5408160. 3 Statistics Canada. Health Fact Sheets: The 10 leaning causes of death, 2013. Ottawa, Canada: Statistics Canada, 2017. 4 Barquera S, Pedroza-Tobías A, Medina C, Hernández-Barrera L, Bibbins-Domingo K, Lozano R, et al. Global Overview of the Epidemiology of Atherosclerotic Cardiovascular Disease. Arch Med Res. 2015;46(5):328-338. Epub 20150629. doi: 10.1016/j.arcmed.2015.06.006. PubMed PMID: 26135634. 5 Maclagan LC, Park J, Sanmartin C, Mathur KR, Roth D, Manuel DG, et al. The CANHEART health index: a tool for monitoring the cardiovascular health of the Canadian population. CMAJ. 2013:cmaj. 131358. 6 Chen G. Farris M.S. Cowling T. Pinto L. Rogoza R.M. MacKinnon E. Prevalence of atherosclerotic cardiovascular disease and subsequent major adverse cardiovascular events in Alberta, Canada: A real-world evidence study Clinical Cardiology 44 11 2021 1613 1620 10.1002/clc.23732 34585767 7 Tran D.T. Ohinmaa A. Thanh N.X. Welsh R.C. Kaul P. The Healthcare Cost Burden of Acute Myocardial Infarction in Alberta, Canada PharmacoEconomics - Open 2 4 2018 433 442 10.1007/s41669-017-0061-0 29623635 8 Wang D, Hu B, Hu C, Zhu F, Liu X, Zhang J, et al. Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China. JAMA. 2020. Epub 2020/02/08. doi: 10.1001/jama.2020.1585. PubMed PMID: 32031570; PubMed Central PMCID: PMC7042881. 9 Boukhris M, Hillani A, Moroni F, Annabi MS, Addad F, Ribeiro MH, et al. Cardiovascular Implications of the COVID-19 Pandemic: A Global Perspective. Can J Cardiol. 2020;36(7):1068-1080. Epub 2020/05/20. doi: 10.1016/j.cjca.2020.05.018. PubMed PMID: 32425328; PubMed Central PMCID: PMC7229739. 10 World Health Organization. WHO Director-General's Opening Remarks at the Media Briefing on COVID-19 - 11 March 2020 2020 [cited 2020 August 14]. Available from: www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020. 11 Hassan A, Arora RC, Adams C, Bouchard D, Cook R, Gunning D, et al. Cardiac Surgery in Canada During the COVID-19 Pandemic: A Guidance Statement From the Canadian Society of Cardiac Surgeons. Can J Cardiol. 2020;36(6):952-955. Epub 2020/04/18. doi: 10.1016/j.cjca.2020.04.001. PubMed PMID: 32299752; PubMed Central PMCID: PMC7194553. 12 World Health Organization. WHO Coronavirsus (COVID-19) Dashboard 2021. Available from: https://covid19.who.int/. 13 Alberta Go. COVID-19 orders and legislation: Government of Alberta; 2022 [cited 2022 November 23]. Available from: https://www.alberta.ca/covid-19-orders-and-legislation.aspx. 14 Impact of COVID-19 on Canada's health care systems. Canadian Institute for Health Information, 2022. 15 Pal R, Bhadada SK. Managing common endocrine disorders amid COVID-19 pandemic. Diabetes Metab Syndr. 2020;14(5):767-771. Epub 2020/06/11. doi: 10.1016/j.dsx.2020.05.050. PubMed PMID: 32521463; PubMed Central PMCID: PMC7271862. 16 Gittoes NJ, Criseno S, Appelman-Dijkstra NM, Bollerslev J, Canalis E, Rejnmark L, et al. ENDOCRINOLOGY IN THE TIME OF COVID-19: Management of calcium metabolic disorders and osteoporosis. Eur J Endocrinol. 2020;183(2):G57-G65. Epub 2020/05/13. doi: 10.1530/EJE-20-0385. PubMed PMID: 32396134. 17 Zheng Y.-Y. Ma Y.-T. Zhang J.-Y. Xie X. COVID-19 and the cardiovascular system Nat Rev Cardiol 17 5 2020 259 260 10.1038/s41569-020-0360-5 PubMed PMID: 32139904 32139904 18 Wang D. Hu B. Hu C. Zhu F. Liu X. Zhang J. Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China JAMA 323 11 2020 1061 1069 10.1001/jama.2020.1585 PubMed PMID: 32031570 32031570 19 Ad N, Luc JGY, Nguyen TC, Arora RC, Balkhy HH, Bender EM, et al. Cardiac surgery in North America and coronavirus disease 2019 (COVID-19): Regional variability in burden and impact. The Journal of Thoracic and Cardiovascular Surgery. 2021;162(3):893-903.e4. doi: https://doi.org/10.1016/j.jtcvs.2020.06.077. 20 Table 17-10-0005-01 Population estimates on July 1st, by age and sex [Internet]. Statistics Canada. 2021 [cited April 26 2021]. Available from: 2021https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1710000501. 21 Rennert-May E. Leal J. Thanh N.X. Lang E. Dowling S. Manns B. The impact of COVID-19 on hospital admissions and emergency department visits: A population-based study PLOS ONE 16 6 2021 e0252441 10.1371/journal.pone.0252441 22 Tam DY, Qiu F, Manoragavan R, Fremes SE, Hassan A, Ko DT, et al. The Impact of the COVID-19 Pandemic on Cardiac Procedure Wait List Mortality in Ontario, Canada. Can J Cardiol. 2021;37(10):1547-1554. Epub 20210930. doi: 10.1016/j.cjca.2021.05.008. PubMed PMID: 34600793; PubMed Central PMCID: PMC8481095. 23 Table 13-10-0768-01 Provisional weekly death counts, by age group and sex [Internet]. Statistics Canada. [cited April 26 2021]. Available from: https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1310076801. 24 Wijeysundera H.C. Machado M. Farahati F. Wang X. Witteman W. van der Velde G. Association of temporal trends in risk factors and treatment uptake with coronary heart disease mortality, 1994-2005 Jama 303 18 2010 1841 1847 10.1001/jama.2010.580 PubMed PMID: 20460623 20460623 25 New app helps Albertans access health care [Internet]. 2020. Available from: https://www.alberta.ca/release.cfm?xID=69851809AA1B8-AEA8-D268-E2D1E54D6DF119C0 26 Chief medical officer of health COVID-19 update – March 17, 2020 [Internet]. 2020. Available from: https://www.alberta.ca/release.cfm?xID=69830284041BB-FA70-03ED-82D4516CFE1CEA01 27 Relaxing the 30-day limit for prescription drugs [Internet]. 2020. Available from: https://www.alberta.ca/release.cfm?xID=725733A0EE850-DA12-F4D3-DD11FE650A730E28 28 Alberta Precision Laboratories. Cessation of Routine Laboratory & Fecal Immunochemical Testing (FIT) Testing during COVID-19 Pandemic Alberta2020. Available from: https://www.albertahealthservices.ca/assets/info/ppih/if-ppih-apl-memo-physicians-cease-non-essential.pdf. 29 Wong L.E. Hawkins J.E. Langness S. Murrell K.L. Iris P. Sammann A. Where are all the patients? Addressing Covid-19 fear to encourage sick patients to seek emergency care Nejm Catalyst 1 3 2020 1 12 30 Einstein Andrew J. Shaw Leslee J. Hirschfeld C. Williams Michelle C. Villines Todd C. Better N. International Impact of COVID-19 on the Diagnosis of Heart Disease Journal of the American College of Cardiology 77 2 2021 173 185 10.1016/j.jacc.2020.10.054 33446311 31 Sharma A, Harrington RA, McClellan MB, Turakhia MP, Eapen ZJ, Steinhubl S, et al. Using Digital Health Technology to Better Generate Evidence and Deliver Evidence-Based Care. J Am Coll Cardiol. 2018;71(23):2680-2690. Epub 2018/06/09. doi: 10.1016/j.jacc.2018.03.523. PubMed PMID: 29880129. 32 Services AH. Historical Hospital Occupancy. https://www.albertahealthservices.ca/br/page17593.aspx: AHS; 2023. 33 Alberta Go. COVID-19 cases in Alberta by day and case status. Probable cases include cases where the lab confirmation is pending. Cases are under investigation and numbers may fluctuate as cases are resolved. Data included up to end of day February 06, 2023.https://www.alberta.ca/stats/covid-19-alberta-statistics.htm#total-cases. February 6, 2023 ed: Alberta; 2023. 34 Domingo L. Comas M. Jansana A. Louro J. Tizón-Marcos H. Cos M.L. Impact of COVID-19 on Hospital Admissions and Healthcare Quality Indicators in Non-COVID Patients: A Retrospective Study of the First COVID-19 Year in a University Hospital in Spain Journal of Clinical Medicine 11 7 2022 1752 10.3390/jcm11071752 PubMed PMID: 35407360 35 Roth G.A. Vaduganathan M. Mensah G.A. Impact of the COVID-19 Pandemic on Cardiovascular Health in 2020: JACC State-of-the-Art Review J Am Coll Cardiol 80 6 2022 631 640 10.1016/j.jacc.2022.06.008 PubMed PMID: 35926937; PubMed Central PMCID: PMC9341480 35926937
PMC010xxxxxx/PMC10286560.txt	==== Front Am J Infect Control Am J Infect Control American Journal of Infection Control 0196-6553 1527-3296 Published by Elsevier Inc. on behalf of Association for Professionals in Infection Control and Epidemiology, Inc. S0196-6553(23)00479-0 10.1016/j.ajic.2023.06.013 Major Article Association Between Ventilator Associated Events and Implementation of Acute Respiratory Distress Syndrome (ARDS) Ventilator Weaning Protocol Marshall Gerard BS, AS, RRT, CIC Manager of Infection Prevention and Control Sanguinet Jennifer DrPh, FAPIC, CIC, MBA-HCM, BSIS Director of Infection Prevention and Control ⁎1 Batra Shreya BS Intern, Infection Prevention and Control Foreman Mary Jo RN, MBA, MHA, BSN, CIC Manager of Infection Prevention and Control Peruchini Justin BS Infection Prevention Patient Safety Coordinator Lopez Sarah BSN, RN, CIC, CRRN Infection Prevention Specialist De Guzman Rian BS Intern, Infection Prevention and Control Rivera Nancy MD, FACS Trauma Surgeon Hightower Todd BSN, RN Trauma Research Coordinator Malone Cheryl BSN, RN, CPN, TCRN Director Trauma Program Stucke Sheri Ph.D, APRN Trauma APRN Sunrise Hospital and Medical Center ⁎ Correspondence to: 3186 S Maryland Pkwy, Las Vegas, NV, 89109. 1 702-961-9571 22 6 2023 22 6 2023 24 5 2023 15 6 2023 16 6 2023 © 2023 Published by Elsevier Inc. on behalf of Association for Professionals in Infection Control and Epidemiology, Inc. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Acute respiratory distress syndrome (ARDS) is a severe and life-threatening condition that can occur in critically ill patients. Mechanical ventilation is a commonly used intervention with ARDS patients, but weaning patients off the ventilator can be challenging. An ARDSnet-like ventilator weaning protocol was implemented with the goal of reducing triggers for ventilator-associated events (VAEs). Methods The implementation of the new protocol was used to complete a retrospective investigation of patient outcomes for 1,233 ventilator periods. Periods were included between April and December 2022 for any ventilated patient lasting at least four days. National Healthcare Safety Network (NHSN) VAE criteria was used to surveille the patient data. Triggers were based on positive end expiratory pressure (PEEP) increases or fraction of inspired oxygen (FiO₂) increases. The pre-set weaning criteria was a reduction by two cmH2O per 24 hours. Results Of the total 1,233 individual ventilator periods, VAE criteria were met in 10%. Of the total 126 periods with VAE, 39.2% met criteria for appropriate protocol implementation. There was a statistically significant relationship between VAE identification and implementation of the protocol. Conclusions The implementation of a protocol for ventilator weaning affects the outcome of developing a VAE. The findings emphasize the importance of implementing ARDS weaning protocol as a template to reduce the triggers for VAEs and improve overall patient outcomes. Keywords Acute Respiratory Distress Syndrome (ARDS) Mechanical ventilation Ventilator–associated Events (VAE) Infection related ventilator-associated complications (IVAC) Possible ventilator- associated pneumonia (PVAP) Trauma Ventilator-Associated Condition (VAC) Burn Ventilator weaning protocol ==== Body pmcBACKGROUND Acute Respiratory Distress Syndrome (ARDS) is a common critical illness that requires mechanical ventilation to support respiratory function.1 ARDS requires appropriate management of mechanical ventilation to improve patient prognosis and reduce incidence of VAEs.2 Prolonged mechanical ventilation is linked to issues like ventilator-induced lung injury (VILI), muscle weakness, and infections, which can result in higher morbidity and mortality rates.3 Multiple complications, like fluid imbalanced and trauma, associated with ARDS may exacerbate existing respiratory issues. 3 The National Healthcare Safety Network (NHSN) uses the term “ventilator-associated events” (VAEs) to refer to a range of problems that may occur in patients who need mechanical ventilation to maintain their respiratory capability.4 Major influences on VAEs may include pneumonia, lung damage, and other ailments that contribute further to the incidence of impaired respiratory function. The NHSN criteria establishes clear guidelines to determine when a new VAE is defined and when the current VAE moves up through the higher risk VAE ( Figure 1).4 A patient starting with mechanical ventilation that is maintained following the protocol may never meet criteria for a ventilator-associated condition (VAC). However, once a VAC is determined, the risk for an infection related ventilator-associated complication (IVAC) or possible ventilator-associated pneumonia (PVAP) greatly increases. Due to multiple variables, including extended mechanical ventilation and reduced immunological function, patients with ARDS are more likely to acquire VAEs.5, 6 Increased pneumonia risk is one of the main ways that ARDS affects VAE.3 Pneumonia may exacerbate general health and cause additional lung damage. Patients with traumatic injuries such as rib fractures, pulmonary contusions, pneumothorax, or aspiration are placed at a higher risk for the development of VAE and pneumonia.7, 8 Fig. 1 VAE categories:. Fig. 1 Ventilator weaning is a complex process that requires an implementation of evidence-based, lung protective strategies.9 Understanding the best practices for ventilator weaning in ARDS patients is critical in reducing morbidity and mortality associated with prolonged mechanical ventilation.9 The outcome of the study was to determine if an association exists between the implementation of an evidence-based protocol for mechanically ventilated patients and subsequent development of VAE. METHODS A new protocol was developed through collaboration of a ventilator work group including intensivists, Chief Medical Officer, infection preventionists, and respiratory therapists. The implementation of the new protocol was used to complete a retrospective investigation of patient outcomes for 1,233 ventilator periods which includes a new ventilator period if there is at least one full calendar day break between ventilator removals. All patients that had a ventilator period of at least four days was included. Those with three or fewer ventilator days were excluded as they would not meet criteria for a VAE. The study included periods between April 2022 and December 2022. The specialty ICU units included services from burn, cardiovascular, medical, neurosurgical, and trauma. In these ICUs there were 173 beds and all ventilated patients had the protocol ordered. The ICUs were chosen due to the broad type of potential ventilator management based on the patient medical needs, including traumatic injuries and/or medical conditions. The protocol was implemented to identify ventilator parameters and adjustments to FiO₂ to optimize the management for each individual patient. The protocol provides a sliding scale method for oxygenation-based positive end expiratory pressure (PEEP) increases or fraction of inspired oxygen (FiO₂) adjustment. The protocol was followed for patients with acute lung injury or (ARDS) when ordered by the physician. FiO₂ and PEEP were adjusted using the criteria listed in Table 1.Table 1 Oxygenation Goal Table:. Table 1Oxygenation Goal TABLE FIO2 0.4 0.5 0.6 0.7 0.8 0.9 1.0 PEEP 8 10 12 14 14 16 18 The protocol scale outlines goals for the initiation and escalation of PEEP values when oxygenation needs increase. The scale was not to be used for rapid de-escalation of PEEP when FiO₂ is decreased. PEEP and FiO₂ adjustments were based on the oxygenation goal when responding to the patient’s whole clinical picture. If a PEEP greater than 12 cmH2O is needed for multiple days in a row, then when weaning PEEP, the baseline should not be reduced by more than two cmH2O every 24 hours.10 The application of PEEP improves gas exchange and lung function. The main effect of increasing PEEP is to maintain the recruitment of alveolar units that were previously collapsed. Thus, since the tidal volume is distributed to more alveoli, peak airway pressure is reduced and elasticity is increased., 11 It is not recommended to lower the PEEP faster than 2 cmH₂O per day or below eight cmH₂O for this population. When determining whether extubation is appropriate, a brief time interval of PEEP of five may be used. If FiO₂ requirements are rising in response to provider orders, PEEP may be increased and adjusted more frequently to achieve adequate oxygenation levels. An added benefit of using the oxygenation goals is that inadvertent triggers for VAE are avoided. A chi-square test of independence was performed to examine the relationship between the implementation of an ARDSnet-like evidence-based lung protective protocol for mechanically ventilated patients and development of VAE. RESULTS In this period, a total of 56,127 patient days occurred with 16,360 ventilator days which were spread among seven intensive care units (ICU). The highest percentage of the total VAE were in medical, cardiovascular and trauma ICUs ( Table 2). Ventilator utilization per 100 patient days for the top three services were medical, cardiac (step down from cardiovascular), and trauma (Table 2). The rate of VAE per 1,000 ventilator days was the highest in order cardiovascular, burn, cardiac, and trauma (Table 2).Table 2 VAE Totals and Ventilator Utilization. Table 2 Total VAE Ventilator Utilization per 100 patient days Rate of VAE per 1,000 vent days Burn 5 11.11 12.76 Cardiac 8 42.1 9.22 Medical Step-down 1 17.45 0.42 Cardiovascular 20 12.29 17.75 Medical 55 53.31 8.90 Neuro 10 30.57 5.19 Trauma 27 35.32 7.71 Total 126 29.15 61.95 NHSN VAE classifications were used for the study period of April through December 2022.The overall total of ventilator periods was 1,233 ( Table 3). Of the total 1,233 periods, 66% of the periods followed the protocol. Of the 66% (813), 6% were complicated by a VAE. Of the total 49 periods where a VAE developed, 53% developed into a VAC, 37% developed into an IVAC, and 10% developed into a PVAP after following the protocol. VAC developed in 44.6% less periods when the protocol was implemented compared to when the protocol was not implemented. Similarly, IVAC developed 18.2% less and PVAP developed 28.6% less when the protocol was implemented. The association between the implementation of an ARDSnet-like evidence-based lung protective protocol for mechanically ventilated patients and development of VAE was significant, X 2(2, N =1,233) = 44.2728, p =.00001. There is an a statistically significant relationship between the implementation of an ARDSnet-like protocol and development of a VAE. Overall, the VAEs identified when protocol was not implemented outweighed the VAEs where the protocol was implemented by 35.5%.Table 3 Crosstabulation of Protocol Followed and Type of VAE Identified Protocol. Table 3 PROTOCOL FOLLOWED NO YES TOTAL VAC Count 47 26 73 IVAC Count 22 18 40 PVAP Count 7 5 12 No VAE Count 344 764 1108 Total Count 420 813 1233 DISCUSSION Implementation of a ventilator weaning protocol has a statistically significant relationship with the development of a VAE. Based on the results, reduction occurred in all levels of the VAE algorithm when the protocol was implemented. The findings show that a decrease in VAE triggers leads to a decrease in progression of the clinical processes that can lead to further respiratory complications including mortality12. In line with other research comparing NHSN and PVAP outcomes, the periods showed similar distribution with trauma having the second highest VAEs identified while having the fourth highest ventilator utilization rates.12 The protocol outlines a consistent process for implementing standardized ventilator management coordination. The time when the highest patient risk for events existed was post implementation of PEEP levels of 13 to 20 cm H₂O. Implementation of the protocol was a mitigation technique. Lung integrity is maintained by reducing the PEEP by only 2 cm H2O per 24 hours. FiO2 levels can be reduced more aggressively when weaning high PEEP levels because the lung integrity is maintained by functional residual capacity which is already sustained by the PEEP. The level of periods where a VAE did not develop showed that regardless of underlying medical issues, the protocol achieves a successful outcome by gradually lowering the ventilator's level of support while maintaining the patient's respiratory function and while appropriately reacting to the needed clinical changes. A properly implemented ARDS ventilator weaning approach is crucial for enhancing patient outcomes, potentially lowering morbidity and mortality while potentially maximizing hospital resource use. ETHICAL CONSIDERATIONS Data access to the secondary data set required permission via a publisher clearing process through the company submitted by the primary author. Confidentiality, honesty, and integrity in all data gathering, storage, and use were consistently maintained. All data were kept secured via password protection for at least 5 years. CONCLUSIONS The results showed a statistically significant association between implementation of the protocol and development of a VAE. The findings reiterate the importance of implementing a protocol to reduce the triggers for VAE and improving overall patient outcomes. Protocol implementation as a part of a multidisciplinary process, can assist patients in recovering their ability to breathe on their own while lowering the chance of consequences related to ventilation such as pneumonia, infections, and other respiratory issues. DISCLAIMER This research was supported (in whole or in part) by HCA Healthcare and/ or an HCA Healthcare affiliated entity. The views expressed in this publication represent those of the author(s) and do not necessarily represent the official views of HCA Healthcare or any of its affiliated entities. Conflict of Interest No conflict of interests to disclose ==== Refs References 1 Saguil, A., & Fargo, M.V. (2012). Acute respiratory distress syndrome: diagnosis and management. American fam. 2 Sim J.K. Oh J. Min K. Hur G. Lee S.H. Lee S.Y. Kim J. Shin C. Shim J. Kang K. Clinical significance of ventilator-associated event Journal of Critical Care 35 2016 19 23 27481731 3 Magill S.S. Rhodes B. Klompas M. Improving ventilator-associated event surveillance in the National Healthcare Safety Network and addressing knowledge gaps: update and review Current opinion in infectious diseases 27 4 2014 394 400 10.1097/QCO.0000000000000083 24945615 4 National Healthcare Safety Network. (2023), Ventilator-Associated Event (VAE) - CDC. 〈https://www.cdc.gov/nhsn/pdfs/pscmanual/10-vae_final.pdf〉. 5 Magill S.S. Klompas M. Balk R. Burns S.M. Deutschman C.S. Dikeman D. Lipsett P. ). Developing a new, national approach to surveillance for ventilator-associated events: executive summary Clinical infectious diseases 57 12 2013 1742 1746 24280662 6 Villar J. The use of positive end-expiratory pressure in the management of the acute respiratory distress syndrome. Minerva Anestesiol 2005;71(6):265-272.Klompas M. Interobserver variability in ventilator-associated pneumonia surveillance. Am J Infect Control 2010;38: 237-9.ily physician, 85(4), 352-358. 7 Cook A. Norwood S. Berne J. Ventilator-associated pneumonia is more common and of less consequence in trauma patients compared with other critically ill patients Nov J Trauma 69 5 2010 1083 1091 10.1097/TA.0b013e3181f9fb51 21068613 8 Rello J. Ollendorf D.A. Oster G. Vera-Llonch M. Bellm L. Redman R. Kollef M.H. VAP Outcomes Scientific Advisory Group Epidemiology and outcomes of ventilator-associated pneumonia in a large US database Dec Chest. 122 6 2002 2115 2121 10.1378/chest.122.6.2115 12475855 9 Wawrzeniak I.C. Regina Rios Vieira S. Almeida Victorino J. Weaning from Mechanical Ventilation in ARDS: Aspects to Think about for Better Understanding, Evaluation, and Management BioMed research international 2018 2018 5423639 10.1155/2018/5423639 10 Papazian L. Aubron C. Brochard L. Chiche J.D. Combes A. Dreyfuss D. Faure H. Formal guidelines: management of acute respiratory distress syndrome Annals of intensive care 9 1 2019 1 18 11 Pintado M.C. de Pablo R. Trascasa M. Milicua J.M. Rogero S. Daguerre M. Sánchez-García M. Individualized PEEP setting in subjects with ARDS: a randomized controlled pilot study Respiratory care 58 9 2013 1416 1423 23362167 12 Michetti C.P. Fakhry S.M. Ferguson P.L. Cook A. Moore F.O. Gross R. AAST Ventilator-Associated Pneumonia Investigators. Ventilator-associated pneumonia rates at major trauma centers compared with a national benchmark: a multi-institutional study of the AAST May J Trauma Acute Care Surg 72 5 2012 1165 1173 10.1097/TA.0b013e31824d10fa 22673241
PMC010xxxxxx/PMC10286561.txt	==== Front Prostaglandins Other Lipid Mediat Prostaglandins Other Lipid Mediat Prostaglandins & Other Lipid Mediators 1098-8823 1098-8823 The Author(s). Published by Elsevier Inc. S1098-8823(23)00059-X 10.1016/j.prostaglandins.2023.106762 106762 Article Specialized Pro-resolving Lipid Mediators and Resolution of Viral Diseases Ferri Giulia 1 Mucci Matteo 1 Mattoscio Domenico ⁎ Recchiuti Antonio ⁎ Department of Medical, Oral, and Biotechnology Science (DSMOB), “G.d’Annunzio” University of Chieti – Pescara, Center for Advanced Studies and Technology (CAST), via Polacchi 13, 66100 Chieti (IT) ⁎ Corresponding authors. 1 These authors contributed equally 22 6 2023 22 6 2023 10676215 3 2023 16 6 2023 19 6 2023 © 2023 The Authors 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The COVID-19 pandemics has made sparkly evident the importance of acute inflammation and its timely resolution to protect humans from pathogenic viruses while sparing them from collateral damages due to an uncontrolled immune response. It is clear now that resolution of inflammation is an active process regulated by endogenous specialized proresolving lipid mediators (SPM) biosynthesized from essential polyunsaturated fatty acids. Accruing evidence indicates that SPM are produced during viral infections and play key roles in controlling the magnitude and duration of the inflammatory response and in regulating adaptive immunity. Here, we reviewed biosynthesis and bioactions of SPM in virus-mediated human diseases. Harnessing SPM and their proresolutive actions can help in providing new therapeutic approaches to current and future human viral diseases by controlling infection, stimulating host immunity, and protecting from organ damage. Keywords inflammation lipid mediators immunity GPCR Neutrophils Macrophages leukocytes COVID-19 respiratory viruses viral infections Resolution ==== Body pmc1 Inflammation and viral infections Viral infectious diseases represent a major threat for human health, especially those sustained by emerging viruses, such as avian influenza, Ebola, and coronaviruses. In viral infections, the acute inflammatory response is meant to be a primordial, necessary protective mechanism to restrain microorganisms, adequately initiate the adaptive cellular and humoral immune response, and to allow tissue repair [1], [2]. The importance of acute inflammation during infections is evident in neutropenic individuals, who typically succumb of disseminated infections [3]. Acute inflammation can be divided into 2 general phases: initiation and resolution. The classical cardinal signs of the initiation phase identified by Celsius, i.e., rubor (redness), tumor (swelling), calor (heat), and dolor (pain) are gross manifestations of molecular and cellular responses. Following infections, increased blood flow and microvascular permeability result into tissue edema, mediated by lipid mediators (eg., cysteinyl leukotrienes and prostaglandins) and other vasoactive mediators. Subsequently, polymorphonuclear neutrophils (PMN) are among the first white blood cell that accumulate – sensing leukotriene (LT) B4 and other chemo attractants - in inflamed tissues. Monocytes enter as a second wave and differentiate into macrophages (MФs). Infiltrated PMN and monocyte-MΦs are crucial for killing microbes, infected cells, and contain the spread of infection. Once the inciting cause is removed, leukocytes play also key roles in clearing the site from dead cells through non-phlogistic phagocytosis and repair the damaged tissue [2], [4]. Countless times every day, viral infections remain unnoticed because acute inflammation protects us and these challenges are timely eliminated and inflammation resolves. Viruses are obligate intracellular parasites that infect and replicate exclusively within cells of many living organisms, including bacteria, fungi, protozoa, plants, and animal. Their identification dates back to late XIX century, when Dmitrii Iwanowski (1864-1920) proposed that the disease Adolph Mayer (1843-1942) named tobacco mosaic disease was caused by an infectious agent several times smaller than bacteria [5], [6]. Almost contemporaneously, Martinus Beijerinck (1851-1931) replicated Iwanoski’s findings and called the pathogenic agent of the tobacco mosaic disease “contagium vivum fluidum” (contagious living fluid) [7]. It was not until the Nobel laureate Wendell M Stanley (1904-1971) obtained the first crystal of the tobacco mosaic virus that viruses were proven to be particulate microorganisms [8]. Their discoveries marked the beginning of virology and made possible to understand the etiology and pathophysiology of diseases that were described much earlier [9]. Viruses are divided according to the Baltimore classification based on the structure of their genome, strandedness, sense, and method of replication into 7 classes encompassing > 30,000 isolates ( Fig. 1), most of which do not cause serious illness to the human population. However, many viruses can cause common, severe, or even life-threatening diseases involving brain, hearth, blood, liver, pancreas, gut, lungs, skin and mucous membranes.Fig. 1 Baltimore classification of viruses The figure represents the Baltimore classification of DNA and RNA viruses. This classification was originally proposed by the Nobel laureate David Baltimore as a scheme for organizing known viruses based on the nature of their genome and replication strategy [92]. Fig. 1 As an arm of innate immunity, acute inflammation represents a formidable barrier mechanism to suppress viral replication and spread. It is also important for activating adaptive immunity and, hence, coordinating the overall host immune system. Acute inflammation is activated upon recognition of viral pathogen associated molecular patterns (PAMPs) by the host pattern recognition receptors (PRRs), which encompass toll-like receptors (TLRs), Nod-like receptors (NLRs), and RIG-I-like receptors (RLRs). These PRRs sense specific viral molecules and signal downstream pathways that culminate with recruitment and activation of leukocytes, enhancement of cytokines and chemokines, and induction of antiviral genes like type I and III interferons [10]. Innate immune responses mediated by acute inflammation normally can clear virally infected cells and resolve virosis. On the contrary, inability to mount a timely and effective pro-resolution and antiviral responses can lead to virus persistence, pathogenic excessive inflammation, and fatal outcomes. Influenza viruses and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) offer clear examples of this [11], [12], emphasizing the crucial role and intriguing therapeutic functions of pro-resolving mechanisms during viral infections. 2 Resolution of inflammation and SPM One of the major strides in our understanding of inflammation was the discovery that resolution is an active process introduced by the biochemical synthesis of specific proresolving molecules [13], among which are lipid mediators derived from ω-6 and ω-6 fatty acids that were dubbed “Specialized Pro-resolving lipid Mediators (SPM) by its discoverer CN Serhan. Using a system lipidomics-informatics approach to self-resolving inflammation, pioneering research from the laboratory of Dr. Serhan led to the discovery of SPM in inflammatory exudates during resolution [14], [24]. SPM act through specific receptors to halt excessive PMN infiltration and activation, counter pro-inflammatory signals, enhance the active clearance of pathogens and death cells by MΦ, protect organ from loss of function, and stimulate tissue regeneration. Notably, SPM biosynthesis is impinged by pro-inflammatory mediators, for instance prostaglandin E2, generated during the onset of the inflammatory response [15], indicating that “the beginning of inflammation programs its end”. The main biosynthetic pathways and members of SPM families are shown in Fig. 2.Fig. 2 Illustration of SPM biosynthesis Precursors AA, EPA, DHA and n-3 DPA polyunsaturated fatty acids (PUFA) are converted via biosynthetic enzymes to SPM. The pie chart visualizes the members of SPM accordingly to they precursor. The size of each slice is proportionate to the number of SPM produced by the specific precursor. See text for details on each SPM structure and biosynthetic mechanisms. Fig. 2 The eicosanoid lipoxins (LX) A4 and B4 are the SPM derived from arachidonic acid (AA) metabolism [16], [17], [18]. The regio- and stereoselective oxidation catalized by 5- and 15-lipoxygenase (LO) constitutes the first pathway for the formation of LX [16], [17], [18]. This pathway occurs in 15-LO expressing epithelial cells or MФ and leukocyte 5-LO. A second pathway relies on the LX synthase activity of 12-LO in platelets during cell-cell interactions with PMN [19], [20]. A the third pathway produces LX epimers, i.e., 15-epi-LXA4 and 15-epi-LXB4, which are formed in the presence of acetylated cyclooxygenase (COX)-2. Since this pathways was originally described with aspirin-treated endothelial cells expressing COX-2, 15-epi-LX are also called “aspirin-triggered lipoxins” (ATL) [21], [22]. RvE1 was the first SPM isolated from eicosapentaenoic acid (EPA)[23]. The current members of the E-series resolvins include RvE1, RvE2, and RvE3, with the recent addition and elucidation of RvE4. They are produced through transcellular biosynthesis with human neutrophils by acetylated cyclooxygenase-2 (COX-2) or microbial cytochrome P450 [24]. Docosahexaenoic acid (DHA)-SPM include D-series resolvins, protectins, and maresins. The D-series resolvins (RvD1-6) are biosynthesized from the sequential oxygenation of precursor ω-3 fatty acid docosahexaenoic acid (DHA) [25], either via aspirin-triggered cyclooxygenase catalysis (17(R) AT-RvDs) or via the lipoxygenase pathway (15-LOX-1 and 15-LOX-2) forming the epimeric 17(S)-RvD1-6 resolvins [26]. Protectin D1 (PD1) is biosynthesized by DHA via 15-LOX and is produced enzymatically by human leucocytes from 16,17-epoxide-intermediates, PMN, macrophages, and eosinophils [27]. The third group produced by DHA biosynthesis is the Maresins (MaR-1 and MaR-2). Maresin biosynthesis occurs from carbon-14 via human 12-LOX, producing a 13(14)epoxide-intermediate (eMaR) that stimulates the conversion of M1 to M2 macrophages and blocks LTA4 hydrolase [28], [29]. Another precursor substrate for SPM formation is n-3 docosapentaenoic acid (n-3 DPA). n-3 DPA is converted into new SPM, including RvDn-3 DPA, MaRn-3 DPA, and PDn-3 DPA, as well as into series 13 resolvins (RvTs). SPM from n-3 DPA are characterized by the presence of an -OH group at position C13 in the PUFA chain [30], [31]. Three novel series of SPM conjugated with peptide-lipids have been recently introduced. They include maresin conjugates for tissue regeneration (MCTR), protectin conjugates for tissue regeneration (PCTR) and resolvin conjugates for tissue regeneration (RCTR), collectively referred to as cysteinyl-SPM (cys-SPM)[30], [31]. Recent studies confirm their pro-resolution action and organ protection in many organs, including lungs [32], [33], [34], [35] (and reviewed in [23]). Several SPM G-protein coupled receptors (GPCRs) have been identified to date, using robust pharmacological approaches including library screening, specific binding with labeled ligands, engineered GPCR-β-arrestin cell for monitoring receptor engagement, and gain and loss of function strategies(recent reviewed in [36], [37]) ( Fig. 3). These GPCRs convey SPM actions transmitting signals to activate intracellular pathways and cell responsesFig. 3 SPM receptors The figure shows molecular graphics and protein structure of identified SPM GPCR. Analyses were carried out with UCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from National Institutes of Health R01-GM129325 and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases. Fig. 3 Readers interested in cell- and tissue/organ-specific SPM bioactions are directed to excellent recent papers [36], [38]. Substantial evidence has accumulated that pro-resolving endogenous mediators also encompass proteins and peptides. One of the first polypeptide identified playing crucial biological roles in regulating acute inflammation is the glucocorticoid-regulated protein annexin (Anx) A1. AnxA1 is a 37 kDa protein initially recognized as an inhibitor of phospholipase A2 and, therefore, as a cellular mediator of glucocorticoid pharmacological anti-inflammatory actions, including the inhibition of prostaglandin and leukotriene biosynthesis [39], [40], [41], [42]. However, it is now clear that the biology of AnxA1 and N-teminal peptides with glucocorticoids is much more complex than initially thought. Readers interested can refer to [43] for excellent coverage of this topic. The development of recombinant human AnxA1 (hr-AnxA1) has helped to understand its biological activities, including the control of leukocyte migration, the promotion of neutrophil apoptosis, and induction efferocytosis, which underlie the therapeutic potential of the AnxA1-centered proresolving pathway that has been demonstrated in various experimental models [43]. The discovery of the lipoxin A4 receptor (ALX/FPR2) as a receptor for AnxA1 actions was also crucial in decoding mechanisms underlying resolution of inflammation, since ALX was the first GPCR identified that binds proresolving ligands of lipid and peptide structure[44]. The AnxA1- ALX/FPR2 axis is associated with key events in the resolution of inflammation, such as decreased neutrophil recruitment, induction of noninflammatory monocyte recruitment, promotion of neutrophil apoptosis and efferocytosis, contribution to tissue repair and resolution program amplification [43]. 3 SPM and viral infection In addition to their well-characterized roles in tissue homeostasis described above, several studies highlighted beneficial functions of SPM in the modulation of host responses to various infectious diseases triggered by viruses ( Table 1). Indeed, a large body of evidence shows that SPM decrease the inflammatory response by 1) promoting resolution and clearance of infection through modulation of host cell activities and 2) directly affecting the life cycle of viruses. These combined actions reduce viral replication in target cells thus providing greater ability of the host to deal with harmful infection. Along these lines, defective SPM production is suppressed during viral challenges and inversely correlated with virus pathogenicity [45], [46]. Therefore, SPM may represent a viable strategy for controlling the viral load and the excessive inflammation during viral infections.Table 1 Main bioactions of SPM in virus-mediated diseases. Table 1Virus SPM Function References Influenza AnxA1 PD-1 and PDX AT-RvD1 MCTR • Promotes viral replication. • Reduce viral RNA nuclear export. • Reduces PMN infiltration and pneumonia severity promoting pro-resolution pathways. • Reduce post-IAV S. pneumonia infection [45], [48] [50] [53] [56] Respiratory syncytial virus LXA4, RvE1 RvD1 PCTR1, PD1 • Induce gene expression of arginase-1 and mannose receptor in mouse Mφ. • Increases the frequency and modulate memory CD8 T cells gene expression by increasing transcript of anti-inflammatory genes II-4, II-10, and Ifng. • Regulate host antiviral immunity and inflammation [59] [60] [61] SARS-CoV-2 RvD1, RvD2 • Restore phagocytic ability of Mφ • Reprogram Mφ toward lower production of pro-inflammatory cytokines. • Abate the inflammatory responses induced by SARS-CoV-2 virion spike 1 glycoprotein (S1). [67], [75] [75] [67], [70] Herpes viruses RvE1 PD1 AT-RvD1 • Reduces PMN and pathological CD4 T cells infiltration while increases anti-inflammatory IL-10. [82] [83] [84] Kaposi’s Sarcoma-Associated Herpesvirus LXA4 • Reduces levels of key pro-inflammatory mediators (PGE2, LTB4, IL-6, IL-8). • Decreases the expression of PD-L1. [86] [87] 3.1 Influenza A virus Respiratory viruses are among the most frequent causative agents of disease in humans, causing illness in nose, throat and breathing passages including lungs, with significant impact on morbidity and mortality. Respiratory viruses include rhinoviruses and enteroviruses (Picornaviridae), influenza viruses (Orthomyxoviridae), parainfluenza, metapneumoviruses and respiratory syncytial viruses (Paramyxoviridae), coronaviruses (Coronaviridae), and several adenoviruses. A number of studies evaluated the relevance of SPM in the context of viral infections of the lung, especially influenza A (IAV), a negative-sense RNA viruses that causes seasonal epidemics of disease in people, particularly harmful in fragile individuals [47]. Using lung tissues lipidomics in mice subjected to intratracheal inoculation of the H1N1 PR8 strain, the lipid protectin D1 (PD1) isomer PDX was identified as one of the most reduced lipid mediators in the lungs of PR8-infected mice. Mechanistically, this reduction could be ascribed to a viral-induced defect of the 12/15-LOX enzyme, a key component of PD biosynthesis. Importantly, treatment of mice with exogenous PD increased survival and improved pulmonary injury trough reduction of viral titers in lungs of PR8 challenged mice. In vitro and in vivo experiments demonstrated that, rather than altering the host inflammatory response, PDX dampened IAV life cycle via attenuation of viral RNA nuclear export, a key step for virus replication [45], [48]. Similarly, the highly pathogenic H5N1 IAV altered the gene expression levels of the lipoxin pathway machinery in lungs to disseminate in multiple organs after infection in mice lungs [46]. Among the most affected genes, these included the suppressor of cytokine signaling (SOCS) 2 gene, an intracellular lipoxin mediator regulating cytokine and immune cells dynamics [49], thus indicating that this reduction could crucially impair pro-resolutive actions against viral infection. Therefore, IAV hijack key pathways of SPM biosynthesis to reduce the production of crucial anti-viral and pro-resolutive SPM that would impair viral proliferation and dissemination. Opposite to these beneficial effects of SPM supplementation, the pro-resolving AnxA1 enhanced IAV infectivity. Indeed, AnxA1 deficient (-/-) mice are protected against IAV infection due to an enhanced leukocyte infiltration, thus suggesting a sustained inflammatory response against viral infection. In addition, AnxA1 silencing and overexpression experiments in vitro suggested that, in addition to regulate host immunity, the presence of AnxA1 promoted viral replication, binding at the host cell membrane, viral uptake by host cells, viral transport to the nucleus and viral-induced apoptosis of target A549 lung cells, all key steps leading to greater virus production. Mechanistically, AnxA1 was incorporated within IAV and co-localized with the IAV protein NS1 in endosomes, indicating that AnxA1 facilitated endosomal trafficking and IAV infection life cycle [50]. These effects could be also due, at least in part, to ALX. As discussed, ALX is a plastic receptor able to sense and to activate a variety of pro-inflammatory and pro-resolving stimulus, such as AnxA1, LXA4 and RvD1 among the latter. IAV infection up-regulated ALX in murine lungs and lungs human cell lines [51], thus suggesting the needed of the virus to exploit the receptor to support the viral cycle. Indeed, activation of ALX with the agonists WKYMVm-NH2 and IAV harboring AnxA1 increased viral replication in vitro and in vivo and altered cytokine release in lungs of infected mice [51]. These effects of AnxA1 on IAV are in sharp contrast with those demonstrated in viral dengue fever, a potentially lethal hemorrhagic disease caused by one of the 4 serotypes of dengue virus (DENV1-4) transmitted through mosquitos that can result in fatal exacerbation of innate and adaptive immune responses. Indeed, Costa and Sugimoto recently demonstrated that therapeutic administration of an AnxA1 derived peptide to DENV-infected mice improves clinical signs of the disease (e.g., reduction in blood platelets and hematocrit), liver damage, and inflammatory markers. Strickingly, the absence of AnxA1 or its receptor ALX in knockout mice resulted in more severe illness of DENV-infected animals, signifying the important protective roles of AnxA1 and ALX in dengue fever [52]. Therapeutic treatment with AT-RvD1, another agonist of ALX, during an acute co-infection pneumonia in mice co-infected with Streptococcus pneumoniae and IAV, markedly reduced PMN infiltration and pneumonia severity promoting pro-resolution pathways [53]. Therefore, even in in a co-infection model, these results signified that diverse stimuli (peptide vs lipid) may differentially fuel ALX to translate pro-inflammatory and pro-viral or pro-resolutive signals that could be explored for therapeutic purposes. SPM also hold the potential to activate adaptive immunity as well. In particular, the DHA-derived SPM 17-hydroxydocosahexaenoic acid (17-HDHA) enhanced plasma cell differentiation and production of specific antibodies (Abs) directed against the recombinant H1N1 hemagglutinin (HA) used to immunize mice. Importantly, 17-HDHA–mediated HA-specific Abs protected mice live influenza infection, indicating that 17-HDHA increase a defensive humoral response sustaining a specific B-cell differentiation and Ab-secreting phenotype [54]. Similarly, studies showing that LXB4 enhances the production of IgG in B lymphocytes derived from donors vaccinated against influenza [55] and others demonstrating that MCTR protect from bacterial pneumonia post-IAV acting on MΦ [56] confirm that SPM could be used to stimulate host immunity against IAV and collateral bacterial infections. 3.2 Respiratory syncytial virus Respiratory syncytial virus (RSV) is a common respiratory virus infects the lungs and respiratory tract causing mild, cold-like symptoms except in infants, older adults, and fragile people where severe infections lead to pneumonia and bronchiolitis [57]. After epithelia infection, RSV elicits a potent inflammatory response mainly sustained by pro-inflammatory (M1) lung MΦ that, as expected, is dampened after MΦ skew to M2 polarization [58]. Along these lines, in vitro treatment with LXA4 or RvE1 induced gene expression of arginase-1 and mannose receptor in mouse MΦ from 5LO-/- transgenic mice, suggestive of M2 alternative activation that stimulate RSV resolution [59]. Similarly to IAV, SPM modulate the adaptive harm of immunity during RSV infection. In particular, exposure of RSV infected mice with RvD1 increased the frequency of specific memory precursors CD8 T cells against virus in the lung, and modulate memory CD8 T cells gene expression by increasing transcript of anti-inflammatory genes II-4, II-10, and Ifng [60]. Recent work shows that intranasal administration of PCTR1 and PD1 in RVS-infected mice decrease viral load and leukocyte infiltration while raising IFN-responses [61]. Collectively, these results highlight the critical role of SPM in the immune and inflammatory host response to RSV. 3.3 SARS-CoV-2 The coronavirus disease (COVID-19) caused by the severe acute respiratory syndrome Coronavirus 2 (SARS-CoV-2) has been an unprecedented global threat for human health. SARS-CoV-2 is a RNA virus that infects the lungs, that encompass a wide range of symptoms and variable clinical outcomes, with many people developing severe, often lethal, pneumonia, sepsis, and respiratory failure, and others showing only a mild illness that resolves in few days [62]. SARS-CoV-2 can also determine blood disturbancies, including clotting and formation of NETs (neutrophil extracellular traps) [12]. Evidence indicates that viral load is not correlated with the worsening of the symptoms, while cytokine storm, increase in inflammatory mediators, and an imbalance in immunity are associated with poor prognosis [12], [63], [64], [65]. Indeed, it is now clear that failure in resolution of inflammation is a key determinant of SARS-CoV-2 infection. Serum and bronchoalveolar lavage fluids of symptomatic SARS-CoV-2 infected patients ad significantly higher concentrations of both omega-6–derived proinflammatory lipids and omega-6– and omega-3–derived SPM age- and sex-matched SARS-CoV-2–negative group, suggestive of an unrelenting inflammation that failed to resolve [66], [67], [68]. Importantly, failure in resolution also characterized worsen outcomes in COVID-19 disease. Patients who recovered from disease showed upregulation of the pro-resolution peptide AnxA1 in peripheral blood monocytes, indicating increased resolution in people that mount an active response against SARS-CoV-2 [69]. Consistent with this, lipid mediator profiling demonstrated that plasma SPM concentrations were downregulated in people with severe COVID-19 disease [67], [70], [71], [72]. More in details, an overall downregulation in PCTR3, MCTR3 [67] and RvD3 [70] were observed in patients with severe disease, with an upregulation of arachidonic acid-derived LM, including LTB4 and LTE4, LTC4 and LTD4. Importantly, in addition to reduced SPM levels, patients with severe disease demonstrated lowered expression of SPM biosynthetic enzymes (ALOX15B in neutrophils, COX-2 and ALOX5 in classical monocytes, and ALOX5 in nonclassical monocytes) [67], [70] and receptors (GPR18 on neutrophils, classical monocytes, and nonclassical monocytes, ChemR23, on classical and intermediate monocytes, GPR32 on intermediate and nonclassical monocytes, GPR101 on classical monocytes) on circulating leukocytes [67]. Collectively, these results pointed to defects in SPM biosynthesis and production as critical determinant for disease severity and suggested that restoration of adequate pro-resolution programs may be beneficial. Indeed, patients treated with dexamethasone, a corticosteroid proved to upregulate SPM formation [73], [74], reduced plasma pro-inflammatory eicosanoids while increasing SPM concentration, along with upregulation of ALOX15, ALOX15B, ALOX12, GPR18 and GPR37 in circulating leukocyte subsets [67]. Exposure of PMN, monocytes and monocyte-derived MΦ to MCTR3, PCTR3, 17R-RvD3, RvD1 and RvD2 restored phagocytic ability of these cells and reprogrammed MΦ toward a pro-resolutive phenotype characterized by lowered production of pro-inflammatory cytokines [67], [75]. Along these lines, we recently reported that RvD1 and RvD2 treatment abated the inflammatory responses induced by SARS-CoV-2 virion spike 1 glycoprotein (S1) by dampening the release of IL-8 and TNF-α and modulating the expression of the inflammatory microRNAs (miRNA) miR-16, miR-29a, miR-223 and miR-125a [75]. These effects are of paramount importance, since the imbalanced pro-inflammatory MΦ-derived cytokine storm may cause severe pulmonary edema, acute respiratory distress, and multi-organ failure [76]. Thus, by broadly inhibiting proinflammatory cytokine production by MΦ and other cells, SPM proved valuable as potential therapeutics to limit SARS-CoV-2-induced inflammation [77]. Finally, since SPM reduce NETosis (e.g., RvD4, RvD1, RvT1, RvT2, RvT3, RvT4) [78], [79], [80], they can also have roles in reducing the severity of COVID-19. 3.4 Herpes viruses Severe infections with ocular Herpes simplex virus (HSV) can lead to scarring of the cornea or blindness mainly due to a chronic inflammatory reaction within cornea [81]. Thus, stimulation of pro-resolution pathways could be an attractive strategy to reduce the incidence of eye defects. Topical treatment with RvE1 of HSV-induced ocular disease reduced PMN and pathological CD4 T cells infiltration, levels of pro-inflammatory cytokines such as IFN-g, IL-6, KC while increasing anti-inflammatory IL-10 in corneas of infected mice [82]. Similar findings were also reported in eyes of HSV-infected mice treated neuro PD1 [83] and AT-RvD1 [84]. These results highlight that SPM could be harnessed as novel approach to control virally-induced immunopathological disease in the eye. 3.5 Kaposi’s Sarcoma-Associated Herpesvirus The Kaposi sarcoma herpesvirus (KSHV) is the causative agents of Kaposi sarcoma, a form of multicentric Castleman disease, and primary effusion lymphoma [85]. In vitro experiments showed that exposure to LXA4 of KSHV positive cell lines or de novo KSHV infected cells not only reduced levels of key pro-inflammatory mediators (PGE2, LTB4, IL-6, IL-8) [86] but also critically impact on reactivation from latency of dormant KSHV. Indeed, LXA4 physically interact with chromatin-remodeling proteins finally leading to viral gene lytic replication and viral progression. These events, together with the decreased expression of the immunomodulatory PD-L1 protein triggered by LXA4 in infected cells, should unleash cellular immunity against active KSHV [87]. 4 Summary and Future Directions Considerable research effort has been made to decipher the underlying mechanisms of active resolution of inflammation. It is becoming clear that failure in specific resolution pathways can contribute to a worse clinical outcome of viral infectious diseases. Therefore, harnessing endogenous proresolution mechanisms is gaining traction as a new therapeutic approach to treating viral diseases given their proresolving actions ( Fig. 4). Conventional anti-inflammatory strategies stop the inception phase of inflammation by inhibiting prostaglandin and/or leukotrienes biosynthesis. However, in viral diseases, this approach may undermine the beneficial effects of inflammation to restrain viral diffusion, lead to immune suppression, or delay resolution. SPM proved to enhance host defenses and lower threshold for antibiotic therapies in bacterial infections [88], [89]. Their roles in virus-mediated infections are of timely paramount importance in view of possible future outbreaks caused by highly pathogenic viruses (e.g., new SARS variants, Ebola and Crimean-Congo hemorragic fever viruses, and zoonotic Nipah viruses) that under surveillance by the WHO [90]. The latest COVID-19 pandemics has shown our unpreparedness to face viruses that had no vaccines or therapeutics available to regulate host immunity. As a result COVID-19 has claimed ~ 7,000,000 human lives worldwide [91]. Further studies on how viruses hijack SPM production, as well as on SPM functions will contribute towards understanding the pathogenesis of viral diseases and finding new ways to encompass resolution of inflammation to protect human health.Fig. 4 General bioactions of SPM in virus-driven infectious diseases. Shown here are main effects of SPM demonstrated in vitro and in vivo. See text for further description and references. Fig. 4 Acknowledgements The support of the Cystic Fibrosis Foundation (Grant RECCHIG1) to A.R is gratefully acknowledged. ==== Refs References 1 Medzhitov R. Inflammation 2010: new adventures of an old flame Cell. 140 2010 771 776 20303867 2 Majno G. Joris I. Cells, tissues, and disease: principles of general pathology 2004 Oxford University Press 3 Gudiol C. Bodro M. Simonetti A. Tubau F. González-Barca E. Cisnal M. Domingo-Domenech E. Jiménez L. Carratalà J. Changing aetiology, clinical features, antimicrobial resistance, and outcomes of bloodstream infection in neutropenic cancer patients Clin Microbiol Infect 19 2013 474 479 10.1111/j.1469-0691.2012.03879.x 22524597 4 Serhan C.N. Savill J. Resolution of inflammation: the beginning programs the end Nat Immunol 6 2005 1191 1197 10.1038/ni1276 16369558 5 Mayer A. Concerning the mosaic disease of tobbacco. Über die Mosaikkrankheit des Tabaks Die Landwirtschaftliche Versuchsstationen 32 1886 451 467 6 Iwanowski D.I. Über die Mosaikkrankheit der Tabakspflanze Bulletin de l’Académie Impériale Des Sciences de St.-Pétersbourg 35 1892 67 70 7 Beijerinck M. Concerning a contagium viwm fluidum as cause of the spot disease of tobacco leaves Phytopathol Class 7 1898 33 52 8 Stanley W.M. Crystalline Tobacco-Mosaic Virus Protein American Journal of Botany 24 1937 59 68 10.2307/2436720 9 Lustig A. Levine A. One hundred years of virology J Virol 66 1992 4629 4631 10.1128/jvi.66.8.4629-4631.1992 1629947 10 Farag N.S. Breitinger U. Breitinger H.G. Azizi M.A.El Viroporins and inflammasomes: A key to understand virus-induced inflammation The International Journal of Biochemistry & Cell Biology 122 2020 105738 10.1016/j.biocel.2020.105738 11 Iwasaki A. Pillai P.S. Innate immunity to influenza virus infection Nat Rev Immunol 14 2014 315 328 10.1038/nri3665 24762827 12 Merad M. Blish C.A. Sallusto F. Iwasaki A. The immunology and immunopathology of COVID-19 Science 375 2022 1122 1127 10.1126/science.abm8108 35271343 13 Ortega-Gómez A. Perretti M. Soehnlein O. Resolution of inflammation: an integrated view EMBO Molecular Medicine 5 2013 661 674 10.1002/emmm.201202382 23592557 14 Bannenberg G.L. Chiang N. Ariel A. Arita M. Tjonahen E. Gotlinger K.H. Hong S. Serhan C.N. Molecular circuits of resolution: formation and actions of resolvins and protectins J Immunol 174 2005 4345 4355 10.4049/jimmunol.174.7.4345 15778399 15 Levy B.D. Clish C.B. Schmidt B. Gronert K. Serhan C.N. Lipid mediator class switching during acute inflammation: signals in resolution Nature Immunology 2 2001 612 619 10.1038/89759 11429545 16 Serhan C.N. Hamberg M. Samuelsson B. Trihydroxytetraenes: a novel series of compounds formed from arachidonic acid in human leukocytes Biochem Biophys Res Commun 118 1984 943 949 10.1016/0006-291x(84)91486-4 6422933 17 Serhan C.N. Hamberg M. Samuelsson B. Lipoxins: novel series of biologically active compounds formed from arachidonic acid in human leukocytes Proc Natl Acad Sci U S A 81 1984 5335 5339 10.1073/pnas.81.17.5335 6089195 18 Serhan C.N. Hamberg M. Samuelsson B. Morris J. Wishka D.G. On the stereochemistry and biosynthesis of lipoxin B Proc Natl Acad Sci U S A 83 1986 1983 1987 10.1073/pnas.83.7.1983 3083410 19 Romano M. Serhan C.N. Lipoxin generation by permeabilized human platelets Biochemistry. 31 1992 8269 8277 1381961 20 Romano M. Chen X.S. Takahashi Y. Yamamoto S. Funk C.D. Serhan C.N. Lipoxin synthase activity of human platelet 12-lipoxygenase Biochem J 296 Pt 1 1993 127 133 8250832 21 Aspirin triggers previously undescribed bioactive eicosanoids by human endothelial cell-leukocyte interactions., (n.d.). https://doi.org/10.1073/pnas.92.21.9475. 22 Clària J. Lee M.H. Serhan C.N. Aspirin-triggered lipoxins (15-epi-LX) are generated by the human lung adenocarcinoma cell line (A549)-neutrophil interactions and are potent inhibitors of cell proliferation Mol Med 2 1996 583 596 8898374 23 Serhan C.N. Chiang N. Resolvins and cysteinyl-containing pro-resolving mediators activate resolution of infectious inflammation and tissue regeneration Prostaglandins & Other Lipid Mediators 166 2023 106718 10.1016/j.prostaglandins.2023.106718 24 Serhan C.N. Clish C.B. Brannon J. Colgan S.P. Chiang N. Gronert K. Novel Functional Sets of Lipid-Derived Mediators with Antiinflammatory Actions Generated from Omega-3 Fatty Acids via Cyclooxygenase 2–Nonsteroidal Antiinflammatory Drugs and Transcellular Processing Journal of Experimental Medicine 192 2000 1197 1204 10.1084/jem.192.8.1197 11034610 25 Serhan C.N. Treating inflammation and infection in the 21st century: new hints from decoding resolution mediators and mechanisms FASEB j 31 2017 1273 1288 10.1096/fj.201601222R 28087575 26 Serhan C.N. Hong S. Gronert K. Colgan S.P. Devchand P.R. Mirick G. Moussignac R.-L. Resolvins Journal of Experimental Medicine 196 2002 1025 1037 10.1084/jem.20020760 12391014 27 Hong S. Gronert K. Devchand P.R. Moussignac R.-L. Serhan C.N. Novel Docosatrienes and 17S-Resolvins Generated from Docosahexaenoic Acid in Murine Brain, Human Blood, and Glial Cells Journal of Biological Chemistry 278 2003 14677 14687 10.1074/jbc.M300218200 12590139 28 Dalli J. Zhu M. Vlasenko N.A. Deng B. Haeggström J.Z. Petasis N.A. Serhan C.N. The novel 13 S,14 S ‐epoxy‐maresin is converted by human macrophages to maresin 1 (MaR1), inhibits leukotriene A 4 hydrolase (LTA 4 H), and shifts macrophage phenotype FASEB j 27 2013 2573 2583 10.1096/fj.13-227728 23504711 29 Serhan C.N. Yang R. Martinod K. Kasuga K. Pillai P.S. Porter T.F. Oh S.F. Spite M. Maresins: novel macrophage mediators with potent antiinflammatory and proresolving actions Journal of Experimental Medicine 206 2009 15 23 10.1084/jem.20081880 19103881 30 Dalli J. Chiang N. Serhan C.N. Identification of 14-series sulfido-conjugated mediators that promote resolution of infection and organ protection Proc. Natl. Acad. Sci. U.S.A 111 2014 10.1073/pnas.1415006111 31 Dalli J. Ramon S. Norris P.C. Colas R.A. Serhan C.N. Novel proresolving and tissue-regenerative resolvin and protectin sulfido-conjugated pathways FASEB Journal: Official Publication of the Federation of American Societies for Experimental Biology 29 2015 2120 2136 10.1096/fj.14-268441 25713027 32 Levy B.D. Abdulnour R.-E.E. Tavares A. Brüggemann T.R. Norris P.C. Bai Y. Ai X. Serhan C.N. Cysteinyl maresins regulate the prophlogistic lung actions of cysteinyl leukotrienes J Allergy Clin Immunol 145 2020 335 344 10.1016/j.jaci.2019.09.028 31622689 33 Säfholm J. Abma W. Bankova L.G. Boyce J.A. Al-Ameri M. Orre A.-C. Wheelock C.E. Dahlén S.-E. Adner M. Cysteinyl-maresin 3 inhibits IL-13 induced airway hyperresponsiveness through alternative activation of the CysLT1 receptor Eur J Pharmacol 934 2022 175257 10.1016/j.ejphar.2022.175257 34 Pistorius K. Ly L. Souza P.R. Gomez E.A. Koenis D.S. Rodriguez A.R. Foster J. Sosabowski J. Hopkinson M. Rajeeve V. Spur B.W. Pitsillides A. Pitzalis C. Dalli J. MCTR3 reprograms arthritic monocytes to upregulate Arginase-1 and exert pro-resolving and tissue-protective functions in experimental arthritis EBioMedicine 79 2022 103974 10.1016/j.ebiom.2022.103974 35 Chiang N. de la Rosa X. Libreros S. Pan H. Dreyfuss J.M. Serhan C.N. Cysteinyl-specialized proresolving mediators link resolution of infectious inflammation and tissue regeneration via TRAF3 activation Proc Natl Acad Sci U S A 118 2021 e2013374118 10.1073/pnas.2013374118 36 Chiang N. Serhan C.N. Specialized pro-resolving mediator network: an update on production and actions Essays in Biochemistry 64 2020 443 462 10.1042/EBC20200018 32885825 37 Romano M. Patruno S. Pomilio A. Recchiuti A. Proresolving Lipid Mediators and Receptors in Stem Cell Biology: Concise Review Stem Cells Transl Med 8 2019 992 998 10.1002/sctm.19-0078 31187940 38 Dyall S.C. Balas L. Bazan N.G. Brenna J.T. Chiang N. da Costa Souza F. Dalli J. Durand T. Galano J.-M. Lein P.J. Serhan C.N. Taha A.Y. Polyunsaturated fatty acids and fatty acid-derived lipid mediators: Recent advances in the understanding of their biosynthesis, structures, and functions Prog Lipid Res 86 2022 101165 10.1016/j.plipres.2022.101165 39 Goulding N.J. Godolphin J.L. Sharland P.R. Peers S.H. Sampson M. Maddison P.J. Flower R.J. Anti-inflammatory lipocortin 1 production by peripheral blood leucocytes in response to hydrocortisone Lancet. 335 1990 1416 1418 10.1016/0140-6736(90)91445-g 1972208 40 Perretti M. Flower R.J. Modulation of IL-1-induced neutrophil migration by dexamethasone and lipocortin 1 J Immunol 150 1993 992 999 8423349 41 Flower R.J. Blackwell G.J. Anti-inflammatory steroids induce biosynthesis of a phospholipase A2 inhibitor which prevents prostaglandin generation Nature 278 1979 456 459 10.1038/278456a0 450050 42 Blackwell G.J. Carnuccio R. Di Rosa M. Flower R.J. Parente L. Persico P. Macrocortin: a polypeptide causing the anti-phospholipase effect of glucocorticoids Nature 287 1980 147 149 10.1038/287147a0 6893620 43 Perretti M. Dalli J. Resolution Pharmacology: Focus on Pro-Resolving Annexin A1 and Lipid Mediators for Therapeutic Innovation in Inflammation Annu. Rev. Pharmacol. Toxicol. 63 2023 449 469 10.1146/annurev-pharmtox-051821-042743 36151051 44 Perretti M. Chiang N. La M. Fierro I.M. Marullo S. Getting S.J. Solito E. Serhan C.N. Endogenous lipid- and peptide-derived anti-inflammatory pathways generated with glucocorticoid and aspirin treatment activate the lipoxin A4 receptor Nat Med 8 2002 1296 1302 10.1038/nm786 12368905 45 Morita M. Kuba K. Ichikawa A. Nakayama M. Katahira J. Iwamoto R. Watanebe T. Sakabe S. Daidoji T. Nakamura S. Kadowaki A. Ohto T. Nakanishi H. Taguchi R. Nakaya T. Murakami M. Yoneda Y. Arai H. Kawaoka Y. Penninger J.M. Arita M. Imai Y. The Lipid Mediator Protectin D1 Inhibits Influenza Virus Replication and Improves Severe Influenza Cell 153 2013 112 125 10.1016/j.cell.2013.02.027 23477864 46 Cilloniz C. Pantin-Jackwood M.J. Ni C. Goodman A.G. Peng X. Proll S.C. Carter V.S. Rosenzweig E.R. Szretter K.J. Katz J.M. Korth M.J. Swayne D.E. Tumpey T.M. Katze M.G. Lethal Dissemination of H5N1 Influenza Virus Is Associated with Dysregulation of Inflammation and Lipoxin Signaling in a Mouse Model of Infection J Virol 84 2010 7613 7624 10.1128/JVI.00553-10 20504916 47 Kumar D. Michaels M.G. Morris M.I. Green M. Avery R.K. Liu C. Danziger-Isakov L. Stosor V. Estabrook M. Gantt S. Marr K.A. Martin S. Silveira F.P. Razonable R.R. Allen U.D. Levi M.E. Lyon G.M. Bell L.E. Huprikar S. Patel G. Gregg K.S. Pursell K. Helmersen D. Julian K.G. Shiley K. Bono B. Dharnidharka V.R. Alavi G. Kalpoe J.S. Shoham S. Reid G.E. Humar A. Outcomes from pandemic influenza A H1N1 infection in recipients of solid-organ transplants: a multicentre cohort study The Lancet Infectious Diseases 10 2010 521 526 10.1016/S1473-3099(10)70133-X 20620116 48 Imai Y. Role of omega-3 PUFA-derived mediators, the protectins, in influenza virus infection Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1851 2015 496 502 10.1016/j.bbalip.2015.01.006 25617737 49 Machado F.S. Johndrow J.E. Esper L. Dias A. Bafica A. Serhan C.N. Aliberti J. Anti-inflammatory actions of lipoxin A4 and aspirin-triggered lipoxin are SOCS-2 dependent Nat Med 12 2006 330 334 10.1038/nm1355 16415877 50 Arora S. Lim W. Bist P. Perumalsamy R. Lukman H.M. Li F. Welker L.B. Yan B. Sethi G. Tambyah P.A. Fairhurst A.-M. Alonso S. Lim L.H.K. Influenza A virus enhances its propagation through the modulation of Annexin-A1 dependent endosomal trafficking and apoptosis Cell Death Differ 23 2016 1243 1256 10.1038/cdd.2016.19 26943321 51 Ampomah P.B. Moraes L.A. Lukman H.M. Lim L.H.K. Formyl peptide receptor 2 is regulated by RNA mimics and viruses through an IFN‐β‐STAT3–dependent pathway FASEB j 32 2018 1468 1478 10.1096/fj.201700584RR 29127186 52 Costa V.V. Sugimoto M.A. Hubner J. Bonilha C.S. Queiroz-Junior C.M. Gonçalves-Pereira M.H. Chen J. Gobbetti T. Libanio Rodrigues G.O. Bambirra J.L. Passos I.B. Machado Lopes C.E. Moreira T.P. Bonjour K. Melo R.C. Oliveira M.A. Andrade M.V.M. Sousa L.P. Souza D.G. da H. Santiago C. Perretti M. Teixeira M.M. Targeting the Annexin A1-FPR2/ALX pathway for host-directed therapy in dengue disease ELife 11 2022 e73853 10.7554/eLife.73853 53 Wang H. Anthony D. Yatmaz S. Wijburg O. Satzke C. Levy B. Vlahos R. Bozinovski S. Aspirin-triggered resolvin D1 reduces pneumococcal lung infection and inflammation in a viral and bacterial coinfection pneumonia model Clinical Science 131 2017 2347 2362 10.1042/CS20171006 28779028 54 Ramon S. Baker S.F. Sahler J.M. Kim N. Feldsott E.A. Serhan C.N. Martínez-Sobrido L. Topham D.J. Phipps R.P. The Specialized Proresolving Mediator 17-HDHA Enhances the Antibody-Mediated Immune Response against Influenza Virus: A New Class of Adjuvant? The Journal of Immunology 193 2014 6031 6040 10.4049/jimmunol.1302795 25392529 55 Kim N. Lannan K.L. Thatcher T.H. Pollock S.J. Woeller C.F. Phipps R.P. Lipoxin B4 Enhances Human Memory B Cell Antibody Production via Upregulating Cyclooxygenase-2 Expression The Journal of Immunology 201 2018 3343 3351 10.4049/jimmunol.1700503 30348736 56 Tavares L.P. Brüggemann T.R. Rezende R.M. Machado M.G. Cagnina R.E. Shay A.E. Garcia C.C. Nijmeh J. Teixeira M.M. Levy B.D. Cysteinyl Maresins Reprogram Macrophages to Protect Mice from Streptococcus pneumoniae after Influenza A Virus Infection MBio 13 2022 e0126722 10.1128/mbio.01267-22 57 Welliver R.C. Review of epidemiology and clinical risk factors for severe respiratory syncytial virus (RSV) infection The Journal of Pediatrics 143 2003 112 117 10.1067/S0022-3476(03)00508-0 58 Shirey K.A. Pletneva L.M. Puche A.C. Keegan A.D. Prince G.A. Blanco J.C.G. Vogel S.N. Control of RSV-induced lung injury by alternatively activated macrophages is IL-4Rα-, TLR4-, and IFN-β-dependent Mucosal Immunology 3 2010 291 300 10.1038/mi.2010.6 20404812 59 Shirey K.A. Lai W. Pletneva L.M. Karp C.L. Divanovic S. Blanco J.C.G. Vogel S.N. Role of the lipoxygenase pathway in RSV-induced alternatively activated macrophages leading to resolution of lung pathology Mucosal Immunology 7 2014 549 557 10.1038/mi.2013.71 24064666 60 do D. de Freitas N. Marinho Franceschina C. Muller D. Hilario G.T. Gassen R.B. Fazolo T. de Lima Kaminski V. Bogo Chies J.A. Maito F. Antunes K.H. Zanin R.F. Rodrigues L.C. Jr Duarte de Souza A.P. RvD1 treatment during primary infection modulates memory response increasing viral load during respiratory viral reinfection Immunobiology 226 2021 152151 10.1016/j.imbio.2021.152151 61 Walker K.H. Krishnamoorthy N. Brüggemann T.R. Shay A.E. Serhan C.N. Levy B.D. Protectins PCTR1 and PD1 Reduce Viral Load and Lung Inflammation During Respiratory Syncytial Virus Infection in Mice Front. Immunol. 12 2021 704427 10.3389/fimmu.2021.704427 62 Huang C. Wang Y. Li X. Ren L. Zhao J. Hu Y. Zhang L. Fan G. Xu J. Gu X. Cheng Z. Yu T. Xia J. Wei Y. Wu W. Xie X. Yin W. Li H. Liu M. Xiao Y. Gao H. Guo L. Xie J. Wang G. Jiang R. Gao Z. Jin Q. Wang J. Cao B. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China The Lancet 395 2020 497 506 10.1016/S0140-6736(20)30183-5 63 Zhang Q. Bastard P. Liu Z. Le Pen J. Moncada-Velez M. Chen J. Ogishi M. Sabli I.K.D. Hodeib S. Korol C. Rosain J. Bilguvar K. Ye J. Bolze A. Bigio B. Yang R. Arias A.A. Zhou Q. Zhang Y. Onodi F. Korniotis S. Karpf L. Philippot Q. Chbihi M. Bonnet-Madin L. Dorgham K. Smith N. Schneider W.M. Razooky B.S. Hoffmann H.-H. Michailidis E. Moens L. Han J.E. Lorenzo L. Bizien L. Meade P. Neehus A.-L. Ugurbil A.C. Corneau A. Kerner G. Zhang P. Rapaport F. Seeleuthner Y. Manry J. Masson C. Schmitt Y. Schlüter A. Le Voyer T. Khan T. Li J. Fellay J. Roussel L. Shahrooei M. Alosaimi M.F. Mansouri D. Al-Saud H. Al-Mulla F. Almourfi F. Al-Muhsen S.Z. Alsohime F. Al Turki S. Hasanato R. van de Beek D. Biondi A. Bettini L.R. D’Angio’ M. Bonfanti P. Imberti L. Sottini A. Paghera S. Quiros-Roldan E. Rossi C. Oler A.J. Tompkins M.F. Alba C. Vandernoot I. Goffard J.-C. Smits G. Migeotte I. Haerynck F. Soler-Palacin P. Martin-Nalda A. Colobran R. Morange P.-E. Keles S. Çölkesen F. Ozcelik T. Yasar K.K. Senoglu S. Karabela Ş.N. Rodríguez-Gallego C. Novelli G. Hraiech S. Tandjaoui-Lambiotte Y. Duval X. Laouénan C. COVID-STORM Clinicians, COVID Clinicians, Imagine COVID GroupFrench COVID Cohort Study GroupCoV-Contact Cohort, Amsterdam UMC Covid-19 Biobank, COVID Human Genetic Effort, NIAID-USUHS/TAGC COVID Immunity GroupSnow A.L. Dalgard C.L. Milner J.D. Vinh D.C. Mogensen T.H. Marr N. Spaan A.N. Boisson B. Boisson-Dupuis S. Bustamante J. Puel A. Ciancanelli M.J. Meyts I. Maniatis T. Soumelis V. Amara A. Nussenzweig M. García-Sastre A. Krammer F. Pujol A. Duffy D. Lifton R.P. Zhang S.-Y. Gorochov G. Béziat V. Jouanguy E. Sancho-Shimizu V. Rice C.M. Abel L. Notarangelo L.D. Cobat A. Su H.C. Casanova J.-L. Foti G. Bellani G. Citerio G. Contro E. Pesci A. Valsecchi M.G. Cazzaniga M. Abad J. Aguilera-Albesa S. Akcan O.M. Darazam I.A. Aldave J.C. Ramos M.A. Nadji S.A. Alkan G. Allardet-Servent J. Allende L.M. Alsina L. Alyanakian M.-A. Amador-Borrero B. Amoura Z. Antolí A. Arslan S. Assant S. Auguet T. Azot A. Bajolle F. Baldolli A. Ballester M. Feldman H.B. Barrou B. Beurton A. Bilbao A. Blanchard-Rohner G. Blanco I. Blandinières A. Blazquez-Gamero D. Bloomfield M. Bolivar-Prados M. Borie R. Bosteels C. Bousfiha A.A. Bouvattier C. Boyarchuk O. Bueno M.R.P. Bustamante J. Cáceres Agra J.J. Calimli S. Capra R. Carrabba M. Casasnovas C. Caseris M. Castelle M. Castelli F. de Vera M.C. Castro M.V. Catherinot E. Chalumeau M. Charbit B. Cheng M.P. Clavé P. Clotet B. Codina A. Colkesen F. Çölkesen F. Colobran R. Comarmond C. Dalmau D. Darley D.R. Dauby N. Dauger S. de Pontual L. Dehban A. Delplancq G. Demoule A. Diehl J.-L. Dobbelaere S. Durand S. Eldars W. Elgamal M. Elnagdy M.H. Emiroglu M. Erdeniz E.H. Aytekin S.E. Euvrard R. Evcen R. Fabio G. Faivre L. Falck A. Fartoukh M. Faure M. Arquero M.F. Flores C. Francois B. Fumadó V. Fusco F. Solis B.G. Gaussem P. Gil-Herrera J. Gilardin L. Alarcon M.G. Girona-Alarcón M. Goffard J.-C. Gok F. González-Montelongo R. Guerder A. Gul Y. Guner S.N. Gut M. Hadjadj J. Haerynck F. Halwani R. Hammarström L. Hatipoglu N. Hernandez-Brito E. Heijmans C. Holanda-Peña M.S. Horcajada J.P. Hoste L. Hoste E. Hraiech S. Humbert L. Iglesias A.D. Íñigo-Campos A. Jamme M. Arranz M.J. Jordan I. Jorens P. Kanat F. Kapakli H. Kara I. Karbuz A. Yasar K.K. Keles S. Demirkol Y.K. Klocperk A. Król Z.J. Kuentz P. Kwan Y.W.M. Lagier J.-C. Lambrecht B.N. Lau Y.-L. Le Bourgeois F. Leo Y.-S. Lopez R.L. Leung D. Levin M. Levy M. Lévy R. Li Z. Linglart A. Loeys B. Lorenzo-Salazar J.M. Louapre C. Lubetzki C. Luyt C.-E. Lye D.C. Mansouri D. Marjani M. Pereira J.M. Martin A. Pueyo D.M. Martinez-Picado J. Marzana I. Mathian A. Matos L.R.B. Matthews G.V. Mayaux J. Mège J.-L. Melki I. Meritet J.-F. Metin O. Meyts I. Mezidi M. Migeotte I. Millereux M. Mirault T. Mircher C. Mirsaeidi M. Melián A.M. Martinez A.M. Morange P. Mordacq C. Morelle G. Mouly S. Muñoz-Barrera A. Naesens L. Nafati C. Neves J.F. Ng L.F.P. Medina Y.N. Cuadros E.N. Ocejo-Vinyals J.G. Orbak Z. Oualha M. Özçelik T. Pan-Hammarström Q. Parizot C. Pascreau T. Paz-Artal E. Pellegrini S. de Diego R.P. Philippe A. Philippot Q. Planas-Serra L. Ploin D. Poissy J. Poncelet G. Pouletty M. Quentric P. Raoult D. Rebillat A.-S. Reisli I. Ricart P. Richard J.-C. Rivet N. Rivière J.G. Blanch G.R. Rodrigo C. Rodriguez-Gallego C. Rodríguez-Palmero A. Romero C.S. Rothenbuhler A. Rozenberg F. Ruiz del Prado M.Y. Riera J.S. Sanchez O. Sánchez-Ramón S. Schluter A. Schmidt M. Schweitzer C.E. Scolari F. Sediva A. Seijo L.M. Sene D. Senoglu S. Seppänen M.R.J. Ilovich A.S. Shahrooei M. Slabbynck H. Smadja D.M. Sobh A. Moreno X.S. Solé-Violán J. Soler C. Soler-Palacín P. Stepanovskiy Y. Stoclin A. Taccone F. Tandjaoui-Lambiotte Y. Taupin J.-L. Tavernier S.J. Terrier B. Thumerelle C. Tomasoni G. Toubiana J. Alvarez J.T. Trouillet-Assant S. Troya J. Tucci A. Ursini M.V. Uzunhan Y. Vabres P. Valencia-Ramos J. Van Braeckel E. Van de Velde S. Van Den Rym A.M. Van Praet J. Vandernoot I. Vatansev H. Vélez-Santamaria V. Viel S. Vilain C. Vilaire M.E. Vincent A. Voiriot G. Vuotto F. Yosunkaya A. Young B.E. Yucel F. Zannad F. Zatz M. Belot A. Bole-Feysot C. Lyonnet S. Masson C. Nitschke P. Pouliet A. Schmitt Y. Tores F. Zarhrate M. Abel L. Andrejak C. Angoulvant F. Bachelet D. Basmaci R. Behillil S. Beluze M. Benkerrou D. Bhavsar K. Bompart F. Bouadma L. Bouscambert M. Caralp M. Cervantes-Gonzalez M. Chair A. Coelho A. Couffignal C. Couffin-Cadiergues S. D’Ortenzio E. Da Silveira C. Debray M.-P. Deplanque D. Descamps D. Desvallées M. Diallo A. Diouf A. Dorival C. Dubos F. Duval X. Eloy P. Enouf V.V. Esperou H. Esposito-Farese M. Etienne M. Ettalhaoui N. Gault N. Gaymard A. Ghosn J. Gigante T. Gorenne I. Guedj J. Hoctin A. Hoffmann I. Jaafoura S. Kafif O. Kaguelidou F. Kali S. Khalil A. Khan C. Laouénan C. Laribi S. Le M. Le Hingrat Q. Le Mestre S. Le Nagard H. Lescure F.-X. Lévy Y. Levy-Marchal C. Lina B. Lingas G. Lucet J.C. Malvy D. Mambert M. Mentré F. Mercier N. Meziane A. Mouquet H. Mullaert J. Neant N. Noret M. Pages J. Papadopoulos A. Paul C. Peiffer-Smadja N. Petrov-Sanchez V. Peytavin G. Picone O. Puéchal O. Rosa-Calatrava M. Rossignol B. Rossignol P. Roy C. Schneider M. Semaille C. Mohammed N.S. Tagherset L. Tardivon C. Tellier M.-C. Téoulé F. Terrier O. Timsit J.-F. Trioux T. Tual C. Tubiana S. van der Werf S. Vanel N. Veislinger A. Visseaux B. Wiedemann A. Yazdanpanah Y. Alavoine L. Amat K.K.A. Behillil S. Bielicki J. Bruijning P. Burdet C. Caumes E. Charpentier C. Coignard B. Costa Y. Couffin-Cadiergues S. Damond F. Dechanet A. Delmas C. Descamps D. Duval X. Ecobichon J.-L. Enouf V. Espérou H. Frezouls W. Houhou N. Ilic-Habensus E. Kafif O. Kikoine J. Le Hingrat Q. Lebeaux D. Leclercq A. Lehacaut J. Letrou S. Lina B. Lucet J.-C. Malvy D. Manchon P. Mandic M. Meghadecha M. Motiejunaite J. Nouroudine M. Piquard V. Postolache A. Quintin C. Rexach J. Roufai L. Terzian Z. Thy M. Tubiana S. van der Werf S. Vignali V. Visseaux B. Yazdanpanah Y. van Agtmael M. Algera A.G. van Baarle F. Bax D. Beudel M. Bogaard H.J. Bomers M. Bos L. Botta M. de Brabander J. de Bree G. Brouwer M.C. de Bruin S. Bugiani M. Bulle E. Chouchane O. Cloherty A. Elbers P. Fleuren L. Geerlings S. Geerts B. Geijtenbeek T. Girbes A. Goorhuis B. Grobusch M.P. Hafkamp F. Hagens L. Hamann J. Harris V. Hemke R. Hermans S.M. Heunks L. Hollmann M.W. Horn J. Hovius J.W. de Jong M.D. Koning R. van Mourik N. Nellen J. Paulus F. Peters E. van der Poll T. Preckel B. Prins J.M. Raasveld J. Reijnders T. Schinkel M. Schultz M.J. Schuurman A. Sigaloff K. Smit M. Stijnis C.S. Stilma W. Teunissen C. Thoral P. Tsonas A. van der Valk M. Veelo D. Vlaar A.P.J. de Vries H. van Vugt M. Wiersinga W.J. Wouters D. (Koos) Zwinderman A.H. van de Beek D. Abel L. Aiuti A. Al Muhsen S. Al-Mulla F. Anderson M.S. Arias A.A. Feldman H.B. Bogunovic D. Bolze A. Bondarenko A. Bousfiha A.A. Brodin P. Bryceson Y. Bustamante C.D. Butte M. Casari G. Chakravorty S. Christodoulou J. Cirulli E. Condino-Neto A. Cooper M.A. Dalgard C.L. David A. DeRisi J.L. Desai M. Drolet B.A. Espinosa S. Fellay J. Flores C. Franco J.L. Gregersen P.K. Haerynck F. Hagin D. Halwani R. Heath J. Henrickson S.E. Hsieh E. Imai K. Itan Y. Karamitros T. Kisand K. Ku C.-L. Lau Y.-L. Ling Y. Lucas C.L. Maniatis T. Mansouri D. Marodi L. Meyts I. Milner J. Mironska K. Mogensen T. Morio T. Ng L.F.P. Notarangelo L.D. Novelli A. Novelli G. O’Farrelly C. Okada S. Ozcelik T. de Diego R.P. Planas A.M. Prando C. Pujol A. Quintana-Murci L. Renia L. Renieri A. Rodríguez-Gallego C. Sancho-Shimizu V. Sankaran V. Barrett K.S. Shahrooei M. Snow A. Soler-Palacín P. Spaan A.N. Tangye S. Turvey S. Uddin F. Uddin M.J. van de Beek D. Vazquez S.E. Vinh D.C. von Bernuth H. Washington N. Zawadzki P. Su H.C. Casanova J.-L. Jing H. Tung W. Luthers C.R. Bauman B.M. Shafer S. Zheng L. Zhang Z. Kubo S. Chauvin S.D. Meguro K. Shaw E. Lenardo M. Lack J. Karlins E. Hupalo D.M. Rosenberger J. Sukumar G. Wilkerson M.D. Zhang X. Inborn errors of type I IFN immunity in patients with life-threatening COVID-19 Science 370 2020 eabd4570 10.1126/science.abd4570 32972995 64 Bastard P. Rosen L.B. Zhang Q. Michailidis E. Hoffmann H.-H. Zhang Y. Dorgham K. Philippot Q. Rosain J. Béziat V. Manry J. Shaw E. Haljasmägi L. Peterson P. Lorenzo L. Bizien L. Trouillet-Assant S. Dobbs K. de Jesus A.A. Belot A. Kallaste A. Catherinot E. Tandjaoui-Lambiotte Y. Le Pen J. Kerner G. Bigio B. Seeleuthner Y. Yang R. Bolze A. Spaan A.N. Delmonte O.M. Abers M.S. Aiuti A. Casari G. Lampasona V. Piemonti L. Ciceri F. Bilguvar K. Lifton R.P. Vasse M. Smadja D.M. Migaud M. Hadjadj J. Terrier B. Duffy D. Quintana-Murci L. van de Beek D. Roussel L. Vinh D.C. Tangye S.G. Haerynck F. Dalmau D. Martinez-Picado J. Brodin P. Nussenzweig M.C. Boisson-Dupuis S. Rodríguez-Gallego C. Vogt G. Mogensen T.H. Oler A.J. Gu J. Burbelo P.D. Cohen J.I. Biondi A. Bettini L.R. D’Angio M. Bonfanti P. Rossignol P. Mayaux J. Rieux-Laucat F. Husebye E.S. Fusco F. Ursini M.V. Imberti L. Sottini A. Paghera S. Quiros-Roldan E. Rossi C. Castagnoli R. Montagna D. Licari A. Marseglia G.L. Duval X. Ghosn J. HGID Lab, NIAID-USUHS Immune Response to COVID GroupCOVID Clinicians, COVID-STORM Clinicians, Imagine COVID GroupFrench COVID Cohort Study Group, The Milieu Intérieur Consortium, CoV-Contact CohortAmsterdam UMC Covid-19 BiobankCOVID Human Genetic EffortTsang J.S. Goldbach-Mansky R. Kisand K. Lionakis M.S. Puel A. Zhang S.-Y. Holland S.M. Gorochov G. Jouanguy E. Rice C.M. Cobat A. Notarangelo L.D. Abel L. Su H.C. Casanova J.-L. Arias A.A. Boisson B. Boucherit S. Bustamante J. Chbihi M. Chen J. Chrabieh M. Kochetkov T. Le Voyer T. Liu D. Nemirovskaya Y. Ogishi M. Papandrea D. Patissier C. Rapaport F. Roynard M. Vladikine N. Woollett M. Zhang P. Kashyap A. Ding L. Bosticardo M. Wang Q. Ochoa S. Liu H. Chauvin S.D. Stack M. Koroleva G. Bansal N. Dalgard C.L. Snow A.L. Abad J. Aguilera-Albesa S. Akcan O.M. Darazam I.A. Aldave J.C. Ramos M.A. Nadji S.A. Alkan G. Allardet-Servent J. Allende L.M. Alsina L. Alyanakian M.-A. Amador-Borrero B. Amoura Z. Antolí A. Arslan S. Assant S. Auguet T. Azot A. Bajolle F. Baldolli A. Ballester M. Feldman H.B. Barrou B. Beurton A. Bilbao A. Blanchard-Rohner G. Blanco I. Blandinières A. Blazquez-Gamero D. Bloomfield M. Bolivar-Prados M. Borie R. Bousfiha A.A. Bouvattier C. Boyarchuk O. Bueno M.R.P. Bustamante J. Cáceres Agra J.J. Calimli S. Capra R. Carrabba M. Casasnovas C. Caseris M. Castelle M. Castelli F. de Vera M.C. Castro M.V. Catherinot E. Chalumeau M. Charbit B. Cheng M.P. Clavé P. Clotet B. Codina A. Colkesen F. Colkesen F. Colobran R. Comarmond C. Corsico A.G. Dalmau D. Darley D.R. Dauby N. Dauger S. de Pontual L. Dehban A. Delplancq G. Demoule A. Di Sabatino A. Diehl J.-L. Dobbelaere S. Durand S. Eldars W. Elgamal M. Elnagdy M.H. Emiroglu M. Erdeniz E.H. Aytekin S.E. Euvrard R. Evcen R. Fabio G. Faivre L. Falck A. Fartoukh M. Faure M. Arquero M.F. Flores C. Francois B. Fumadó V. Fusco F. Solis B.G. Gaussem P. Gil-Herrera J. Gilardin L. Alarcon M.G. Girona-Alarcón M. Goffard J.-C. Gok F. González-Montelongo R. Guerder A. Gul Y. Guner S.N. Gut M. Hadjadj J. Haerynck F. Halwani R. Hammarström L. Hatipoglu N. Hernandez-Brito E. Holanda-Peña M.S. Horcajada J.P. Hraiech S. Humbert L. Iglesias A.D. Íñigo-Campos A. Jamme M. Arranz M.J. Jordan I. Kanat F. Kapakli H. Kara I. Karbuz A. Yasar K.K. Keles S. Demirkol Y.K. Klocperk A. Król Z.J. Kuentz P. Kwan Y.W.M. Lagier J.-C. Lau Y.-L. Le Bourgeois F. Leo Y.-S. Lopez R.L. Leung D. Levin M. Levy M. Lévy R. Li Z. Linglart A. Lorenzo-Salazar J.M. Louapre C. Lubetzki C. Luyt C.-E. Lye D.C. Mansouri D. Marjani M. Pereira J.M. Martin A. Pueyo D.M. Martinez-Picado J. Marzana I. Mathian A. Matos L.R.B. Matthews G.V. Mayaux J. Mège J.-L. Melki I. Meritet J.-F. Metin O. Meyts I. Mezidi M. Migeotte I. Millereux M. Mirault T. Mircher C. Mirsaeidi M. Melián A.M. Martinez A.M. Morange P. Mordacq C. Morelle G. Mouly S. Muñoz-Barrera A. Nafati C. Neves J.F. Ng L.F.P. Medina Y.N. Cuadros E.N. Ocejo-Vinyals J.G. Orbak Z. Oualha M. Özçelik T. Hammarström Q.P. Parizot C. Pascreau T. Paz-Artal E. de Diego R.P. Philippe A. Philippot Q. Planas-Serra L. Ploin D. Poissy J. Poncelet G. Pouletty M. Quentric P. Raoult D. Rebillat A.-S. Reisli I. Ricart P. Richard J.-C. Rivet N. Rivière J.G. Blanch G.R. Rodrigo C. Rodriguez-Gallego C. Rodríguez-Palmero A. Romero C.S. Rothenbuhler A. Rozenberg F. Ruiz del Prado M.Y. Riera J.S. Sanchez O. Sánchez-Ramón S. Schluter A. Schmidt M. Schweitzer C.E. Scolari F. Sediva A. Seijo L.M. Sene D. Senoglu S. Seppänen M.R.J. Ilovich A.S. Shahrooei M. Smadja D. Sobh A. Moreno X.S. Solé-Violán J. Soler C. Soler-Palacín P. Stepanovskiy Y. Stoclin A. Taccone F. Tandjaoui-Lambiotte Y. Taupin J.-L. Tavernier S.J. Terrier B. Thumerelle C. Tomasoni G. Toubiana J. Alvarez J.T. Trouillet-Assant S. Troya J. Tucci A. Ursini M.V. Uzunhan Y. Vabres P. Valencia-Ramos J. Van Den Rym A.M. Vandernoot I. Vatansev H. Vélez-Santamaria V. Viel S. Vilain C. Vilaire M.E. Vincent A. Voiriot G. Vuotto F. Yosunkaya A. Young B.E. Yucel F. Zannad F. Zatz M. Belot A. Foti G. Bellani G. Citerio G. Contro E. Pesci A. Valsecchi M.G. Cazzaniga M. Bole-Feysot C. Lyonnet S. Masson C. Nitschke P. Pouliet A. Schmitt Y. Tores F. Zarhrate M. Abel L. Andrejak C. Angoulvant F. Bachelet D. Basmaci R. Behillil S. Beluze M. Benkerrou D. Bhavsar K. Bompart F. Bouadma L. Bouscambert M. Caralp M. Cervantes-Gonzalez M. Chair A. Coelho A. Couffignal C. Couffin-Cadiergues S. D’ortenzio E. Da Silveira C. Debray M.-P. Deplanque D. Descamps D. Desvallées M. Diallo A. Diouf A. Dorival C. Dubos F. Duval X. Eloy P. Enouf V.V.E. Esperou H. Esposito-Farese M. Etienne M. Ettalhaoui N. Gault N. Gaymard A. Ghosn J. Gigante T. Gorenne I. Guedj J. Hoctin A. Hoffmann I. Jaafoura S. Kafif O. Kaguelidou F. Kali S. Khalil A. Khan C. Laouénan C. Laribi S. Le M. Le Hingrat Q. Le Mestre S. Le Nagard H. Lescure F.-X. Lévy Y. Levy-Marchal C. Lina B. Lingas G. Lucet J.C. Malvy D. Mambert M. Mentré F. Mercier N. Meziane A. Mouquet H. Mullaert J. Neant N. Noret M. Pages J. Papadopoulos A. Paul C. Peiffer-Smadja N. Petrov-Sanchez V. Peytavin G. Picone O. Puéchal O. Rosa-Calatrava M. Rossignol B. Rossignol P. Roy C. Schneider M. Semaille C. Mohammed N.S. Tagherset L. Tardivon C. Tellier M.-C. Téoulé F. Terrier O. Timsit J.-F. Treoux T. Tual C. Tubiana S. van der Werf S. Vanel N. Veislinger A. Visseaux B. Wiedemann A. Yazdanpanah Y. Abel L. Alcover A. Aschard H. Astrom K. Bousso P. Bruhns P. Cumano A. Demangel C. Deriano L. Di Santo J. Dromer F. Eberl G. Enninga J. Fellay J. Gomperts-Boneca I. Hasan M. Hercberg S. Lantz O. Mouquet H. Patin E. Pellegrini S. Pol S. Rausell A. Rogge L. Sakuntabhai A. Schwartz O. Schwikowski B. Shorte S. Tangy F. Toubert A. Touvier M. Ungeheuer M.-N. Albert M.L. Duffy D. Quintana-Murci L. Alavoine L. Amat K.K.A. Behillil S. Bielicki J. Bruijning P. Burdet C. Caumes E. Charpentier C. Coignard B. Costa Y. Couffin-Cadiergues S. Damond F. Dechanet A. Delmas C. Descamps D. Duval X. Ecobichon J.-L. Enouf V. Espérou H. Frezouls W. Houhou N. Ilic-Habensus E. Kafif O. Kikoine J. Le Hingrat Q. Lebeaux D. Leclercq A. Lehacaut J. Letrou S. Lina B. Lucet J.-C. Malvy D. Manchon P. Mandic M. Meghadecha M. Motiejunaite J. Nouroudine M. Piquard V. Postolache A. Quintin C. Rexach J. Roufai L. Terzian Z. Thy M. Tubiana S. van der Werf S. Vignali V. Visseaux B. Yazdanpanah Y. van Agtmael M. Algera A.G. van Baarle F. Bax D. Beudel M. Bogaard H.J. Bomers M. Bos L. Botta M. de Brabander J. Bree G. Brouwer M.C. de Bruin S. Bugiani M. Bulle E. Chouchane O. Cloherty A. Elbers P. Fleuren L. Geerlings S. Geerts B. Geijtenbeek T. Girbes A. Goorhuis B. Grobusch M.P. Hafkamp F. Hagens L. Hamann J. Harris V. Hemke R. Hermans S.M. Heunks L. Hollmann M.W. Horn J. Hovius J.W. de Jong M.D. Koning R. van Mourik N. Nellen J. Paulus F. Peters E. van der Poll T. Preckel B. Prins J.M. Raasveld J. Reijnders T. Schinkel M. Schultz M.J. Schuurman A. Sigaloff K. Smit M. Stijnis C.S. Stilma W. Teunissen C. Thoral P. Tsonas A. van der Valk M. Veelo D. Vlaar A.P.J. de Vries H. van Vugt M. Wiersinga W.J. Wouters D. (Koos) Zwinderman A.H. van de Beek D. Abel L. Aiuti A. Al Muhsen S. Al-Mulla F. Anderson M.S. Arias A.A. Feldman H.B. Bogunovic D. Bolze A. Bondarenko A. Bousfiha A.A. Brodin P. Bryceson Y. Bustamante C.D. Butte M. Casari G. Chakravorty S. Christodoulou J. Cirulli E. Condino-Neto A. Cooper M.A. Dalgard C.L. DeRisi J.L. Desai M. Drolet B.A. Espinosa S. Fellay J. Flores C. Franco J.L. Gregersen P.K. Haerynck F. Hagin D. Halwani R. Heath J. Henrickson S.E. Hsieh E. Imai K. Itan Y. Karamitros T. Kisand K. Ku C.-L. Lau Y.-L. Ling Y. Lucas C.L. Maniatis T. Mansouri D. Marodi L. Meyts I. Milner J.D. Mironska K. Mogensen T. Morio T. Ng L.F.P. Notarangelo L.D. Novelli G. Novelli A. O’Farrelly C. Okada S. Ozcelik T. de Diego R.P. Planas A.M. Prando C. Pujol A. Quintana-Murci L. Renia L. Renieri A. Rodríguez-Gallego C. Sancho-Shimizu V. Sankaran V. Barrett K.S. Shahrooei M. Snow A. Soler-Palacín P. Spaan A.N. Tangye S. Turvey S. Uddin F. Uddin M.J. van de Beek D. Vazquez S.E. Vinh D.C. von Bernuth H. Washington N. Zawadzki P. Su H.C. Casanova J.-L. Autoantibodies against type I IFNs in patients with life-threatening COVID-19 Science 370 2020 eabd4585 10.1126/science.abd4585 65 Blanco-Melo D. Nilsson-Payant B.E. Liu W.-C. Uhl S. Hoagland D. Møller R. Jordan T.X. Oishi K. Panis M. Sachs D. Wang T.T. Schwartz R.E. Lim J.K. Albrecht R.A. tenOever B.R. Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19 e9 Cell 181 2020 1036 1045 10.1016/j.cell.2020.04.026 32416070 66 Turnbull J. Jha R.R. Ortori C.A. Lunt E. Tighe P.J. Irving W.L. Gohir S.A. Kim D.-H. Valdes A.M. Tarr A.W. Barrett D.A. Chapman V. Serum Levels of Proinflammatory Lipid Mediators and Specialized Proresolving Molecules Are Increased in Patients With Severe Acute Respiratory Syndrome Coronavirus 2 and Correlate With Markers of the Adaptive Immune Response The Journal of Infectious Diseases 225 2022 2142 2154 10.1093/infdis/jiab632 34979019 67 Koenis D.S. Beegun I. Jouvene C.C. Aguirre G.A. Souza P.R. Gonzalez-Nunez M. Ly L. Pistorius K. Kocher H.M. Ricketts W. Thomas G. Perretti M. Alusi G. Pfeffer P. Dalli J. Disrupted Resolution Mechanisms Favor Altered Phagocyte Responses in COVID-19 Circ Res 129 2021 10.1161/CIRCRESAHA.121.319142 68 Archambault A. Zaid Y. Rakotoarivelo V. Turcotte C. Doré É. Dubuc I. Martin C. Flamand O. Amar Y. Cheikh A. Fares H. El Hassani A. Tijani Y. Côté A. Laviolette M. Boilard É. Flamand L. Flamand N. High levels of eicosanoids and docosanoids in the lungs of intubated COVID‐19 patients The FASEB Journal 35 2021 10.1096/fj.202100540R 69 Wen W. Su W. Tang H. Le W. Zhang X. Zheng Y. Liu X. Xie L. Li J. Ye J. Dong L. Cui X. Miao Y. Wang D. Dong J. Xiao C. Chen W. Wang H. Immune cell profiling of COVID-19 patients in the recovery stageby single-cell sequencing Cell Discov 6 2020 31 10.1038/s41421-020-0168-9 32377375 70 Schwarz B. Sharma L. Roberts L. Peng X. Bermejo S. Leighton I. Casanovas-Massana A. Minasyan M. Farhadian S. Ko A.I. Yale IMPACT TeamDela Cruz C.S. Bosio C.M. Cutting Edge: Severe SARS-CoV-2 Infection in Humans Is Defined by a Shift in the Serum Lipidome, Resulting in Dysregulation of Eicosanoid Immune Mediators The Journal of Immunology 206 2021 329 334 10.4049/jimmunol.2001025 33277388 71 Palmas F. Clarke J. Colas R.A. Gomez E.A. Keogh A. Boylan M. McEvoy N. McElvaney O.J. McElvaney O. Alalqam R. McElvaney N.G. Curley G.F. Dalli J. Dysregulated plasma lipid mediator profiles in critically ill COVID-19 patients PLoS ONE 16 2021 e0256226 10.1371/journal.pone.0256226 72 Biagini D. Franzini M. Oliveri P. Lomonaco T. Ghimenti S. Bonini A. Vivaldi F. Macera L. Balas L. Durand T. Oger C. Galano J.-M. Maggi F. Celi A. Paolicchi A. Di Francesco F. MS-based targeted profiling of oxylipins in COVID-19: A new insight into inflammation regulation Free Radical Biology and Medicine 180 2022 236 243 10.1016/j.freeradbiomed.2022.01.021 35085774 73 Pyrillou K. Chairakaki A.-D. Tamvakopoulos C. Andreakos E. Dexamethasone induces ω3-derived immunoresolvents driving resolution of allergic airway inflammation e4 Journal of Allergy and Clinical Immunology 142 2018 691 695 10.1016/j.jaci.2018.04.004 29698628 74 Andreakos E. Papadaki M. Serhan C.N. Dexamethasone, pro‐resolving lipid mediators and resolution of inflammation in COVID‐19 Allergy 76 2021 626 628 10.1111/all.14595 32956495 75 Recchiuti A. Patruno S. Mattoscio D. Isopi E. Pomilio A. Lamolinara A. Iezzi M. Pecce R. Romano M. Resolvin D1 and D2 reduce SARS-CoV-2-induced inflammatory responses in cystic fibrosis macrophages FASEB J 35 2021 e21441 10.1096/fj.202001952R 76 Dockrell D.H. Russell C.D. McHugh B. Fraser R. Does autonomous macrophage-driven inflammation promote alveolar damage in COVID-19 Eur Respir J 60 2022 2201521 10.1183/13993003.01521-2022 77 Panigrahy D. Gilligan M.M. Huang S. Gartung A. Cortés-Puch I. Sime P.J. Phipps R.P. Serhan C.N. Hammock B.D. Inflammation resolution: a dual-pronged approach to averting cytokine storms in COVID-19? Cancer Metastasis Rev 39 2020 337 340 10.1007/s10555-020-09889-4 32385712 78 Cherpokova D. Jouvene C.C. Libreros S. DeRoo E.P. Chu L. de la Rosa X. Norris P.C. Wagner D.D. Serhan C.N. Resolvin D4 attenuates the severity of pathological thrombosis in mice Blood 134 2019 1458 1468 10.1182/blood.2018886317 31300403 79 Chiang N. Sakuma M. Rodriguez A.R. Spur B.W. Irimia D. Serhan C.N. Resolvin T-series reduce neutrophil extracellular traps Blood 139 2022 1222 1233 10.1182/blood.2021013422 34814186 80 Recchiuti A. Federti E. Matte A. Mazzi F. Ceolan J. Porreca A. Di Nicola M. Menotti S. Alivernini S. De Franceschi L. Impaired pro-resolving mechanisms promote abnormal NETosis, fueling autoimmunity in sickle cell disease Am J Hematol 98 2023 E45 E48 10.1002/ajh.26797 36540948 81 Liesegang T.J. Herpes Simplex Virus Epidemiology and Ocular Importance Cornea 20 2001 1 13 10.1097/00003226-200101000-00001 11188989 82 Rajasagi N.K. Reddy P.B.J. Suryawanshi A. Mulik S. Gjorstrup P. Rouse B.T. Controlling Herpes Simplex Virus-Induced Ocular Inflammatory Lesions with the Lipid-Derived Mediator Resolvin E1 The Journal of Immunology 186 2011 1735 1746 10.4049/jimmunol.1003456 21187448 83 Rajasagi N.K. Reddy P.B.J. Mulik S. Gjorstrup P. Rouse B.T. Neuroprotectin D1 Reduces the Severity of Herpes Simplex Virus–Induced Corneal Immunopathology Invest. Ophthalmol. Vis. Sci. 54 2013 6269 10.1167/iovs.13-12152 23942967 84 Rajasagi N.K. Bhela S. Varanasi S.K. Rouse B.T. Frontline Science: Aspirin-triggered resolvin D1 controls herpes simplex virus-induced corneal immunopathology Journal of Leukocyte Biology 102 2017 1159 1171 10.1189/jlb.3HI1216-511RR 28584076 85 Goncalves P.H. Ziegelbauer J. Uldrick T.S. Yarchoan R. Kaposi sarcoma herpesvirus-associated cancers and related diseases Current Opinion in HIV and AIDS 12 2017 47 56 10.1097/COH.0000000000000330 27662501 86 Marginean A. Sharma-Walia N. Lipoxins exert antiangiogenic and anti-inflammatory effects on Kaposi’s sarcoma cells Translational Research 166 2015 111 133 10.1016/j.trsl.2015.02.009 25814167 87 Asha K. Balfe N. Sharma-Walia N. Concurrent Control of the Kaposi’s Sarcoma-Associated Herpesvirus Life Cycle through Chromatin Modulation and Host Hedgehog Signaling: a New Prospect for the Therapeutic Potential of Lipoxin A4 e02177-19 J Virol 94 2020 10.1128/JVI.02177-19 88 Chiang N. la Rosa X. Libreros S. Serhan C.N. Novel Resolvin D2 Receptor Axis in Infectious Inflammation (Baltimore, Md.: 1950) Journal of Immunology 198 2017 842 851 10.4049/jimmunol.1601650 89 Codagnone M. Cianci E. Lamolinara A. Mari V.C. Nespoli A. Isopi E. Mattoscio D. Arita M. Bragonzi A. Iezzi M. Romano M. Recchiuti A. Resolvin D1 enhances the resolution of lung inflammation caused by long-term Pseudomonas aeruginosa infection Mucosal Immunol 11 2018 35 49 10.1038/mi.2017.36 28422188 90 WHO to identify pathogens that could cause future outbreaks and pandemics, (n.d.). 〈https://www.who.int/news/item/21-11-2022-who-to-identify-pathogens-that-could-cause-future-outbreaks-and-pandemics〉 (accessed March 15, 2023). 91 WHO, WHO Coronavirus Disease (COVID-19) Dashboard, (n.d.). 〈https://covid19.who.int/〉. 92 Baltimore D. Expression of animal virus genomes Bacteriol Rev 35 1971 235 241 10.1128/br.35.3.235-241.1971 4329869
PMC010xxxxxx/PMC10286569.txt	==== Front Am J Infect Control Am J Infect Control American Journal of Infection Control 0196-6553 1527-3296 Published by Elsevier Inc. on behalf of Association for Professionals in Infection Control and Epidemiology, Inc. S0196-6553(23)00480-7 10.1016/j.ajic.2023.06.015 Article Addressing an increase in surgical site infections during the COVID-19 pandemic – identifying opportunities during a chaotic time Plummer Traci MPH a1 Zepeda Jordan MPH, CIC b2 Reese Sara M PhD, MPH, CIC, FAPIC c⁎3 a University of Colorado Anschutz, Colorado School of Public Health, 13001 E. 17th Pl, B119 Aurora, CO 80045 b St. Vincent Healthcare, 1233 N 30th St, Billings, MT 59101 c Intermountain Health, Clinical Excellence, 500 Eldorado Blvd, Broomfield, CO 80021 ⁎ Corresponding author. 1 (319) 480-4181 2 406-231-0544 3 970-420-8256 22 6 2023 22 6 2023 © 2023 Published by Elsevier Inc. on behalf of Association for Professionals in Infection Control and Epidemiology, Inc. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Healthcare systems saw increases in device-associated infections and decreases in surgical site infections (SSI) during the COVID-19 pandemic. However, following an increase in SSI, an acute care hospital assessed risk and preventative factors of SSIs among patients. Methods A retrospective cohort study on surgeries performed between January 2020 and September 2021 analyzed associations of SSI with risk and preventive factors utilizing Chi-Square, t-tests and odds ratios. A secondary analysis was utilized to determine association of case urgency and prevention practice performance. Results There was significant difference in administration of correct perioperative antibiotic selection between SSI (78.1%) and non-SSI (86.2%) along with 2.9 greater odds of developing an SSI with incorrect perioperative antibiotics. Patients who had urgent cases were significantly less likely than elective to receive pre-operative chlorhexidine gluconate wipes (81.6%, 61.5%, respectively), correct antibiotic selection and timing (93.2%, 70.8%, respectively) and chlorhexidine/alcohol skin preparation (81.6%, 67.5%, respectively). Discussion Disruption on perioperative workflow during the COVID-19 pandemic likely resulted in an increase in SSI. Numerous opportunities were identified for focused prevention efforts. Conclusions Next steps include implementing strategies to improve SSI prevention and establish a culture that can withstand workflow disruptions to maintain a safe environment during significant changes. Keywords COVID19 SSI Prevention ==== Body pmcBackground The COVID-19 pandemic brought many challenges to the healthcare system. The pandemic resulted in an influx of patients, staffing challenges with quarantine and disease, significantly sick patients and supply chain challenges to name a few. Numerous papers have demonstrated the impact of the pandemic on device-associated infections like central line associate bloodstream infections (CLABSI) and catheter-associated urinary tract infections (CAUTI).1, 2 However, the minimal literature describing the impact of the pandemic on surgical care and subsequently, surgical site infections (SSI) suggested the pandemic either resulted in no significant difference in SSI3, 4 or even observed a decrease.5, 6 The perioperative space was significantly impacted during the first year of the COVID-19 pandemic as operating rooms (OR) shut down and millions of elective surgeries were cancelled for months. Perioperative services were impacted in multiple ways including shortage of blood components, surgeons and perioperative staff members’ responsibilities shifted to other units, OR spaces converted to critical care units, limited opportunities to train new surgeons, educational conferences cancelled and significant backlog of cases resulting in greater complications.7 When elective procedures resumed, there were many new protocols put into place to keep healthcare workers and patients safe from COVID including pre-operative screening, considerations of COVID patient flow, consideration of aerosol generating procedures, minimal staff present for intubation and extubation and changes in patient flow.3 Additionally, post-operative protocols were abbreviated with the goal of discharging patients to limit their potential exposure to COVID in hospitals.8 The resumption of elective surgical services also had multiple downstream impacts including the challenge of scheduling millions of backlogged patients, perioperative staff returning to the OR after a hiatus, implementing the challenging and confusing pre-operative screening protocols and operating on patients who were more complicated due to the postponement of their surgery.7 Evidence has suggested that interruptions to the workflow during the pandemic was a common stressor for healthcare workers, and it potentially compromised patient safety and health outcomes.9 Following an over 40% increase in the standardized infection ratio (SIR) between the year prior to the COVID-19 pandemic (March 2019 – March 2020, 0.76) and the year shortly after the COVID shutdown (September 2020 – September 2021, 1.06) across multiple surgical specialties, a quality assurance project was developed to identify potential modifiable practices used among cases with identified SSI occurring between January 2020 and September 2021 at one rural acute care hospital. Methods Setting This retrospective cohort study was conducted at an acute care hospital located in Billings, MT. The hospital performs approximately 9,500 surgeries per year in sixteen ORs. This care site provides numerous surgical service lines including gastrointestinal, neurosurgical, spinal, gynecological, orthopedic and cardiac. This project was deemed a quality assurance project by the SCL Health Institutional Review Board as a measure of compliance with standards in a particular unit. Case definition and identification Surgical site infections were identified utilizing the NHSN SSI surveillance definitions10 following select surgical procedures performed at the care site between January 1, 2020, and September 30, 2021. Selection of controls A list of patients was generated based on ICD-10 codes defined by NHSN as procedures surveilled for SSI.10 Exclusion criteria applied to the overall data set included an ASA rating of 6 (i.e., represented an organ donation surgery), recorded minimum body temperature pre- and post-operatively < 94°, pediatric patients aged 17 years or younger, and “John Doe” patients with an incomplete medical record listed as 120 years of age. For each record, numerous modifiable and non-modifiable risk factors were abstracted from the patient’s medical record. Non-modifiable risk factors included age (≥18 years old), body mass index (BMI), sex (male, female), diabetes diagnosis (yes, no), surgery length (minutes), wound class (clean, clean contaminated, contaminated and dirty), intraoperative estimated blood loss (mL), American Society of Anesthesiologists (ASA) classification (1 – 5), surgery duration (minutes to perform the surgery) and urgency of surgery, as determined by the surgical team (elective, urgent or emergent). Elective surgeries were scheduled surgeries, emergent surgeries were those deemed necessary to be sent to the OR as soon as possible, and urgent case classification signaled the patient needed the operation within 24 – 48 hours. The modifiable risk factors were chosen based on standard pre-, intra- and post-operative protocols for all procedures which included:- Patient pre- and post-operative temperature (≥96.8˚F/36.0˚C) utilized as a continuous variable - Pre- and post-operative glucose level (dichotomous variable created to represent an elevated blood glucose of ≥ 200 mg/dL or < 200 mg/dL with the first measurement of blood glucose prior to and the second measurement following the conclusion of surgery) - Administration of chlorhexidine gluconate (CHG) wipes pre-operatively (yes, no) - Appropriate skin antisepsis product defined by case type (70% alcohol, povidone/iodine, chlorhexidine without alcohol, or chlorhexidine and alcohol) - Appropriate perioperative antibiotic type defined by case type (yes, no, none) - Appropriate perioperative antibiotic timing defined by antibiotic type (yes, no, none) - Appropriate perioperative antibiotic type and timing (yes, no, none). Correct antibiotic administration (yes, no, none) was determined if the antibiotic administered to the patient prior to incision time was first deemed appropriate according to the procedure type based on the healthcare system’s guidelines. A value of ‘None’ represented no antibiotic was administered to the patient in the time period assessed. The timing of antibiotic administration was determined by utilizing the guidelines for infusion. Administration could occur either within 60 minutes prior to surgery start or between 60 and 120 minutes prior to surgery, as recommended for specific antibiotics based on system established and evidenced-based guidelines. A secondary retrospective cohort analysis evaluating the association between case classification (i.e., elective, urgent and emergent) and modifiable risk factors (i.e., pre-operative CHG wipes, appropriate peri-operative antibiotics and timing and skin prep) was completed to determine if there was an association between operation urgency and modifiable risk factors. Univariate analyses were completed to assess frequencies, proportions of covariates and odds ratios to compare and identify potential differences in preventative factors between SSI and non-SSI. Additionally, univariate analyses were used to determine the odds of receiving specific preventative tasks with varying case urgency. Next, bivariate analyses were conducted on each covariate to measure its associated odds with the outcome of an SSI. Any covariates with a p-value of <0.05 was considered significant. All analyses were completed using SAS OnDemand: SAS Studio Version 9.4. Results There were 8,022 eligible surgeries identified for potential inclusion following application of exclusion criteria. With a sample size of 96 in the SSI group and 7926 in non-SSI group (with alpha = 0.05 and d = 0.5), the estimated power was >99%.11 Of the 8,022 eligible surgeries, 96 (1.2%) resulted in an SSI. Knee prostheses (12.5%; 1,005), spinal fusions (10.0%; 804), hip prothesis (9.1%; 731), open reduction of fractures (8.4%; 674) and cesarean sections (6.8%;548) were the five most common procedures in the population. In cases that resulted in SSI (96), colon surgeries (15.5%; 15), laminectomies (9.4%; 9) and limb amputations (9.4%; 9) were the most prevalent. Of the 8,022 eligible surgeries, there were 5,554 (69.2%) elective surgeries, 2,194 urgent surgeries (27.4%) and 274 (3.4%) emergent surgeries. The median time from admission to surgery varied by case type with elective (3.0 hours) being the shortest time, followed by emergent cases (7.5 hours) and urgent cases (21.0 hours). In the cohort, the average age of patients was 59.7 (± 15.3) and 58.5 (±17.3) for cases and controls, respectively (p = 0.55) ( Table 1). There was not a statistically significant difference in BMI between cases (32.1) and controls (31.6; p = 0.2). Diabetes was significantly different between cases (25.0%; 24) and controls (14.2%; 41; p = 0.02). Almost half of the SSI (42.7%; 41) were urgent cases compared to only 27.1% of the non-SSI cases (2154; p = 0.003).Table 1 Risk and preventative factors and association with SSI and non-SSI. Table 1Risk Factor SSI N = 96 Non-SSI N = 7926 P-value Age (Mean ± SD) 59.7 ± 15.3 58.5 ± 17.3 0.55 BMI (Mean ± SD) 32.1 ± 9.1 31.6 ± 54.4 0.20 Surgery Length in Minutes (Mean ± SD) 122.4 ± 74.6 99.9 ± 71.6 0.002 Minimum temperature Pre-op minimum temperature 97.5 ± 0.7 97.6 ± 0.7 0.32 Post-op minimum temperature 96.8 ± 0.6 96.8 ± 0.6 0.48 N(%) N(%) Sex 0.38 Male 40 (41.7) 3655 (46.1) Female 56 (58.3) 4271 (53.9) Diabetes Diagnosis 0.0002 No 72 (75.0) 6944 (87.6) Yes 24 (25.0) 982 (12.4) Wound class <0.0001 Clean 49 (51.0) 5644 (71.2) Clean Contaminated 19 (19.8) 1393 (17.6) Contaminated 18 (18.8) 479 (6.0) Dirty 10 (10.4) 410 (5.2) Case class 0.003 Elective 52 (54.2) 5501 (69.4) Urgent 41 (42.7) 2154 (27.1) Emergent 3 (3.1) 271 (3.4) ASA 0.2 1 0 (0) 334 (4.2) 2 42 (43.8) 3951 (49.8) 3 44 (45.8) 3044 (38.4) 4 10 (10.4) 580 (7.3) 5 0 (0) 17 (0.2) CHG pre-operative wipes 0.23 Yes 71 (74.0) 6262 (79.0) No 25 (26.0) 1664 (21.0) Correct antibiotics 0.002 Yes 75 (78.1) 6834 (86.2) No 12 (12.5) 375 (4.7) None 9 (9.4) 717 (9.0) Correct antibiotics and timing <0.001 Yes 64 (66.7) 6298 (79.5) No 23 (24.0) 911 (11.5) None 9 (9.4) 717 (9.0) Skin preparation⁎ 0.006 Alcohol 5 (5.2) 671 (8.5) CHG 3 (3.1) 178 (2.3) CHG/alcohol 63 (65.6) 6148 (78.3) Povidine/Iodine 24 (25.0) 846 (10.9) Pre-op glucose 0.29 < 200 mg/Dl 46 (97.9) 2576 (94.3) ≥ 200 mg/dL 1 (2.1) 155 (5.7) Post-op glucose 0.12 < 200 mg/Dl 89 (92.7) 7601 (95.9) ≥ 200 mg/dL 7 (7.3) 325 (4.1) SSI: Surgical Site Infections; SD: Standard deviation; BMI: body mass index; ASA: American Society of Anesthesiology; CHG: chlorhexidine gluconate ⁎ Missing values were omitted from the analysis There was significant difference in administration of correct antibiotic between SSI cases (78.1%; 75) and non-SSI (86.2%; 6834; p = 0.002). Among the non-SSI group, 79.5% (6298) of patients received correct antibiotic administration and timing, compared to 66.7% (64) of SSI (p <0.001). A significant difference was also observed with the type of skin preparation product used between the SSI and non-SSI where 25% (24) of the SSI had povidone/iodine used as a skin preparation product, compared to only 11% (846) of the non-SSI (p = 0.006). Unadjusted odds ratios were assessed for their potential association with the outcome of SSIs ( Table 2). Patients who did not receive the correct antibiotics had a 2.9 times greater odds (95% CI: 1.5, 4.0; p <0.001) of developing an SSI compared to those who received the correct antibiotics. Additionally, patients who were prepped preoperatively with povidone/iodine had 2.8 times greater odds of develop an SSI (95% CI: 1.7, 4.5; p < 0.0001) than patients prepped with chlorhexidine/alcohol. Patients who had an urgent case had 2.0 times greater odds to develop an SSI than patients who were an elective case (95% CI: 1.3, 3.0; p <0.001).Table 2 Odds Ratios of Risk Factors by Primary Explanatory Variable. Table 2 Crude Association w/SSI Risk Factor Unadjusted OR 95% CI P-Value Pre-operative Glucose (≥ 200 mg/dL vs. < 200 mg/dL) 0.36 0.05,2.63 0.32 Post-operative Glucose (≥ 200 mg/dL vs. < 200 mg/dL) 1.84 0.84, 4.00 0.12 Correct Antibiotic + Timing (No vs Yes) 2.48 1.54, 4.02 <0.001 Correct Antibiotic (No vs Yes) 2.92 1.57, 5.41 <0.001 CHG pre-operative wipes (No vs Yes) 1.33 (0.84, 2.1) 0.23 Skin Preparation Agent (Chlorhexidine w/o Alcohol vs. Chlorhexidine w/ Alcohol) 1.64 0.51, 5.29 0.40 (Alcohol 70% vs. Chlorhexidine w/ Alcohol) 0.72 0.29,1.81 0.49 (Povidone Iodine vs. Chlorhexidine w/ Alcohol) 2.77 1.72,4.45 <0.0001 Case Class (Urgent vs Elective) 2.01 (1.33, 3.04) <0.001 (Emergent vs Elective) 1.17 (0.36, 3.77) 0.79 SSI: surgical site infection; OR: Odds ratio; CHG: chlorhexidine Patients who had urgent cases (67.5%; 1481) were significantly less likely to have skin preparation with chlorhexidine/alcohol compared to patients who had elective cases (81.6%; 4531; p <0.0001) ( Table 3). Patients who underwent urgent surgery were significantly more likely to receive incorrect ABX (p < 0.0001) compared to patients who had elective surgeries. Additionally, patients who had elective surgery were significantly more likely to have the correct antibiotic and timing (5792, 86.3%) compared to those patients who had urgent procedures (1413, 64.4%, p <0.0001). Patients who underwent urgent procedures were also significantly less likely to receive a CHG bath in pre-operative unit (1326, 60.4) compared to those patients who had elective procedures (4969, 89.5, p <0.0001).Table 3 Univariate analysis of SSI prevention tactics among case classification. Table 3 Elective N (%) Urgent N (%) Emergent N (%) P-value Skin prep <0.0001 Chlorhexidine/Alcohol 4531 (81.6) 1481 (67.5) 199 (72.6) Chlorhexidine 131 (2.4) 45 (2.1) 5 (1.8) Povidone/Iodine 449 (8.1) 365 (16.6) 56 (20.4) 70% alcohol 376 (6.8) 288 (13.1) 12 (4.4) Correct antibiotics <0.0001 Yes 5176 (93.2) 1554 (70.8) 179 (65.3) No 133 (2.4) 210 (9.6) 44 (16.0) Correct antibiotics and timing <0.0001 Yes 5792 (86.3) 1413 (64.4) 158 (57.7) No 518 (9.3) 351 (16.0) 65 (23.7) Pre-operative CHG wipes <0.0001 Yes 4969 (89.5) 1326 (60.4) 38 (13.9) No 584 (10.5) 869 (39.6) 236 (86.1) SSI: surgical site infection; CHG: chlorhexidine Discussion The goal of this project was to determine opportunities for improvement following an increase in SSI in multiple service lines during the COVID pandemic. It was identified that incorrect antibiotic selection, timing of antibiotic administration and skin preparation in the OR were modifiable risk factors all significantly associated with SSI. Additionally, it was determined that basic SSI prevention practices like CHG bathing in the pre-operative unit, skin preparation utilizing chlorhexidine/alcohol and administration of appropriate peri-operative antibiotics within the correct time are practices that are significantly less likely to occur during urgent cases compared to elective cases. This approach utilizing a retrospective cohort analysis identified numerous opportunities for improvement across multiple service lines for the resource-limited infection prevention team during the COVID-19 pandemic. The opportunities identified in this retrospective cohort study were not surprising as the importance of perioperative antibiotics (type and timing) and alcohol-based skin preparation product are well documented in the literature.12, 13 High-quality evidence supports intraoperative preparation of the skin using an alcohol-based antiseptic agent unless otherwise contraindicated.12 Additionally, meta-analyses have demonstrated significantly lower risk of infection with the use of a CHG/alcohol skin prep compared to povidone/iodine alone.14, 15 The utilization of perioperative antibiotics has significantly reduced the incidence of SSIs, specifically within 60 minutes of incision for most antibiotics.16 The analysis also revealed the increased risk of SSI based on the urgency of the surgical case with urgent cases (cases that need to be performed within 24-48 hours) having 2.0 increased odds of an SSI compared to elective cases. Elective surgeries allow time for appropriate skin preparation and pre-operative order sets that contain the recommended perioperative antibiotics. This project underscored several workflows around urgent cases that need to be addressed to improve patient safety and decrease risk of SSI. Evidence has suggested that patients who undergo urgent surgical procedures have an increased risk of morbidity and mortality.17, 18 The findings from this project highlight the increased risk of infection as well as the prevention tactics that are less likely to occur in urgent cases, even when patients are admitted 24-48 hours prior to their urgent surgery. This increase in SSI during the COVID-19 pandemic is different from what the literature suggests was reported at other facilities. There is evidence that many facilities saw no change,3, 4 or even a decrease in SSI incidence5, 6 during the COVID-19 shutdown and subsequent reopening of the ORs due to improvements in infection prevention practices like hand hygiene, glove use and mask compliance, limiting surgical case load, shorter post-operative length of stay and decreased interactions with healthcare workers.19, 20 There could be numerous reasons there was an increase observed in this setting including the variation in surveillance tactics during this time,21 rural health disparities22 or varying impacts of the pandemic on the healthcare community.23 The increase in SSI in multiple service lines potentially resulted from the workflow disruption in the perioperative space during the COVID-19 pandemic in 2020 and 2021. Disruptions in the daily workflow can distract nurses from patient care, causing delayed or missed tasks or medical errors and the impact of those disruptions were multiplied during the pandemic.7, 24 The missed opportunities for pre-operative bathing, appropriate skin preparation, appropriate antibiotic administration and timing and performing specific prevention tactics in urgent cases are prime examples of missed tasks or medical errors. Koch et al’s 2020 systematic review suggested that workflow disruptions result in negative associations with surgical outcomes.25 While the significant disruption in workflow due to the pandemic was an unlikely event, the importance of maintaining SSI prevention tactics in a time of disruption was highlighted. In a time where the infection prevention team was overwhelmed with the pandemic response and little resources to investigate the increase in SSIs, this approach identified opportunities that could be addressed systematically with minimal effort. Strengths of this project included a large sample size of over 8000 patient surgeries. In addition, there was an extensive number of recorded risk factors in each patient medical record, as well as those included on required NHSN surveillance forms for procedures and SSIs. Extensive variables included both non-modifiable factors and modifiable surgical techniques that aided in quality improvement and contributed to broader public health knowledge. Limitations of this single facility project included data collection beginning in 2020, omitting the time period before the pandemic as comparison. As surgeries were performed from January 2020 through September 2021, there was also no control over what variables were recorded or how they were measured. This led to occasional data entry errors in patient medical records. There were also only three SSIs identified in emergent cases, which limited any comparisons with process and outcome metrics. Additionally, as the time frame of interest was over a year and a half, variations in measurement, recording, or surveillance may have occurred. Another limitation of this study is that it was not stratified by service line or matched by surgical procedure. As part of a quality assurance project, this study aimed to inform resource-deprived infection preventionists in a pandemic setting on opportunities for preventing SSIs throughout the perioperative space. This approach allowed a quick overview of the opportunities that needed to be addressed without expending significant resources. The infection preventionists can partner with perioperative leadership to not only address the opportunities identified in this quality assurance project, but also to address the response of the workflow disruption due to the pandemic. Next steps include implementing strategies to improve SSI prevention efforts and establish a perioperative culture that can withstand workflow disruptions to maintain the safest patient environment possible during significant changes. Acknowledgements The authors would like to thank Zachary Felix for abstracting the dataset for this project and responding to the numerous changes in the dataset to meet the needs of the project. The authors would also like to thank the Centralized Surveillance Team that assisted the infection prevention team in SSI surveillance to assure that the daily infection prevention tasks during the pandemic continued. ==== Refs References 1 Fakih M.G. Bufalino A. Sturm L. Huang R.H. Ottenbacher A. Saake K. Coronavirus disease 2019 (COVID-19) pandemic, central-line-associated bloodstream infection (CLABSI), and catheter-associated urinary tract infection (CAUTI): The urgent need to refocus on hardwiring prevention efforts Infect Control Hosp Epidemiol 43 1 2022 Jan 26 31 10.1017/ice.2021.70 Epub 2021 Feb 19 33602361 2 Baker M.A. Sands K.E. Huang S.S. Kleinman K. Septimus E.J. Varma N. CDC Prevention Epicenters Program. The Impact of Coronavirus Disease 2019 (COVID-19) on Healthcare-Associated Infections Clin Infect Dis 74 10 2022 May 30 1748 1754 10.1093/cid/ciab688 PMID: 34370014; PMCID: PMC8385925. 34370014 3 Humphrey T. Daniell H. Chen A.F. Hollenbeck B. Talmo C. Fang C.J. Effect of the COVID-19 Pandemic on Rates of Ninety-Day Peri-Prosthetic Joint and Surgical Site Infections after Primary Total Joint Arthroplasty: A Multicenter, Retrospective Study Surg Infect (Larchmt) 23 5 2022 Jun 458 464 10.1089/sur.2022.012 Epub 2022 May 20. PMID: 35594331 35594331 4 Smith B.B. Bosch W. O'Horo J.C. Girardo M.E. Bolton P.B. Murray A.W. Surgical site infections during the COVID-19 era: A retrospective, multicenter analysis Am J Infect Control 51 6 2023 Jun 607 611 10.1016/j.ajic.2022.09.022 Epub 2022 Sep 23. PMID: 36162605; PMCID: PMC9500048 36162605 5 Sybert M. Oakley C.T. Christensen T. Bosco J. Schwarzkopf R. Slover J. Impact of COVID-19 Protocols on Primary and Revision Total Hip Arthroplasty J Arthroplasty 37 11 2022 Nov 2193 2198 10.1016/j.arth.2022.05.035 Epub 2022 May 19. PMID: 35598760; PMCID: PMC9119172 35598760 6 Chacón-Quesada T. Rohde V. von der Brelie C. Less surgical site infections in neurosurgery during COVID-19 times-one potential benefit of the pandemic? Neurosurg Rev 44 6 2021 Dec 3421 3425 10.1007/s10143-021-01513-5 Epub 2021 Mar 5. PMID: 33674981; PMCID: PMC7935474 33674981 7 Al-Jabir A. Kerwan A. Nicola M. Alsafi Z. Khan M. Sohrabi C. Impact of the Coronavirus (COVID-19) pandemic on surgical practice - Part 1 Int J Surg 79 2020 Jul 168 179 10.1016/j.ijsu.2020.05.022 Epub 2020 May 12. PMID: 32407799; PMCID: PMC7214340 32407799 8 Wright E.V. Musbahi O. Singh A. Somashekar N. Huber C.P. Wiik A.V. Increased perioperative mortality for femoral neck fractures in patients with coronavirus disease 2019 (COVID-19): experience from the United Kingdom during the first wave of the pandemic Patient Saf Surg 15 1 2021 Jan 10 8 10.1186/s13037-020-00279-x PMID: 33423685; PMCID: PMC7797178 33423685 9 Baethge A. Rigotti T. Interruptions to workflow: Their relationship with irritation and satisfaction with performance, and the mediating roles of time pressure and mental demands https://doi.org/ Work & Stress 27 1 2013 43 63 10.1080/02678373.2013.761783 10 Surgical Site Infection Event. National Healthcare Safety Network. 〈https://www.cdc.gov/nhsn/PDFs/pscManual/9pscSSIcurrent.pdf?agree=yes〉. Published January 2021. Accessed October 22, 2021. 11 Faul F. Erdfelder E. Buchner A. Lang A.-G. Statistical power analyses using G⁎Power 3.1: Tests for correlation and regression analyses Behavior Research Methods 41 2009 1149 1160 10.3758/BRM.41.4.1149 19897823 12 Yokoe D.S. Anderson D.J. Berenholtz S.M. Calfee D.P. Dubberke E.R. Ellingson K.D. A compendium of strategies to prevent healthcare-associated infections in acute care hospitals: 2014 updates Infect Control Hosp Epidemiol 35 Suppl 2 2014 Sep S21 S31 10.1017/s0899823x00193833 25376067 13 Berríos-Torres S.I. Umscheid C.A. Bratzler D.W. Centers for Disease Control and Prevention Guideline for the Prevention of Surgical Site Infection 2017 [published correction appears in JAMA Surg 152 8 2017 Aug 1 803 10.1001/jamasurg.2017.0904 14 Hasegawa T. Tashiro S. Mihara T. Kon J. Sakurai K. Tanaka Y. Efficacy of surgical skin preparation with chlorhexidine in alcohol according to the concentration required to prevent surgical site infection: meta-analysis BJS Open 6 5 2022 Sep 2 zrac111 10.1093/bjsopen/zrac111 PMID: 36124902; PMCID: PMC9487656 36124902 15 Mastrocola M. Matziolis G. Böhle S. Linderman C. Schlattmann P. Eijer H. Meta-analysis of the efficacy of preoperative skin preparation with alcoholic chlorhexidine compared to povidone iodine in orthopedic surgery Sci Rep 11 2021 18634 10.1038/s41598-021-97838-8 34545135 16 de Jonge S.W. Gans S.L. Atema J.J. Solomkin J.S. Dellinger P.E. Boermeester M.A. Timing of preoperative antibiotic prophylaxis in 54,552 patients and the risk of surgical site infection: A systematic review and meta-analysis Medicine (Baltimore) 96 29 2017 e6903 10.1097/MD.0000000000006903 17 Mullen M.G. Michaels A.D. Mehaffey J.H. Guidry C.A. Turrentine F.E. Hedrick T.L. Risk Associated With Complications and Mortality After Urgent Surgery vs Elective and Emergency Surgery: Implications for Defining “Quality” and Reporting Outcomes for Urgent Surgery JAMA Surg 152 8 2017 768 774 10.1001/jamasurg.2017.0918 28492821 18 Protopapa K.L. Simpson J.C. Smith N.C.E. Moonesinghe S.R. Development and validation of the Surgical Outcome Risk Tool (SORT https://doi.org/ British Journal of Surgery 101 13 2014 Dec 1774 1783 10.1002/bjs.9638 25388883 19 Chacón-Quesada T. Rohde V. von der Brelie C. Less surgical site infections in neurosurgery during COVID-19 times-one potential benefit of the pandemic? Neurosurg Rev 44 6 2021 Dec 3421 3425 10.1007/s10143-021-01513-5 Epub 2021 Mar 5. PMID: 33674981; PMCID: PMC7935474. 33674981 20 Smith B.B. Bosch W. O'Horo J.C. Girardo M.E. Bolton P.B. Murray A.W. Surgical site infections during the COVID-19 era: A retrospective, multicenter analysis Am J Infect Control S0196-6553 22 2022 Sep 23 10.1016/j.ajic.2022.09.022 Epub ahead of print. PMID: 36162605; PMCID: PMC9500048 21 Troillet N. Aghayev E. Eisenring M.C. Widmer A.F. Swissnoso. First Results of the Swiss National Surgical Site Infection Surveillance Program: Who Seeks Shall Find Infect Control Hosp Epidemiol 38 6 2017 Jun 697 704 10.1017/ice.2017.55 Erratum in: Infect Control Hosp Epidemiol. 2017 Nov;38(11):1389. PMID: 28558862 28558862 22 Baljepally V.S. Metheny W. Rural-urban disparities in baseline health factors and procedure outcomes J Natl Med Assoc 114 2 2022 Apr 227 231 10.1016/j.jnma.2022.01.001 Epub 2022 Jan 31 35109969 23 Cuadros D.F. Branscum A.J. Mukandavire Z. Miller F.D. MacKinnon N. Dynamics of the COVID-19 epidemic in urban and rural areas in the United States Ann Epidemiol 59 2021 Jul 16 20 10.1016/j.annepidem.2021.04.007 Epub 2021 Apr 22. PMID: 33894385; PMCID: PMC8061094 33894385 24 Maniya Chandaniben "Workflow Interruptions: Risk Factors and Outcomes in Nursing" Honors Capstones 2018 1401 〈https://huskiecommons.lib.niu.edu/studentengagement-honorscapstones/1401〉 25 Koch A. Burns J. Catchpole K. Associations of workflow disruptions in the operating room with surgical outcomes: a systematic review and narrative synthesis BMJ Quality & Safety 29 2020 1033 1045
PMC010xxxxxx/PMC10286570.txt	==== Front Clin Nutr Clin Nutr Clinical Nutrition (Edinburgh, Scotland) 0261-5614 1532-1983 Published by Elsevier Ltd. S0261-5614(23)00203-0 10.1016/j.clnu.2023.06.017 Original Article Association of Gut microbiota and Dietary component intake with COVID-19：A Mendelian randomization study Zhang Hanyu a∗1 Zhou Zengyuan b1 a Department of General Practice, Clinical Medical College& Affiliated Hospital of Chengdu University, Chengdu, China b Department of Nutrition, Chengdu Women's and Children's Central Hospital, School of Medicine, Chengdu, University of Electronic Science and Technology of China, China ∗ Corresponding author: 1 These authors have contributed equally to this work. 22 6 2023 22 6 2023 27 4 2023 3 6 2023 16 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Growing evidence has indicated that alterations in the gut microbiota and nutritional quality of dietary intake were associated with COVID-19. Whether these associations reflect causality is still unknown. Methods We performed a two-sample Mendelian randomization analysis using genetic variants as instrumental variables for gut microbiota, dietary component intake, and COVID-19. Findings We found that the Ruminococcustorques group genus was significantly associated with COVID-19. The Ruminococcaceae UCG013 genus and Ruminococcus1 genus were suggestively associated with COVID-19. The Actinobacteria class, Bifidobacteriales order, Bifidobacteriaceae genus, Ruminococcustorques group, and Tyzzerella3 genus were potentially associated with severe COVID-19. COVID-19 was significantly associated with the Lachnospira genus, Oscillospira, and RuminococcaceaeUCG009 genus and potentially associated with the Victivallis genus. Severe COVID-19 was significantly associated with the Turicibacter and Olsenella genus and potentially associated with Ruminococcus1, CandidatusSoleaferrea, and Parasutterella genus. Moreover, processed meat intake was significantly associated with COVID-19. Beef intake was suggestively associated with COVID-19. Salt added to food intake, and fresh fruit intake was suggestively associated with severe COVID-19. Conclusions Our findings provide evidence supporting a causal effect of gut microbiota and dietary intake on COVID-19. We also found the causal effect of COVID-19 on the alteration of gut microbiota. Keywords Mendelian randomization gut microbiota dietary component intake COVID-19 ==== Body pmc
PMC010xxxxxx/PMC10286572.txt	==== Front J Mol Biol J Mol Biol Journal of Molecular Biology 0022-2836 1089-8638 Published by Elsevier Ltd. S0022-2836(23)00285-1 10.1016/j.jmb.2023.168187 168187 Research Article Free Energy Perturbation Calculations of Mutation Effects on SARS-CoV-2 RBD::ACE2 Binding Affinity Sergeeva Alina P. a Katsamba Phinikoula S. b Liao Junzhuo c Sampson Jared M. cd Bahna Fabiana b Mannepalli Seetha b Morano Nicholas C. b Shapiro Lawrence be⁎ Friesner Richard A. c⁎ Honig Barry abef⁎ a Department of Systems Biology, Columbia University Medical Center, New York, NY 10032 b Zuckerman Mind Brain and Behavior Institute, Columbia University, New York, NY 10027, USA c Department of Chemistry, Columbia University, New York, NY 10027, USA d Schrödinger, Inc., New York, NY 10036, USA e Department of Biochemistry and Molecular Biophysics, Columbia University, New York, NY 10032, USA f Department of Medicine, Columbia University, New York, NY 10032 ⁎ Corresponding authors. 22 6 2023 22 6 2023 16818710 1 2023 13 6 2023 15 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Graphical abstract The strength of binding between human angiotensin converting enzyme 2 (ACE2) and the receptor binding domain (RBD) of viral spike protein plays a role in the transmissibility of the SARS-CoV-2 virus. In this study we focus on a subset of RBD mutations that have been frequently observed in infected individuals and probe binding affinity changes to ACE2 using surface plasmon resonance (SPR) measurements and free energy perturbation (FEP) calculations. Our SPR results are largely in accord with previous studies but discrepancies do arise due to differences in experimental methods and to protocol differences even when a single method is used. Overall, we find that FEP performance is superior to that of other computational approaches examined as determined by agreement with experiment and, in particular, by its ability to identify stabilizing mutations. Moreover, the calculations successfully predict the observed cooperative stabilization of binding by the Q498R N501Y double mutant present in Omicron variants and offer a physical explanation for the underlying mechanism. Overall, our results suggest that despite the significant computational cost, FEP calculations may offer an effective strategy to understand the effects of interfacial mutations on protein-protein binding affinities and, hence, in a variety of practical applications such as the optimization of neutralizing antibodies. Keywords FEP SPR protein-protein interactions ddG prediction receptor binding domain and angiotensin converting enzyme 2 ==== Body pmcIntroduction The ability to accurately predict binding affinity changes upon mutations of interfacial residues is a problem of significant importance, ranging from the general problem of understanding of interaction specificity and the design of therapeutics such as potent monoclonal antibodies that target antigens to revealing the mechanism of action of cancer driver mutations. Multiple approaches to the problem have been developed including machine learning 1, 2, 3, 4, 5, statistical potentials [6] and various force field related scoring functions 7, 8, 9, 10, 11 embedded in programs such as FoldX [11] and Rosetta [7]. Each approach is associated with its own set of issues, such as conformational changes upon mutation, that are nicely discussed in reviews of Bonvin and co-workers [12]. Moreover, some methods succeed on some test sets and fail on others, suggesting either over-training or simply that some protein-protein interfaces have different properties than others. Issues of experimental validation can also arise [13]; not all experimental methods are equally accurate and, as discussed below, the nuances of the experimental system can have significant effects on the outcome. Detailed atomic-level simulations have not been extensively applied to the prediction of mutation effects, in part due to the computational requirements involved. Free-energy perturbation (FEP) methods have the potential to impact the field as physics-based force fields are, in principle, agnostic to the system being studied. Most current applications have involved the optimization of ligand-protein interaction in the context of small molecule drug design (reviewed in [14]) but recent publications have begun to explore the use of FEP methods to the study of protein-protein interactions (PPIs); specifically, to the effects of interfacial mutations on protein-protein binding free energies 8, 9, 15, 16, 17, 18, 19. This is an inherently complex problem since, as opposed to relatively rigid ligand binding pockets, protein-protein interfaces are often quite large and less constrained so that they can more easily undergo conformational change as a result of a mutation. Moreover, FEP calculations involve a complex computational infrastructure and are extremely time consuming. However, fast graphical processing units (GPUs) make such calculations feasible and a number of recent publications, involving different software packages, suggest that the methodology has reached the point that good correlation with experiment is to be expected 8, 9, 15, 16, 17, 18, 19. Clearly, if FEP methods are capable of providing meaningful results, then in many applications, the computational cost will be worthwhile. Here we explore the ability of FEP calculations to reproduce the effects of mutations on the binding of the receptor binding domain (RBD) of the SARS-CoV-2 spike protein with the human angiotensin converting enzyme 2 (ACE2) using the FEP+ implementation (see Methods). Given that the pathogen entry into the host cell is mediated by RBD::ACE2 binding, the problem has attracted considerable interest and multiple experimental 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46 and computations studies 33, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63 have been reported. We chose to study a set of 23 frequently observed RBD mutations (Table S1) located in the RBD::ACE2 interface (Fig. 1 ) of Alpha, Beta, Gamma, Delta, or Omicron SARS-CoV-2 variants (Table S2). Surface Plasmon Resonance (SPR) experiments were carried out for each and compared to previous experimental work20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46. A number of computational methods were applied to predict binding free energy changes upon mutations, Δ Δ G. We found FEP to be the best performer and, moreover, FEP trajectory analysis facilitates the characterization of the biophysical effects that underlie mutational effects. We show that FEP successfully recapitulates the stabilizing epistatic effect of the Q498R N501Y mutant present in every Omicron variant. The ability to anticipate non-additive effects of multiple mutations is likely to be an important element of future efforts in protein interface design.Figure 1 The ACE2/RBD system. (A) Ribbon representation of the ACE2/RBD complex. (B) Side chains of interfacial RBD residues in contact with ACE2 probed in this study are shown in stick representation with C⍺ atoms shown as spheres. Results SPR Measurements of Binding Affinity Changes. The second column in Table 1 lists experimental changes in binding affinity (ΔΔG) of the ACE2::RBD complex when RBD is mutated. Of 23 single point RBD mutations probed, only four (N501Y, Y453F, S477N and N501T, Table 1, see Fig. S1 for corresponding fitted SPR data) were identified as stabilizing with ΔΔG values ≤ -0.4 which is our measure of experimental accuracy (see Methods). The third column lists ΔΔG values from the deep mutational scanning study of Starr et al.[20]. Although the methods are quite different and our SPR results are obtained with monomeric ACE2 while Starr et al. used dimeric ACE2, the results are in good agreement (high Pearson correlation coefficient (PCC)=0.9 and low root mean square error (RMSE)=0.2 kcal/mol). Given that our calculations are carried out on a structure containing monomeric ACE2 and the likelihood that the SPR results are more accurate than the high-throughput yeast display values, we use the SPR values to compare to computational predictions.Table 1 Experimental ACE2/RBD binding affinity changes for RBD mutants. Mutation ΔΔG (SPR, this study) ΔΔG (Yeast Display, Starr et al.)[20] ΔΔG (other studies)SPR21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33BLI34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46Yeast Display[64](range) N501Y -0.8 -0.3 (-1.7 … -0.5) Y453F -0.7 -0.3 (-1.2 … -0.8) S477N -0.5 -0.1 (-0.6 … -0.1) N501T -0.5 -0.1 (-0.9 … 0.0) N439K -0.1 0.0 (-0.4 … 0.0) N440K 0.0 -0.1 F490S 0.0 0.0 L452M 0.0 -0.1 L452R 0.0 0.0 -0.2 E484Q 0.1 0.0 T478K 0.1 0.0 0.2 N481K 0.1 0.0 E484K 0.1 0.1 (-0.6 … 0.3) Q498R 0.2 0.1 (-0.5 … 0.5) S477I 0.2 0.1 G446V 0.2 0.4 0.5 T478R 0.2 0.1 S477R 0.3 0.0 -0.5 A475V 0.3 0.2 0.2 L455F 0.4 0.3 K417T 0.4 0.4 (0.1 … 0.7) F486L 0.6 0.6 0.9 K417N 0.6 0.6 (-0.5 … 1.1) Experimental binding affinity difference values were calculated based on binding affinity (KD) measurements for wild-type (WT) and single mutant (MT) proteins using the following formula: ΔΔG = RT ln (KD(MT) / KD(WT)), in units of kcal/mol Table 1 also lists ΔΔG values obtained previously with SPR 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33 and other experimental methods; bio-layer interferometry (BLI) 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46 and yeast display [64]. Direct comparisons are difficult since, for example, different constructs were used in different experiments and different proteins were used as analytes in some SPR experiments (see Methods for details). The differences in constructs result from the choice of monomeric vs. multimeric forms of interacting proteins as well as the selection of protein domain boundaries. Nevertheless, overall, there is good agreement among most experimental results with the outliers attributable to the factors mentioned here. Moreover, there is good consensus regarding the identity of the most stabilizing mutations. For example, our results for N501Y (-0.8 kcal/mol), Y453F (-0.7 kcal/mol), S477N (-0.5 kcal/mol) and N501T (-0.5 kcal/mol) are in good agreement with previously published values 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 34, 36, 37, 38, 40, 41, 42, 43, 44, 45, 46, 65 regardless of constructs/experimental setup differences. Computational Prediction of the Effect of Mutations on Binding Affinity. Table 2 presents FEP results for the 23 experimental ΔΔG values obtained from SPR measurements listed in Table 1. Correlation plots for the data in Table 2 are given in Fig. S2. The overall performance of FEP is in line with previous work 8, 9, 14, 66, 67, 68; a PCC of 0.6 and an RMSE of 0.8 kcal/mol (Table 2). The FEP calculations clearly predict that all four stabilizing mutations, N501Y, N501T, S477N, and Y453F have ΔΔ G values < 0 although the prediction for S477N is a weak one. Of note, previous FEP calculations on N501Y 53, 55, 56 have yielded results very similar to ours attesting to the robustness of the method. The most significant failure of FEP is its prediction that A475V is stabilizing when the SPR results indicate that it is weakly destabilizing. A likely explanation for this result is that A475 is located close to the N-terminal residue of ACE (Q18) for which no coordinates were assigned in the crystal structure [69]. Another problematic result, a largely overestimated ΔΔ G for Q498R, will be discussed below. The FEP results in Table 2 were obtained from 100 ns simulations. Extensive prior work 8, 9 has demonstrated that shorter simulations are often inadequate for some mutations, typically those involving residues with a low degree of solvent exposure (i.e. partially or fully buried at the protein-protein interface). The results of 10 ns FEP simulations are shown in the Supplementary Material (Table S3) for comparison. Of note, the N501Y mutation, which is predicted to be destabilizing at 10 ns is correctly predicted to be stabilizing at 100 ns.Table 2 Calculated ACE2/RBD binding affinity changes for RBD mutants. Pearson correlation coefficient (PCC), and root mean square error (RMSE) are calculated for every method tested based on comparison to SPR results in Table 1. Stabilizing mutations with ΔΔG ≤ -0.4 kcal/mol in green, destabilizing mutations with ΔΔG ≥ 0.4 kcal/mol in red. Experimental binding affinity difference values were calculated based on SPR binding affinity (KD) measurements for wild-type (WT) and single mutant (MT) proteins using the following formula: ΔΔG = RT ln (KD(MT) / KD(WT)), in units of kcal/mol. Correlation plots for all theoretical methods are provided in Fig. S2. Calculations were performed on a crystal structure of ACE2/RBD (PDBID 6M0J). Protein specific residue numbering of all the mutants as in Uniprot ID P0DTC2. Mutation ΔΔG ΔΔG ΔΔG ΔΔG ΔΔG ΔΔG ΔΔG ΔΔG ΔΔG ΔΔG experiment SPR FEP 100ns Mutabind2 mCSM-PPI2 SAAMBE-3D BeAtMusic FoldX Rosetta flex ddG MM / PB-SA MM / GB-SA N501Y -0.8 -1.2 0.7 0.5 0.0 0.1 6.0 0.4 5.8 2.6 Y453F -0.7 -0.6 -0.2 0.1 0.1 0.3 -0.4 -0.2 1.3 2.2 S477N -0.5 -0.1 -0.1 -0.1 0.2 0.1 0.0 0.1 5.4 9.3 N501T -0.5 -1.8 -0.6 -0.9 -0.1 0.4 -0.9 -0.4 3.1 1.8 N439K -0.1 0.6 -0.1 0.5 0.6 0.3 0.0 0.0 -8.8 -0.8 N440K 0.0 -0.4 0.1 -0.1 0.5 0.0 -0.1 0.0 -1.9 -2.9 F490S 0.0 -0.1 0.6 0.2 0.8 0.7 0.0 0.0 -1.0 2.2 L452M 0.0 0.2 0.0 0.5 -0.3 0.2 0.0 0.0 6.3 2.1 L452R 0.0 -0.3 -0.8 0.0 0.1 0.2 -0.3 0.0 -6.8 -3.8 E484Q 0.1 0.3 0.1 0.3 0.5 0.2 -0.1 0.0 7.7 11.3 T478K 0.1 0.0 0.2 0.0 0.1 0.2 0.0 0.1 0.8 4.9 N481K 0.1 -0.1 0.0 0.0 0.6 0.2 0.0 0.0 -13.2 -7.4 E484K 0.1 0.2 0.2 0.3 0.3 0.1 -0.2 -0.1 3.1 7.2 Q498R 0.2 2.7 0.4 1.4 1.1 0.8 -0.5 -0.6 -1.0 -6.4 S477I 0.2 0.0 0.1 0.0 0.0 0.2 0.1 0.0 -1.3 6.1 G446V 0.2 0.4 -0.7 0.1 -0.3 1.3 0.1 0.0 6.0 6.6 T478R 0.2 -0.1 0.1 0.0 0.3 0.2 0.0 0.0 1.6 3.4 S477R 0.3 -0.2 0.2 -0.1 0.0 0.1 0.0 -0.2 0.6 3.1 A475V 0.3 -1.1 0.9 0.0 0.1 0.1 1.0 0.2 10.0 12.2 L455F 0.4 1.7 1.5 -0.7 0.2 -0.1 5.0 -0.6 6.5 5.1 K417T 0.4 0.9 0.2 0.4 0.5 0.5 0.8 0.7 11.4 14.1 F486L 0.6 1.3 0.3 0.9 0.5 0.7 1.1 1.8 5.9 7.8 K417N 0.6 0.9 0.6 0.5 0.4 0.5 0.8 0.9 12.2 13.1 PCC 0.6 0.3 0.2 0.3 0.2 -0.1 0.4 0.2 0.3 RMSE 0.8 0.5 0.5 0.5 0.5 1.7 0.5 6.5 6.9 Evaluation of theoretical methods on the ACE2::RBD dataset (Table 2) shows that machine learning (ML) algorithms (Mutabind2 [5], mCSM-PPI2 [3] and SAAMBE-3D [1]) uniformly have a weak correlation with experiment (PCC<0.4). As we have pointed out previously (60) ML methods tend to overpredict destabilizing mutations presumably due to the preponderance of destabilizing mutations in training sets. A related factor likely accounts for relatively low RMSEs of ML methods since most mutations in training sets have only small effects on binding affinities. BeAtMuSiC evaluates mutation effects using a statistical potential (6) while FoldX uses an empirical physics-based force field (10, 11). Both methods assume a rigid backbone although FoldX allows for side chain rearrangement upon mutation. Neither method produces a meaningful correlation with experiment. Of note, poor performance of FoldX can be attributed to outliers (Fig. S2) whereby mutations require backbone rearrangement. BeAtMuSiC identifies no stabilizing mutations while FoldX correctly identifies N501T and Y453F. Other than FEP, Rosetta flex ddG [7] is the only method that allows for backbone flexibility but its PCC is still significantly smaller than that of FEP and, as can be seen in Table 2, Rosetta flex ddG predicts two of the four stabilizing mutations (although the prediction for Y453F is a weak one). Nevertheless, even its partial success highlights the need to account for the ability of proteins to relax in response to interfacial mutations. Finally, we also tested Molecular Mechanics/Poisson−Boltzmann Surface Area (MM/PB-SA) and Molecular Mechanics/Generalized Born Surface Area (MM/GB-SA) methods, which utilize molecular mechanics energies along with a continuum representation of the solvent. As can be seen in Table 2, these methods performed poorly (see Methods for the protocol details). The ACE2::RBD dataset has limitations for statistical analysis given the narrow affinity range of the experimental values (from -0.8 to +0.6 kcal/mol, Table 2) and the preponderance of neutral (15 out of 23) over stabilizing and destabilizing mutations (4 each). For a more robust assessment, we have compared performance of various theoretical methods on a larger and more balanced dataset. Specifically, we have expanded the ACE2::RBD dataset with mutations from another system (DIP::Dpr). DIPs and Dprs are families of neuronal adhesion proteins that bind to one another with isoform-specific affinities [70]. We have previously studied a number of DIP::Dpr mutants using SPR 71, 72, 73 to establish specificity deteminants in this family. Here, we chose 18 single point mutations in Dpr6::DIP-α and Dpr10::DIPα complexes to combined with 23 ACE2::RBD mutations for a total of 41 data points. The combined ACE2::RBD DIP::Dpr dataset has a 2.5 times wider range of Δ Δ G values (-1.3 to +2.2 kcal/mol) than just ACE2::RBD alone and a more balanced distribution of stabilizing, neutral, and destabilizing mutations (Table S3). Table S3 contains the results of different computational methods applied to the combined data set. We use multiple statistical tests to measure performance. Based on the same measures used in Table 2 (PCC and RMSE), all methods exhibit similar performance but with somewhat improved PCCs (except MM/GB-SA). FEP gives the best results with a PCC of 0.8 and RMSE of 1.0 kcal/mol. The next best performing method was Mutabind2 (PCC of 0.6 and RMSE of 0.8). Rosetta flex ddG and FoldX have a PCC of 0.5 and 0.4, respectively, with FoldX having a large RMSE error (1.8 kcal/mol) mostly due to high energetic penalties for mutations that introduce clashes at the interface which are difficult to correct when the backbone fixed. The continuum methods displayed poor performance (PCC<0.3) and large errors (RMSE∼6-7). Table S3 also reports PCC values for the 18 observations in the DIP::Dpr data set alone. As is evident from the Table, FEP is by far the best performer with a PCC of 0.9, well above that of other methods. Of note, all methods perform better on the DIP::Dpr than on the ACE2::RBD data set. We attribute this observation to the fact that the DIP:Dpr interface is not prone to backbone rearrangement upon mutation. We categorized mutations as stabilizing (ΔΔG ≤ -0.4 kcal/mol, green), neutral (-0.4 < ΔΔG < 0.4 kcal/mol), or destabilizing (ΔΔG ≥ 0.4 kcal/mol, red). The cutoff of ±0.4 kcal/mol was chosen based on expected reproducibility error in SPR measurements (see Methods). We evaluated the ability of various theoretical methods to correctly classify mutations into each of these three categories - essentially yes or no regarding placement in a particular category as well as their ability to correctly classify the effects of each mutant into one of three categories (using Matthews correlation coefficient, MCC). As shown in Table S3, FoldX is best at classification into three categories (MCC (triple class) = 0.7; 32 out of 41 mutations correctly classified) and destabilizing mutations (MCC (dest) = 0.9). FEP is the best method for prediction of stabilizing mutations (MCC (stab) = 0.5) and second best for three category classification (MCC (triple class) = 0.6; 31 out of 41 mutations correctly classified). The probability of achieving this result by random chance is approximately 10-8. To assess the statistical significance of our FEP predictions, we calculated a p-value for the classification of stabilizing mutations, which was approximately 3.0 x 10-11 (Table S3) under the assumption that the reproducibility error of the calculations was close to zero. To test this assumption, we performed ten independent repeats of the 100ns FEP simulations for a select set of mutants. The standard deviation of ΔΔG values and the standard error of the mean (SEM) are reported in Table S4). On average, the ΔΔG values obtained starting with different random velocities deviate by approximately 0.2 kcal/mol (SEM ∼0.06). These small fluctuations would only affect the classification of mutations whose ΔΔ G values are very close to the cutoff. In fact, categorizing stabilizing mutations is not affected by using a mean value of multiple independent runs (Table S4). Physical Insights from Trajectory Analysis. The N501Y mutation is responsible for a high infectivity and transmissibility of the Alpha variant of SARS-CoV-2 [74] and has the largest stabilizing effect (Table 1). The N501T mutation found in the SARS-CoV-2 variants transmitted from mink to humans 75, 76 occurs at the same position. Analysis of FEP trajectories reveals that the stabilization effect associated with N501Y and N501T mutations is due to substitution of the asparagine with less polar side chains of tyrosine and threonine. N501 has only one of its polar groups satisfied in the wild-type (WT) structure while, throughout the course of the relevant trajectories, the hydroxyl groups of both tyrosine and threonine participate in hydrogen bonds (see dashed lines, Fig. 2 A). In addition to enhanced stability due to the absence of unsatisfied hydrogen bonds, both mutants undergo stabilizing interactions with Y41 of ACE2; the aromatic ring of Y501 participates in π-π stacking interactions (see purple lines in Fig. 2A) while the methyl group of T501 forms a hydrophobic contact with the aromatic ring of Y41 (see gray shading in Fig. 2A). Of note, previous studies have identified the role of π-π stacking as a source of stabilization of the N501Y mutant 47, 53, 56, 58 but the role of the unsatisfied hydrogen bond in the WT protein has not been emphasized.Figure 2 Structural origins of stabilizing ACE2/RBD interactions. Closeups of the ACE2(cyan)/RBD(pink) showing key interactions in wild-type (WT) and mutant proteins. Hydrophobic contacts are in grey, π-π interactions are in purple, hydrogen bonds are shown as black dashed lines, unsatisfied polar group of N501 is in cyan, favorable hydrophobic contact between T501 and Y41 in grey, and all distances are in Å. Analysis of trajectories associated with the Y453F mutation shows that in the WT protein, a hydrogen bond between the hydroxyl group of Y453 and the Nε atom of H34 is present in ∼25% of the WT trajectory (Fig. 2B) while, in most cases, these two residues form hydrogen bonds with trapped solvent molecules. In the Phe mutant, there is no need to satisfy the buried hydroxyl of the tyrosine while the Nε of H34 is satisfied by structured waters or backbone atoms (see dashed lines showing hydrogen bonds in Fig. 2B). Thus, the enhanced stability of the mutant is likely due to the greater hydrophobicity of a Phe relative to a Tyr. Of note, our trajectory analysis is in agreement with a previous study comparing crystallographic structures of the WT and Y453F mutant [32]. The simulations correctly predict that the S477N mutation is stabilizing but only weakly so. This residue faces the N-terminal of ACE2, and, specifically, the Q18 residue for which coordinates were missing in the crystal structure and, hence, modelled as an acetyl group cap in our simulations (Fig. 2C). Thus, the calculations may suffer from conformational uncertainties in this region. The trajectories reveal that the longer Asn side chain makes more contacts with the ACE2 N-terminal than does the WT Ser. We show only one snapshot from wild-type and mutant trajectories in Fig. 2C, but in reality, due to the flexibility of the ACE2 N-terminal, no specific contact is retained throughout the simulations. The largest error in the FEP calculations is for the Q498R mutant whose destabilizing effect is over-predicted by about 2.5 kcal/mol. Analysis of the 100 ns Q498R trajectory reveals that the Arg side chain samples different conformations with the most prevalent state (∼53%) involving an unfavorable polar-hydrophobic contact between N501 and the aliphatic chain of the Arg, leaving both polar groups of the Asn unsatisfied (Fig. 3 A). This is likely a contributor to the destabilizing ΔΔG value. As we have noted previously [9], a computationally unfavorable outlier of this magnitude is typically attributable to the failure of the molecular dynamics trajectories (100 ns in this study) to achieve a converged reorganization of the protein structure. To test this possibility, we carried out a 300 ns simulation which appears to reach convergence at ∼200ns (Fig. S3) where the destabilizing effect of Q498R is calculated to be 1.9 kcal/mol; reduced from 3.6 kcal/mol at 10 ns and 2.7 kcal/mol at 100 ns. It may well be the case that the system has converged to a metastable state at 300ns and that there are lower energy states not sampled in the course of the simulation.Figure 3 Epistatic effect of the double Q498R N501Y mutant. (A) A closeup of the complex between ACE2 (in cyan) and RBD (in pink) showing chemical interactions involving RBD residues 501 and 498. Hydrophobic contacts are in grey, π-π interactions are in purple, hydrogen bonds are shown as black dashed lines, unsatisfied polar groups are in cyan, and all distances are measured in Å. (B) Cooperativity of the Q498R N501Y double mutant probed by FEP and SPR computed as a difference between a hypothetical Δ Δ G (if the double mutant had an additive effect of two single point mutations) and the actual Δ Δ G of the double mutant. Yeast display assay values are from Zahradnik et a [64]. (C) Absence of cooperativity probed by FEP and SPR for the L452R T478K double mutant. Experimental result for the Delta RBD variant is from Liu et al. [92] (see Methods for details), Δ Δ G of single mutants are from the current study (Table 2). The FEP+ results are from 100ns trajectories. Epistatic Effect of the Q498R N501Y Double Mutant in the Omicron variant. Zahradnik et al. [64] demonstrated that the Q498R N501Y double mutant is more stabilizing than the additive effect of two single point mutations as estimated by their yeast display assay (a “cooperativity energy” of ∼ -1.7 kcal/mol). Our SPR results on the single Q498R and N501Y mutants predict, if their effects were additive, that the double mutant would be stabilizing by -0.6 kcal/mol (-0.8 + 0.2 kcal/mol for the single mutations, respectively) while the experimental value for the double mutant is -1.2 kcal/mol yielding a cooperativity energy of -0.6 kcal/mol (see Fig. 3B for the experimental ΔΔG values and Fig. S1 for corresponding fitted data and dissociation constants). While the experimental SPR and yeast display ΔΔG values differ, both methods indicate that a substantial epistatic effect is playing a role, perhaps contributing to the greater infectivity of the Omicron variants where these mutations are present. FEP calculations on the double mutant predict a ΔΔ G of -1.4 kcal/mol whereas the predicted additive effect of the two single mutants (Table 2) is +1.5 kcal/mol – corresponding to a cooperativity energy of -2.9 kcal/mol (Fig. 3B). Thus, the 100 ns calculations successfully predict the existence of cooperativity. A detailed examination of the structure of the double mutant provides a compelling physical interpretation as to why it behaves so differently from the single Q498R mutation. In the double mutant, Arg 498 forms a favorable pairing with the side chain of Tyr 501. The aliphatic portion of the Arg side chain packs against the Tyr aromatic ring creating an enhanced hydrophobic contact in the double mutant compared to the N501Y single mutant (grey shading, Fig. 3A), while one hydrogen of the Arg guanidinium head group forms a hydrogen bond with the oxygen of the Tyr hydroxyl group (dashed lines, Fig. 3A). The geometries of the two residues are such that this pairing can be carried out without introducing any conformational strain, either in the backbone or in the side chains themselves. Thus, in the presence of N501Y, the mutation from Gln to Arg actually enhances binding rather than diminishing it. As a control, we considered a second double mutant (L452R T478K) where no cooperativity is observed experimentally (Fig. 3C). The FEP calculations at 100ns predict values of both single and double mutant within 0.4 kcal/mol from the experiment and a cooperativity energy of -0.1 kcal/mol (same as SPR) (Fig. 3C). Thus, FEP accurately predicts the absence of cooperativity in the double mutant belonging to the Delta SARS-CoV-2 variant (Table S2). Discussion We have carried out experimental and computational studies of a series of RBD mutants located in the RBD::ACE2 interface. The choice was based in part on their frequency in infective SARS-CoV-2 variants and in part because they have been extensively studied 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63. A central goal has been to assess the ability of the FEP free energy perturbation methodology to predict the effect of mutations at the protein-protein interface on binding affinities but this in turn required an assessment of experimental accuracy. To this end we carried out SPR experiments on each mutant and compared our findings to those previously reported. We also constructed a balanced dataset of mutations with sufficient energy spread and compared performance of easily accessible computational methods to those obtained from FEP. As shown in Table 1 and discussed above, there is excellent agreement between our SPR results and previous yeast display results from the Bloom group [20]. Moreover, both sets of results are in the range reported in other studies. In particular, all methods agree on the identification of stabilizing mutations although the experimental values vary by as much as 1 kcal/mol, likely due to some of the issues discussed in Methods. As pointed out above, FEP yields the best correlation with our SPR results and, crucially, is the most effective in identifying stabilizing mutations (Table 2, Table S3). The mutations we used in our study are not a part of common ML training databases such as SKEMPI [77]; hence, our dataset for methods evaluation can be considered as “blind”. A modest level of performance for ML methods (Table 2, Table S3) is in line with a study by Bonvin and co-workers where various ML methods were tested on a blind dataset of 487 mutations in 56 complexes [78]. FoldX and Rosetta flex ddG categorize neutral and destabilizing mutations well and are at least partially successful in identifying stabilizing mutations (Table S3), though their overall performance (PCC of 0.4-0.5) is inferior to FEP (PCC of 0.8). In addition to the methods we were able to test on our data set, many other studies on RBD mutants have been reported. Table S5 lists published results on RDB::ACE2 on different sets of mutants than studied here but it is of interest to compare cases where there is overlap with previous work. Previous FEP studies 53, 55, 56 also carried out long simulations and all identify N501Y and N501T as stabilizing. TopNetTree[49], a topology-based ML method, does not perform well on our data set. The MM/GBSA study from the Bahar group [58] successfully predicts that N501Y is stabilizing but was not run on other mutants in our data set. We could not reproduce this result in our MM/GB-SA calculations likely due to protocol differences and use of different crystal structures as input. Finally, Maranas and co-workers [47] used physical interactions extracted from MM/GBSA simulations to train a neural network to predict ΔΔ G values (NN_MM-GBSA method). Since many of the mutants in our set were used in the NN_MM-GBSA training (Table S5) it is difficult to make comparisons with our own results. Despite the success of the FEP approach revealed in this work, significant challenges remain. In previous work on a variety of systems, we have demonstrated that the FEP results correlate well with experiment (PCC values on the order of 0.6-0.8) and display RMS errors in the range expected for the OPLS4 [79] and related molecular mechanics force fields (0.5-1 kcal/mol) 8, 9, 14, 66, 67, 68. Precision beyond the above cited statistics is very difficult to obtain, indeed (see Methods) experimental reproducibility errors are typically on the order of 0.4 kcal/mol, even for the high quality SPR results that we report here. As highlighted in this work, the major challenge in using FEP to predict mutation effects on protein-protein binding affinities, as opposed to small molecule binding, is the possibility of significant conformational change induced by mutation, for example if a buried charge is created by mutating a buried hydrophobic residue at the interface to one with a net charge. The key issue is whether the conformational changes required to make accurate predictions are accessible on the timescale of the FEP simulation. When the conformational changes consist primarily of side chain rearrangements and relatively minor backbone motion, FEP will typically deliver reliable results (as in N501Y). When there are significant backbone conformational changes, the accuracy of the FEP results will depend upon whether the barrier to conformational change can be surmounted on the time scale of the simulation. In terms of overall accuracy, the results obtained here are consistent, in terms of both RMSD errors and correlation coefficients, with those we have reported previously in studying HIV derived gp120-antibody binding and also more diverse sets of protein-protein complexes via FEP simulations (8,9). The most significant errors may result from structural uncertainties (as in A475V) or from the prediction of overly unfavorable free energy changes (as in Q498R) which often result from electrostatic or steric clashes or from the burial of a charge in a hydrophobic pocket [9]. However, mutations that are predicted to be highly destabilizing would be of little or no consequence in the context of binding optimization project since such mutations would be rejected in an initial screen even if converged results were obtained. The accuracy of binding free energy differences determined by FEP or other structure-based methods is crucially dependent on the quality and completeness of the protein- complex structures used in the simulations. While FEP performs well with experimentally determined structures (as shown in Table S3), achieving comparable success with modeled complexes obtained by docking or homology models of poor quality is much harder. Our study reveals that using experimental structures helps with general convergence, as multiple independent FEP runs result in SEM < 0.1 kcal/mol (as shown in Table S4). However, when FEP simulations were performed on modeled protein-peptide complexes, a much larger deviation (SEM ∼1.5 kcal/mol) was observed [79]. Although FEP is a successful method for predicting the effect of point mutations on binding free energy differences, it is not suitable for direct evaluation of deletions or insertions effects using alchemical transformations. A combination of MD and advanced sampling techniques to guarantee the adequate exploration of each relevant degree of freedom can be applied to obtain binding free energies within experimental accuracy in the process of progressive separation of proteins via the potential of mean force (PMF) 80, 81. The PMF approaches are not limited for single point mutations, but their success requires sufficiently accurate approximation of the native wild-type and mutant states. Our results suggest FEP might be used in a strategy to optimize the binding of two proteins, for example in the optimization of antibodies. FEP calculations appear to have sufficient rank ordering capability to enable prioritization of specific single residue mutations. The initial step in a design strategy would be an exhaustive screening of single mutations at the various positions across the protein-protein interface (perhaps a few hundred to a few thousand calculations, which would require relatively modest expenditure, given the steadily decreasing cost of GPU-based computation). A key advantage of FEP is that analysis of trajectories can reveal insights as to the response of the interface to different perturbations, enhanced when needed by loop modeling procedures that more efficiently sample conformational space. These insights could then be translated into the investigation of a selected subset of double and triple mutations, with the goal of achieving favorable nonadditive effects. The understanding we gained about the Q498R N501Y double mutant could not have been accomplished by any of the other methods tested here. It is important to clarify that, despite identifying cooperativity of the N501Y Q498R double mutant, the FEP results have not reached the point where the calculated values have experimental accuracy. It is then interesting to consider how a researcher interested in designing a stabilizing mutation would respond to calculated single mutant values of -1.2 kcal/mol for N501Y, +2.7 kcal/mol for Q498R and -1.4 kcal/mol for the double mutant. We would argue that the large calculated cooperativity energy of -2.9 kcal/mol, and the physical basis of this effect revealed by the simulations, would provide a strong hint that the double mutant is worth testing experimentally. Moreover, the experience we gained in this study would make that decision more likely. In conclusion, a careful exploration of a particular system of biological importance as enabled by FEP simulations would then appear to offer a way forward in many practical applications. Materials and Methods ACE2::RBD dataset. We focused on missense RBD mutations at the interface with ACE2 that occurred most frequently in the US at the beginning of 2021 or were a part of known variants of concern. Among single point mutations with a frequency above 100 as of Jan 4, 2021, only seven mutations (S477N, N439K, N501Y, L452R, Y453F, S477R and S477I) were both missense and interfacial. To expand the dataset, we added missense interfacial mutations with a lower frequency (in the range of 10-100) if mutations were stabilizing or destabilizing in the study of Bloom and co-workers [20] while nearly neutral positions were ignored. Stabilizing mutations were of interest as potentially increasing infectivity of the virus, while destabilizing mutations were a necessary addition to create a balanced dataset for a proper testing of ΔΔG predictors. The K417T and Q498R mutations were added due to occurrence in the SARS-CoV-2 variants of concern and not based on frequency counts. The K417T mutation was added due to its emergence in Brazil (as a part of the Gamma variant) at the time. The Q498R mutant was studied in the context of double mutant effects alongside N501Y due to the co-occurrence of these two mutations in Omicron variant (Table S2) and a recent study on in vitro evolution suggesting cooperativity between the two mutations [64]. Although the cumulative frequency (i.e. a number of times a given mutation has been found in the sequenced SARS-CoV-2 genomes) has changed during 2021, the majority of the most frequently observed mutations in January 2021 were still among the most recurrent mutations at the end of the year, with our data set including 16 out of 20 interfacial missense RBD mutations that had the highest frequency (>1000) on December 31, 2021 (Table S2). Frequencies of SARS-CoV-2 mutations were obtained from the Mutation Tracker resource (https://users.math.msu.edu/users/weig/SARS-CoV-2_Mutation_Tracker.html) [82] that relies on data from the GISAID database of coronavirus genomes (https://www.gisaid.org/). DIP::Dpr dataset. We included single point mutations of DIP-α, Dpr6 or Dpr10 that have been previously tested by SPR 71, 72, 73. ΔΔG calculations. All calculations presented in Table 2 and Table S3 were performed using crystal structures of highest available resolution: the SARS-CoV-2 RBD complex with ACE2 (2.45Å, PDBID: 6m0j [69]) , the DIP-α::Dpr6 complex (2.3Å, PDBID: 5EO9 [70]) or the DIP-α::Dpr10 complex (1.8Å, PDBID: 6NRQ [83]). Mutabind2 predictions were run on the https://lilab.jysw.suda.edu.cn/research/mutabind2/ webserver. mCSM-PPI2 predictions were submitted as a query to the following webserver: http://biosig.unimelb.edu.au/mcsm_ppi2/. Standalone version (http://compbio.clemson.edu/saambe_webserver/standaloneCode.zip) was used for binding affinity calculations of SAAMBE-3D. FoldX calculations were performed as described in Sergeeva et al. [72]. Rosetta flex ddG calculations were run using a standalone version (https://github.com/Kortemme-Lab/flex_ddG_tutorial) with the following parameters (considered to give optimized performance of this method [7]): nstruct = 35, max_minimization_iter = 5000, abs_score_convergence_thresh = 1.0, number_backrub_trials = 35000, backrub_trajectory_stride = 35000. The MM/GB-SA and MM/PB-SA methods rely on the end-point approximation of the wild-type (WT) and mutant (MT) states to calculate binding free energies using the following formula: ΔΔG = ΔGbinding(MT)- ΔGbinding(WT), where ΔGbinding(P1::P2) = G(P1::P2) - G(P1) - G(P2). To generate the mutated proteins, we used the “Mutate Residue” function followed by a local minimization of the mutated residue in Maestro, using the crystal structure of the WT as input. We employed the AMBER software package and the pmemd.cuda 84, 85 for minimization. The P1::P2 complex is minimized in explicit OPC water, followed by removal of water molecules and energy calculations in implicit solvent. The isolated P1 and P2 proteins are assumed to have the same conformation as in the complex. The energy calculations were carried out using MMPBSA.py [86]. The mbondi2 radii set was used in both GB and PB calculations, and the OBC2 model was used for GB calculations. We used Schrödinger software (2021-2 release) and default FEP+ protocols (implementing guidelines for protein-protein interactions published earlier 8, 9) for our predictions of binding affinity changes in the ACE2/RBD complex upon RBD mutations. The release incorporates a recently developed OPLS4 force field [79], replica exchange with solute tempering (REST) enhanced sampling methodology for mutated residues, and an improved grand canonical Monte Carlo (GCMC) protocol for sampling solvent molecules around mutated residues [87]. FEP+ is a fully physics-based model that uses explicitly represented water. During FEP+ simulations, an alchemical transformation of a wild type amino acid residue into a mutant residue is conducted, which is implemented by running a series of separate molecular dynamics simulations (“lambda windows”) with varied energy weighting. The differences between each adjacent lambda window are first calculated using a perturbative expansion and then summed up to estimate the total free energy change between a wild-type and mutant states. To enhance the convergence of the free energy calculations, our default protocols use 12 lambda windows for charge conserving mutations and 24 lambda windows for the charge changing mutations. All mutations were run for 10 ns and 100 ns (see Table S3). Calculations of double mutant effects were performed by simultaneous alchemical transformation of the two mutated residues. This procedure minimizes errors associated with a more common FEP protocol where mutations are introduced sequentially. For example, the ΔΔG (Q498R N501Y) is predicted to be -1.4 kcal/mol using the simultaneous protocol and -0.6 kcal/mol using the sequential protocol. Of note, the results from the simultaneous protocol are in better agreement with the experimental value of -1.2 kcal/mol. The FEP+ methodology automates generation of the mutated end-point, while the user provides a prepared wild-type complex as input. This process involves sampling and ranking the rotamers of the mutated residue, followed by minimization of the mutated residue. Additionally, the protein-protein complex is equilibrated before the FEP/REST2 production run. This protocol is typically effective in accurately identifying the mutated end-point throughout the FEP/REST2 trajectory, except in cases where a significant conformational change is required upon mutation. FEP+ requires GPU computing and the time required per single point mutation depends on the length of simulation (in nanoseconds), system size (in atoms), and type of mutation (with charge-changing mutations taking longer to compute compared to charge-neutral mutations as the number of lambda windows is twice as large). In the ACE2/RBD system (∼100,000 atoms including explicit waters in the solvent box), the shortest 10 ns charge-neutral mutations take less than a day, while the longest 100 ns charge-changing simulations take ∼2 weeks when complex and solvent legs are run in parallel with each simulation leg using 4GPUs. Protein Expression and Purification. The SARS-CoV-2 RBD wild-type and its mutants (residues 331-528), were cloned into the pVRC-8400 mammalian expression plasmid, with a C-terminal 6XHis-tag and an intervening HRV-3C protease cleavage site. Plasmid constructs were transfected into HEK293 cells using polyethyleneimine (Polysciences). Cell growths were harvested four days after transfection, and the secreted proteins were purified from supernatant by nickel affinity chromatography using Ni-NTA IMAC Sepharose 6 Fast Flow resin (Cytiva) followed by size exclusion chromatography on a Superdex 200 column (Cytiva) in 10 mM Tris, 150 mM NaCl, pH 7.4. A plasmid encoding ACE2 residues 1-615 (pcDNA3-sACE2(WT)-8his) was a gift from Erik Procko (Addgene plasmid # 149268 ; http://n2t.net/addgene:149268 ; RRID:Addgene_149268) [88]. This plasmid was then mutated to encode ACE2 residues 1-620, followed by a C terminal HRV-3C protease cleavage site, and an 8X HIS tag. This construct was transfected into ExpiHEK293 cells using Expifectamine according to manufacturer’s instructions. Supernatants were harvested seven days after transfection, and ACE2 was purified by nickel affinity chromatography using His60 Ni Superflow Resin (Takara) followed by size exclusion chromatography on a Superdex 200 column (Cytiva) in 10 mM Tris, 150 mM NaCl, pH 8.0. 400 ug of purified ACE2 was then digested overnight at 4 degrees with 20 units of HRV-3C protease (Thermo Scientific). Digested ACE2 was then incubated with His60 Ni Superflow Resin, which was then washed with 2 column volumes of 10 mM Tris, 150 mM NaCl, 5 mM imidazole, pH 8.0. The flow through and wash were determined by SDS-PAGE to contain cleaved ACE2, which was purified by size exclusion chromatography on a Superdex 200 column (Cytiva) in 10 mM Tris, 150 mM NaCl, pH 8.0. Surface Plasmon Resonance. SPR binding assays for monomeric ACE2 binding to RBDs were performed using a Biacore T200 biosensor, equipped with a Series S CM5 chip, at 25°C, in a running buffer of 10mM HEPES pH 7.4, 150mM NaCl, 0.1mg/mL BSA and 0.01% (v/v) Tween-20 at 25°C. Each RBD was captured through its C-terminal his-tag over an anti-his antibody surface, generated using the His-capture kit (Cytiva, MA) according to the instructions of the manufacturer. During a binding cycle, each RBDs was captured over individual flow cells at approximately 250 RU. An anti-his antibody surface was used as a reference flow cell to remove bulk shift changes from the binding signals. Monomeric ACE2 was prepared at six concentrations in running buffer using a three-fold dilution series, ranging from 1.1-270 nM. Samples were tested in order of increasing protein concentration, with each series tested in triplicate. Blank buffer cycles were performed by injecting running buffer instead of the analyte, after two ACE2 injections to remove systematic noise from the binding signal. The association and dissociation rates were each monitored for 180s and 300s respectively, at 50μL/min. Bound RBD/Fab complexes were removed using a 10s pulse of 15 mM H3PO4 at 100μL/min, thus regenerating the anti-his surface for a new cycle of recapturing of each RBD, followed by a 60s buffer wash at 100μL/min. The data was processed and fit to 1:1 interaction model using the Scrubber 2.0 (BioLogic Software). The number in brackets for each kinetic parameter represents the error of the fit. We have developed an SPR assay to measure the binding kinetics and affinities of interactions between ACE2 and wild-type or mutant RBD with the RBD captured to the chip to avoid compromised binding activity resulting from chemical immobilization or repeated surface harsh regeneration steps during the experiments. The SARS-CoV-2 RBD is a basic molecule with a pI ∼9, so capture to the chip surface will also minimize artifacts such as non-specific interactions between the positively charged RBDs and the negatively charged dextran layer of the sensor chips at physiological pH. Monomeric ACE2 was flowed over as analyte to avoid avidity effects resulting from using dimeric ACE2. Studies that have performed such experiments in both orientations (RBD tethered to the surface vs in solution as an analyte) showed that using RBD as an analyte yielded affinities that were approximately three-fold stronger for the wild type RBD/ACE2 interaction, suggesting the presence of such non-specific interactions [24]. In our SPR experiments we have determined the KD for wild type RBD binding to monomeric ACE2 is 162.9 nM (Fig. S1), consistent with similar KDs reported from other groups that have used similar methodologies to perform their biosensor-based measurements 21, 89, 90. Fig. S1A shows the binding kinetics for interactions of mutant RBD proteins with ACE2 and Fig. S1B shows the kinetic parameters along with affinities calculated for each binding interaction, while Table 1 and Fig. 3B list experimental changes in binding affinities (ΔΔG=RTln(KD(MT)/KD(WT))) when RBD is mutated. Experimental reproducibility errors in our SPR data (association and dissociation rate constants) is expected to be ∼15-20% according to previous estimates based on multiple independent measurements[72]; a 15-20% error corresponds to ∼0.4 kcal/mol experimental error in the ΔΔG(SPR) values reported in this study. Of 23 single point RBD mutations probed, four mutations were identified as stabilizing: N501Y, Y453F, S477N and N501T (Table 1). Differences in Experimental Protocol Affecting Binding Affinity Changes. Previously reported experimental binding affinity changes upon RBD mutation of the ACE2/RBD complex span different choice of protein constructs and orientation of molecules used in the binding assays. The differences in the constructs lie in the choice of monomeric vs. multimeric forms of interacting proteins and selection of protein domain boundaries. Some studies relied on monomeric ACE2 and RBD 21, 22, 24, 25, 27, 28, 29, 30, whereas others used at least one of the molecule in a multimeric form (trimeric spike, dimeric ACE2 or monomeric ACE2 fused to a dimeric Fc tag) 20, 25, 26, 31, 32, 64. RBD domain starts at residue 333 and ends at residue 526. It is common that constructs used in studies flank the RBD construct with a few residues before and after the domain boundary (e.g. 331-528 (this study), 331-531 [20], 328-531 21, 22) though some studies use constructs where such flanking regions are too long so they could result in non-specific binding (especially when containing unpaired cysteine residues, e.g. RBD 319-591 [25]). Poor selection of protein domain boundaries can affect protein folding/integrity when a construct has incomplete domain sequence (e.g. RBD 343-532 construct is missing a β-strand [23]). Orientation of molecules in the binding experiments (which molecule is tethered to the chip in SPR) can affect both absolute and relative binding affinities (discussed in SPR methods). Studies using a highly positively charged RBD molecule as analyte and ACE2 immobilized on a chip 27, 28, 29, 30, 31, 32 can be affected by non-specific binding of RBD to the chip. For example, using RBD as analyte in experiments measuring binding affinity of Alpha, Beta, Gamma, and Delta variants [91] results in stronger binding (by 0.4-0.6 kcal/mol) compared to a setup minimizing non-specific binding by immobilizing RBD on a chip [92]. We used the latter as a reference to assess performance of ΔΔG on predicting double mutant effects of the Delta variant (Fig. 3C). Data Availability. All study data are included in the article and/or supporting information. CRediT authorship contribution statement Alina P. Sergeeva: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Data curation, Software, Visualization, Writing – original draft, Writing – review & editing. Phinikoula S. Katsamba: Validation, Formal analysis, Investigation, Visualization. Junzhuo Liao: Formal analysis, Investigation, Data curation, Software. Jared M. Sampson: Resources. Fabiana Bahna: Resources. Seetha Mannepalli: Resources. Nicholas C. Morano: Resources. Lawrence Shapiro: Conceptualization, Supervision, Funding acquisition. Richard A. Friesner: Conceptualization, Supervision, Funding acquisition, Writing – review & editing. Barry Honig: Conceptualization, Supervision, Funding acquisition, Writing – review & editing. Data availability Data will be made available on request. Acknowledgments The work was supported by Bill and Melinda Gates Foundation, INV-016167 (to B.H., L.S., and R. F. We thank Peter Kwong and his lab for providing the monomeric ACE2 protein used in a subset of our experiments. We thank Stephen J. Trudeau for discussion on statistical analysis. ==== Refs References 1 Pahari S. Li G. Murthy A.K. Liang S. Fragoza R. Yu H. SAAMBE-3D: Predicting Effect of Mutations on Protein-Protein Interactions Int J Mol Sci. 21 2020 2563 32272725 2 Pires D.E. Ascher D.B. Blundell T.L. mCSM: predicting the effects of mutations in proteins using graph-based signatures Bioinformatics. 30 2014 335 342 24281696 3 Rodrigues C.H.M. Myung Y. Pires D.E.V. Ascher D.B. mCSM-PPI2: predicting the effects of mutations on protein-protein interactions Nucleic Acids Res. 47 2019 W338 W344 31114883 4 Li M. Simonetti F.L. Goncearenco A. Panchenko A.R. MutaBind estimates and interprets the effects of sequence variants on protein-protein interactions Nucleic Acids Res. 44 2016 W494 W501 27150810 5 Zhang N. Chen Y. Lu H. Zhao F. Alvarez R.V. Goncearenco A. MutaBind2: Predicting the Impacts of Single and Multiple Mutations on Protein-Protein Interactions iScience. 23 2020 100939 6 Dehouck Y. Kwasigroch J.M. Rooman M. Gilis D. BeAtMuSiC: Prediction of changes in protein-protein binding affinity on mutations Nucleic Acids Res. 41 2013 W333 W339 23723246 7 Barlow KA, S OC, Thompson S, Suresh P, Lucas JE, Heinonen M, et al. Flex ddG: Rosetta Ensemble-Based Estimation of Changes in Protein-Protein Binding Affinity upon Mutation. J Phys Chem B. 2018;122:5389-99. 8 Clark A.J. Gindin T. Zhang B. Wang L. Abel R. Murret C.S. Free Energy Perturbation Calculation of Relative Binding Free Energy between Broadly Neutralizing Antibodies and the gp120 Glycoprotein of HIV-1 J Mol Biol. 429 2017 930 947 27908641 9 Clark A.J. Negron C. Hauser K. Sun M. Wang L. Abel R. Relative Binding Affinity Prediction of Charge-Changing Sequence Mutations with FEP in Protein-Protein Interfaces J Mol Biol. 431 2019 1481 1493 30776430 10 Guerois R. Nielsen J.E. Serrano L. Predicting changes in the stability of proteins and protein complexes: a study of more than 1000 mutations J Mol Biol. 320 2002 369 387 12079393 11 Schymkowitz J. Borg J. Stricher F. Nys R. Rousseau F. Serrano L. The FoldX web server: an online force field Nucleic Acids Res. 33 2005 W382 W388 15980494 12 Geng C. Xue L.C. Roel-Touris J. Bonvin A.M.J.J. Finding the ΔΔG spot: Are predictors of binding affinity changes upon mutations in protein–protein interactions ready for it? WIREs Computational Molecular Science. 9 2019 e1410 13 Geng C. Vangone A. Bonvin A.M.J.J. Exploring the interplay between experimental methods and the performance of predictors of binding affinity change upon mutations in protein complexes Protein Engineering, Design and Selection. 29 2016 291 299 14 Abel R. Wang L. Harder E.D. Berne B.J. Friesner R.A. Advancing Drug Discovery through Enhanced Free Energy Calculations Acc Chem Res. 50 2017 1625 1632 28677954 15 Katz D. DiMattia M.A. Sindhikara D. Li H. Abraham N. Leffler A.E. Potency- and Selectivity-Enhancing Mutations of Conotoxins for Nicotinic Acetylcholine Receptors Can Be Predicted Using Accurate Free-Energy Calculations Mar Drugs. 19 2021 16 Katz D. Sindhikara D. DiMattia M. Leffler A.E. Potency-Enhancing Mutations of Gating Modifier Toxins for the Voltage-Gated Sodium Channel NaV1.7 Can Be Predicted Using Accurate Free-Energy Calculations Toxins (Basel) 2021 13 35050990 17 La Serra M.A. Vidossich P. Acquistapace I. Ganesan A.K. De Vivo M. Alchemical Free Energy Calculations to Investigate Protein-Protein Interactions: the Case of the CDC42/PAK1 Complex J Chem Inf Model. 62 2022 3023 3033 35679463 18 Nandigrami P. Szczepaniak F. Boughter C.T. Dehez F. Chipot C. Roux B. Computational Assessment of Protein-Protein Binding Specificity within a Family of Synaptic Surface Receptors J Phys Chem B. 126 2022 7510 7527 35787023 19 Patel D. Patel J.S. Ytreberg F.M. Implementing and Assessing an Alchemical Method for Calculating Protein-Protein Binding Free Energy J Chem Theory Comput. 17 2021 2457 2464 33709712 20 Starr T.N. Greaney A.J. Hilton S.K. Ellis D. Crawford K.H.D. Dingens A.S. Deep Mutational Scanning of SARS-CoV-2 Receptor Binding Domain Reveals Constraints on Folding and ACE2 Binding Cell. 182 1295–310 2020 e20 21 McCallum M. Bassi J. De Marco A. Chen A. Walls A.C. Di Iulio J. SARS-CoV-2 immune evasion by the B.1.427/B.1.429 variant of concern Science. 373 2021 648 654 34210893 22 McCallum M. Walls A.C. Sprouse K.R. Bowen J.E. Rosen L.E. Dang H.V. Molecular basis of immune evasion by the Delta and Kappa SARS-CoV-2 variants Science. 374 2021 1621 1626 34751595 23 Deshpande A. Harris B.D. Martinez-Sobrido L. Kobie J.J. Walter M.R. Epitope Classification and RBD Binding Properties of Neutralizing Antibodies Against SARS-CoV-2 Variants of Concern Front Immunol. 12 2021 691715 24 Thomson E.C. Rosen L.E. Shepherd J.G. Spreafico R. da Silva F.A. Wojcechowskyj J.A. Circulating SARS-CoV-2 spike N439K variants maintain fitness while evading antibody-mediated immunity Cell. 184 1171–87 2021 e20 25 Tian F. Tong B. Sun L. Shi S. Zheng B. Wang Z. N501Y mutation of spike protein in SARS-CoV-2 strengthens its binding to receptor ACE2 Elife. 2021 10 26 Gobeil S.M. Janowska K. McDowell S. Mansouri K. Parks R. Stalls V. Effect of natural mutations of SARS-CoV-2 on spike structure, conformation, and antigenicity Science. 373 2021 27 Laffeber C. de Koning K. Kanaar R. Lebbink J.H.G. Experimental Evidence for Enhanced Receptor Binding by Rapidly Spreading SARS-CoV-2 Variants J Mol Biol. 433 2021 167058 28 Liu H. Wei P. Zhang Q. Chen Z. Aviszus K. Downing W. 501Y.V2 and 501Y.V3 variants of SARS-CoV-2 lose binding to bamlanivimab in vitro MAbs. 13 2021 1919285 34074219 29 Liu H. Zhang Q. Wei P. Chen Z. Aviszus K. Yang J. The basis of a more contagious 501Y.V1 variant of SARS-CoV-2 Cell Res. 31 2021 720 722 33893398 30 Barton M.I. MacGowan S.A. Kutuzov M.A. Dushek O. Barton G.J. van der Merwe P.A. Effects of common mutations in the SARS-CoV-2 Spike RBD and its ligand, the human ACE2 receptor on binding affinity and kinetics Elife. 2021 10 31 Raghu D. Hamill P. Banaji A. McLaren A. Hsu Y.T. Assessment of the binding interactions of SARS-CoV-2 spike glycoprotein variants J Pharm Anal. 2021 32 Han P. Su C. Zhang Y. Bai C. Zheng A. Qiao C. Molecular insights into receptor binding of recent emerging SARS-CoV-2 variants Nat Commun. 12 2021 6103 34671049 33 Xue T. Wu W. Guo N. Wu C. Huang J. Lai L. Single point mutations can potentially enhance infectivity of SARS-CoV-2 revealed by in silico affinity maturation and SPR assay RSC Advances. 11 2021 14737 14745 35423963 34 Mannar D. Saville J.W. Zhu X. Srivastava S.S. Berezuk A.M. Zhou S. Structural analysis of receptor binding domain mutations in SARS-CoV-2 variants of concern that modulate ACE2 and antibody binding Cell Rep. 37 2021 110156 35 Saville J.W. Mannar D. Zhu X. Srivastava S.S. Berezuk A.M. Demers J.P. Structural and biochemical rationale for enhanced spike protein fitness in delta and kappa SARS-CoV-2 variants Nat Commun. 13 2022 742 35136050 36 Zhu X. Mannar D. Srivastava S.S. Berezuk A.M. Demers J.P. Saville J.W. Cryo-electron microscopy structures of the N501Y SARS-CoV-2 spike protein in complex with ACE2 and 2 potent neutralizing antibodies PLoS Biol. 19 2021 e3001237 33914735 37 Cai Y. Zhang J. Xiao T. Lavine C.L. Rawson S. Peng H. Structural basis for enhanced infectivity and immune evasion of SARS-CoV-2 variants Science. 373 2021 642 648 34168070 38 Zhang J. Xiao T. Cai Y. Lavine C.L. Peng H. Zhu H. Membrane fusion and immune evasion by the spike protein of SARS-CoV-2 Delta variant Science. 374 2021 1353 1360 34698504 39 Yang T-J, Yu P-Y, Chang Y-C, Chang N-E, Tsai Y-X, Liang K-H, et al. Structure-activity relationships of B.1.617 and other SARS-CoV-2 spike variants. bioRxiv. 2021:2021.09.12.459978. 40 Bayarri-Olmos R. Jarlhelt I. Johnsen L.B. Hansen C.B. Helgstrand C. Rose Bjelke J. Functional Effects of Receptor-Binding Domain Mutations of SARS-CoV-2 B.1.351 and P.1 Variants Front Immunol. 12 2021 757197 41 Bayarri-Olmos R. Johnsen L.B. Idorn M. Reinert L.S. Rosbjerg A. Vang S. The alpha/B.1.1.7 SARS-CoV-2 variant exhibits significantly higher affinity for ACE-2 and requires lower inoculation doses to cause disease in K18-hACE2 mice Elife. 2021 10 42 Bayarri-Olmos R. Rosbjerg A. Johnsen L.B. Helgstrand C. Bak-Thomsen T. Garred P. The SARS-CoV-2 Y453F mink variant displays a pronounced increase in ACE-2 affinity but does not challenge antibody neutralization J Biol Chem. 296 2021 100536 43 Supasa P. Zhou D. Dejnirattisai W. Liu C. Mentzer A.J. Ginn H.M. Reduced neutralization of SARS-CoV-2 B.1.1.7 variant by convalescent and vaccine sera Cell. 184 2201–11 2021 e7 44 Yuan M. Huang D. Lee C.D. Wu N.C. Jackson A.M. Zhu X. Structural and functional ramifications of antigenic drift in recent SARS-CoV-2 variants Science. 373 2021 818 823 34016740 45 Lopez E. Haycroft E.R. Adair A. Mordant F.L. O'Neill M.T. Pymm P. Simultaneous evaluation of antibodies that inhibit SARS-CoV-2 variants via multiplex assay. JCI Insight. 2021 6 46 Collier D.A. De Marco A. Ferreira I. Meng B. Datir R.P. Walls A.C. Sensitivity of SARS-CoV-2 B.1.1.7 to mRNA vaccine-elicited antibodies Nature. 593 2021 136 141 33706364 47 Chen C. Boorla V.S. Banerjee D. Chowdhury R. Cavener V.S. Nissly R.H. Computational prediction of the effect of amino acid changes on the binding affinity between SARS-CoV-2 spike RBD and human ACE2 Proc Natl Acad Sci U S A. 118 2021 48 Liu X. Luo Y. Li P. Song S. Peng J. Deep geometric representations for modeling effects of mutations on protein-protein binding affinity PLoS Comput Biol. 17 2021 e1009284 34347784 49 Chen J. Wang R. Wang M. Wei G.W. Mutations Strengthened SARS-CoV-2 Infectivity J Mol Biol. 432 2020 5212 5226 32710986 50 Strokach A. Lu T.Y. Kim P.M. ELASPIC2 (EL2): Combining Contextualized Language Models and Graph Neural Networks to Predict Effects of Mutations J Mol Biol. 433 2021 166810 51 Laurini E. Marson D. Aulic S. Fermeglia A. Pricl S. Computational Mutagenesis at the SARS-CoV-2 Spike Protein/Angiotensin-Converting Enzyme 2 Binding Interface: Comparison with Experimental Evidence ACS Nano. 15 2021 6929 6948 33733740 52 Laurini E. Marson D. Aulic S. Fermeglia M. Pricl S. Computational Alanine Scanning and Structural Analysis of the SARS-CoV-2 Spike Protein/Angiotensin-Converting Enzyme 2 Complex ACS Nano. 14 2020 11821 11830 32833435 53 Fratev F. N501Y and K417N Mutations in the Spike Protein of SARS-CoV-2 Alter the Interactions with Both hACE2 and Human-Derived Antibody: A Free Energy of Perturbation Retrospective Study J Chem Inf Model. 61 2021 6079 6084 34806876 54 Fratev F. The High Transmission of SARS-CoV-2 Omicron (B.1.1.529) Variant is Not Only Due to Its hACE2 binding: A Free Energy of Perturbation Study. bioRxiv. 2021:2021.12.04.471246. 55 Pavlova A. Zhang Z. Acharya A. Lynch D.L. Pang Y.T. Mou Z. Machine Learning Reveals the Critical Interactions for SARS-CoV-2 Spike Protein Binding to ACE2 J Phys Chem Lett. 12 2021 5494 5502 34086459 56 Luan B. Wang H. Huynh T. Enhanced binding of the N501Y-mutated SARS-CoV-2 spike protein to the human ACE2 receptor: insights from molecular dynamics simulations FEBS Lett. 595 2021 1454 1461 33728680 57 Khan A. Zia T. Suleman M. Khan T. Ali S.S. Abbasi A.A. Higher infectivity of the SARS-CoV-2 new variants is associated with K417N/T, E484K, and N501Y mutants: An insight from structural data J Cell Physiol. 236 2021 7045 7057 33755190 58 Cheng M.H. Krieger J.M. Banerjee A. Xiang Y. Kaynak B. Shi Y. Impact of new variants on SARS-CoV-2 infectivity and neutralization: A molecular assessment of the alterations in the spike-host protein interactions iScience. 25 2022 103939 59 Buratto D. Saxena A. Ji Q. Yang G. Pantano S. Zonta F. Rapid Assessment of Binding Affinity of SARS-COV-2 Spike Protein to the Human Angiotensin-Converting Enzyme 2 Receptor and to Neutralizing Biomolecules Based on Computer Simulations Front Immunol. 12 2021 730099 60 Villoutreix B.O. Calvez V. Marcelin A.G. Khatib A.M. In Silico Investigation of the New UK (B.1.1.7) and South African (501Y.V2) SARS-CoV-2 Variants with a Focus at the ACE2-Spike RBD Interface Int J Mol Sci. 2021 22 35008458 61 Wu L. Zhou L. Mo M. Liu T. Wu C. Gong C. SARS-CoV-2 Omicron RBD shows weaker binding affinity than the currently dominant Delta variant to human ACE2 Signal Transduct Target Ther. 7 2022 8 34987150 62 Gan H.H. Twaddle A. Marchand B. Gunsalus K.C. Structural Modeling of the SARS-CoV-2 Spike/Human ACE2 Complex Interface can Identify High-Affinity Variants Associated with Increased Transmissibility J Mol Biol. 433 2021 167051 63 Bozdaganyan M.E. Shaitan K.V. Kirpichnikov M.P. Sokolova O.S. Orekhov P.S. Computational Analysis of Mutations in the Receptor-Binding Domain of SARS-CoV-2 Spike and Their Effects on Antibody Binding Viruses. 14 2022 64 Zahradnik J. Marciano S. Shemesh M. Zoler E. Harari D. Chiaravalli J. SARS-CoV-2 variant prediction and antiviral drug design are enabled by RBD in vitro evolution Nat Microbiol. 6 2021 1188 1198 34400835 65 Liu H. Wei P. Zhang Q. Aviszus K. Linderberger J. Yang J. The Lambda variant of SARS-CoV-2 has a better chance than the Delta variant to escape vaccines bioRxiv. 2021 66 Harder E. Damm W. Maple J. Wu C. Reboul M. Xiang J.Y. OPLS3: A Force Field Providing Broad Coverage of Drug-like Small Molecules and Proteins J Chem Theory Comput. 12 2016 281 296 26584231 67 Roos K. Wu C. Damm W. Reboul M. Stevenson J.M. Lu C. OPLS3e: Extending Force Field Coverage for Drug-Like Small Molecules J Chem Theory Comput. 15 2019 1863 1874 30768902 68 Wang L. Wu Y. Deng Y. Kim B. Pierce L. Krilov G. Accurate and reliable prediction of relative ligand binding potency in prospective drug discovery by way of a modern free-energy calculation protocol and force field J Am Chem Soc. 137 2015 2695 2703 25625324 69 Lan J. Ge J. Yu J. Shan S. Zhou H. Fan S. Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor Nature. 581 2020 215 220 32225176 70 Carrillo R.A. Ozkan E. Menon K.P. Nagarkar-Jaiswal S. Lee P.T. Jeon M. Control of Synaptic Connectivity by a Network of Drosophila IgSF Cell Surface Proteins Cell. 163 2015 1770 1782 26687361 71 Cosmanescu F. Katsamba P.S. Sergeeva A.P. Ahlsen G. Patel S.D. Brewer J.J. Neuron-Subtype-Specific Expression, Interaction Affinities, and Specificity Determinants of DIP/Dpr Cell Recognition Proteins Neuron 2018 72 Sergeeva A.P. Katsamba P.S. Cosmanescu F. Brewer J.J. Ahlsen G. Mannepalli S. DIP/Dpr interactions and the evolutionary design of specificity in protein families Nat Commun. 11 2020 2125 32358559 73 Xu S. Sergeeva A.P. Katsamba P.S. Mannepalli S. Bahna F. Bimela J. Affinity requirements for control of synaptic targeting and neuronal cell survival by heterophilic IgSF cell adhesion molecules Cell Rep. 39 2022 110618 74 Liu Y. Liu J. Plante K.S. Plante J.A. Xie X. Zhang X. The N501Y spike substitution enhances SARS-CoV-2 infection and transmission Nature 2021 75 Burkholz S. Pokhrel S. Kraemer B.R. Mochly-Rosen D. Carback R.T. 3rd Hodge T. Paired SARS-CoV-2 spike protein mutations observed during ongoing SARS-CoV-2 viral transfer from humans to minks and back to humans Infect Genet Evol. 93 2021 104897 76 Munnink B.B.O. Sikkema R.S. Nieuwenhuijse D.F. Molenaar R.J. Munger E. Molenkamp R. Transmission of SARS-CoV-2 on mink farms between humans and mink and back to humans Science. 371 2021 172 177 33172935 77 Jankauskaite J. Jimenez-Garcia B. Dapkunas J. Fernandez-Recio J. Moal I.H. SKEMPI 2.0: An updated benchmark of changes in protein-protein binding energy, kinetics and thermodynamics upon mutation Bioinformatics. 2018 78 Geng C. Vangone A. Folkers G.E. Xue L.C. Bonvin A. iSEE: Interface structure, evolution, and energy-based machine learning predictor of binding affinity changes upon mutations Proteins 2018 79 Lu C. Wu C. Ghoreishi D. Chen W. Wang L. Damm W. OPLS4: Improving Force Field Accuracy on Challenging Regimes of Chemical Space J Chem Theory Comput. 17 2021 4291 4300 34096718 80 Blazhynska M. Coderc G. de Lacam E. Chen H. Roux B. Chipot C. Hazardous Shortcuts in Standard Binding Free Energy Calculations J Phys Chem Lett. 13 2022 6250 6258 35771686 81 Fu H. Chen H. Blazhynska M. Coderc G. de Lacam E. Szczepaniak F. Pavlova A. Accurate determination of protein:ligand standard binding free energies from molecular dynamics simulations Nat Protoc. 17 2022 1114 1141 35277695 82 Wang R. Chen J. Gao K. Hozumi Y. Yin C. Wei G.W. Analysis of SARS-CoV-2 mutations in the United States suggests presence of four substrains and novel variants Commun Biol. 4 2021 228 33589648 83 Cheng S. Ashley J. Kurleto J.D. Lobb-Rabe M. Park Y.J. Carrillo R.A. Molecular basis of synaptic specificity by immunoglobulin superfamily receptors in Drosophila Elife. 8 2019 84 Gotz A.W. Williamson M.J. Xu D. Poole D. Le Grand S. Walker R.C. Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 1. Generalized Born J Chem Theory Comput. 8 2012 1542 1555 22582031 85 Salomon-Ferrer R. Gotz A.W. Poole D. Le Grand S. Walker R.C. Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 2. Explicit Solvent Particle Mesh Ewald J Chem Theory Comput. 9 2013 3878 3888 26592383 86 Miller B.R. 3rd McGee T.D. Jr. Swails J.M. Homeyer N. Gohlke H. Roitberg A.E. MMPBSA.py: An Efficient Program for End-State Free Energy Calculations J Chem Theory Comput. 8 2012 3314 3321 26605738 87 Ross G.A. Russell E. Deng Y. Lu C. Harder E.D. Abel R. Enhancing Water Sampling in Free Energy Calculations with Grand Canonical Monte Carlo J Chem Theory Comput. 16 2020 6061 6076 32955877 88 Chan K.K. Dorosky D. Sharma P. Abbasi S.A. Dye J.M. Kranz D.M. Engineering human ACE2 to optimize binding to the spike protein of SARS coronavirus 2 Science. 369 2020 1261 1265 32753553 89 Meng B. Abdullahi A. Ferreira I. Goonawardane N. Saito A. Kimura I. Altered TMPRSS2 usage by SARS-CoV-2 Omicron impacts infectivity and fusogenicity Nature. 603 2022 706 714 35104837 90 Zhou T. Teng I.T. Olia A.S. Cerutti G. Gorman J. Nazzari A. Structure-Based Design with Tag-Based Purification and In-Process Biotinylation Enable Streamlined Development of SARS-CoV-2 Spike Molecular Probes Cell Rep. 33 2020 108322 91 Lan J. Chen P. Liu W. Ren W. Zhang L. Ding Q. Structural insights into the binding of SARS-CoV-2, SARS-CoV, and hCoV-NL63 spike receptor-binding domain to horse ACE2 Structure. 30 1432–42 2022 e4 92 Li L. Liao H. Meng Y. Li W. Han P. Liu K. Structural basis of human ACE2 higher binding affinity to currently circulating Omicron SARS-CoV-2 sub-variants BA.2 and BA.1.1 Cell. 185 2952–60 2022 e10
PMC010xxxxxx/PMC10286573.txt	==== Front Case Studies in Chemical and Environmental Engineering 2666-0164 2666-0164 Published by Elsevier Ltd. S2666-0164(23)00115-9 10.1016/j.cscee.2023.100410 100410 Case Report Effects of fine particulate matter (PM2.5) and meteorological factors on the daily confirmed cases of COVID-19 in Bangkok during 2020–2021, Thailand Sangkham Sarawut a∗1 Islam Md Aminul bc1 Sarndhong Kritsada d Vongruang Patipat ae Hasan Mohammad Nayeem f Tiwari Ananda g Bhattacharya Prosun h∗∗ a Department of Environmental Health, School of Public Health, University of Phayao, Phayao, 56000, Thailand b COVID-19 Diagnostic Lab, Department of Microbiology, Noakhali Science and Technology University, Noakhali, 3814, Bangladesh c Advanced Molecular Lab, Department of Microbiology, President Abdul Hamid Medical College, Karimganj, Kishoreganj, Bangladesh d Department of Community Health, School of Public Health, University of Phayao, Phayao, 56000, Thailand e Atmospheric Pollution and Climate Change Research Unit, School of Energy and Environment, University of Phayao, Phayao, 56000, Thailand f Department of Statistics, Shahjalal University of Science & Technology, Sylhet, Bangladesh g Expert Microbiology Unit, Department of Health Security, Finnish Institute for Health and Welfare, 70701, Kuopio, Finland h COVID-19 Research, Department of Sustainable Development, Environmental Science and Engineering, KTH Royal Institute of Technology, Teknikringen 10B, SE, 10044, Stockholm, Sweden ∗ Corresponding author. ∗∗ Corresponding author. 1 Authors are contributed equally and share the first authorship. 22 6 2023 22 6 2023 1004102 5 2023 20 6 2023 21 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The ongoing global pandemic caused by the SARS-CoV-2 virus, known as COVID-19, has disrupted public health, businesses, and economies worldwide due to its widespread transmission. While previous research has suggested a possible link between environmental factors and increased COVID-19 cases, the evidence regarding this connection remains inconclusive. The purpose of this research is to determine whether or not there is a connection between the presence of fine particulate matter (PM2.5) and meteorological conditions and COVID-19 infection rates in Bangkok, Thailand. The study employs a statistical method called Generalized Additive Model (GAM) to find a positive and non-linear association between RH, AH, and R and the number of verified COVID-19 cases. The impacts of the seasons (especially summer) and rainfall on the trajectory of COVID-19 cases were also highlighted, with an adjusted R-square of 0.852 and a deviance explained of 85.60%, both of which were statistically significant (p < 0.05). The study results assist in preventing the future seasonal spread of COVID-19, and public health authorities may use these findings to make informed decisions and assess their policies. Graphical abstract Image 1 Keywords Fine particulate matter PM2.5 Meteorological factors Rianfall COVID-19 pandemic Bangkok Abbreviations GAM Generalized Additive Model PM2.5 Particulate Matter 2.5 WHO World Health Organization DDC Department of Disease Control PCD The Pollution Control Department ACE-2 Angiotensin-converting enzyme 2 RH Relative humidity AH Absolute humidity BMA Bangkok Metropolitan Administration ==== Body pmc1 Introduction SARS-CoV-2, commonly known as COVID-19, is a novel coronavirus disease that was first reported in 2019. It has emerged as one of the most severe global influenza pandemics, resulting in over 636 million confirmed cases and causing approximately 6 million deaths worldwide [1,2]. This highly contagious virus has rapidly spread across the globe within a short period, posing a significant public health challenge [[3], [4], [5]]. SARS-CoV-2 infections present with a range of potential influenza-like symptoms, and the incubation period typically spans from 2 to 12 days. However, this virus's exact mode of transmission for this virus is still not fully understood [6,7]. Recent reports from the World Health Organization (WHO) have identified the Delta variant (B.1.617.2) and the Omicron variant (B.1.1.529) as highly transmissible variants of the coronavirus [2,[8], [9], [10], [11]]. The first formal notification of COVID-19 verified cases by the Department of Disease Control (DDC) in Thailand was issued on January 12, 2020 [12]. According to several studies, long-term exposure to specific air pollutants makes COVID-19 more severe and complicated for those who have contracted the disease to recover from Refs. [[13], [14], [15]]. Wu et al. (2020) observed in earlier investigations that an increase in PM2.5 by only 1 g/m3 was related to an 8% rise in COVID-19 mortality (95% CI: 2 %-15%) [16]. In addition to the SARS-CoV-2 virus, COVID-19 may be more severe because of chronic lung inflammation, which is exacerbated by long-term exposure to outdoor air pollutants [17]. Populations exposed to high levels of PM2.5 particles are therefore more prone to experiencing respiratory conditions that are conducive to infectious pathogens [18]. Additionally, to evidence suggests prolonged exposure to ambient air pollution may increase COVID-19 mortality [16,19]. Weather-related air pollution could modify the host's immunity and the virus's survival ability, which could influence the patterns of severe acute respiratory syndrome viral outbreaks [20]. Numerous investigations conducted worldwide have revealed that COVID-19-induced humidity plays a pivotal function in morbidity and death [21]. Furthermore, the air quality index (AQI), NO2, PM2.5, and temperature are parameters that may endorse the ongoing transmission of COVID-19 [22]. A previous study discovered an association between COVID-19 cases and average changes in temperature worldwide [23]. The transmission of SARS-CoV-2 and the consequences of droplet aerosol released into the environment due to weather and air pollution are two examples of phenomena for which little is known. Marquès and Domingo (2022) noted the challenges in drawing definitive conclusions regarding the relationship between meteorological factors, air pollution, and daily COVID-19 cases due to the variability in terrain and seasons across different countries [24]. However, a brief study by Sangkham et al. (2021) in the Bangkok Metropolitan Area established a connection between the average annual number of COVID-19 cases and the local climate variables of temperature, wind speed (WS), relative humidity (RH), and absolute humidity (AH) [25]. Notably, their study had no observed relationship between rainfall and air pollution. The World Health Organization (WHO) has established guidelines for PM10 and PM2.5, two air pollutants known to harm the respiratory system, contribute to COVID-19 transmission, and increase health risks during the pandemic. Of particular concern are PM2.5 particles, which can penetrate the lower respiratory system. However, it is unclear how PM2.5 and meteorological conditions affect the pathogens in the environment [26]. There is inconclusive scientific evidence and challenges in studying the impact of air pollution and meteorological variables on the daily number of confirmed COVID-19 cases and the study correlation between climatic parameters and PM2.5 with verified daily COVID-19 instances in BKK has never been analyzed in a nonlinear manner before. Therefore, evaluated the focus on consequences throughout 2020–2021 and provided a hypothesis related to the analysis of the association between meteorological conditions, PM2.5, and the number of daily COVID-19 cases in Bangkok. It could have important ramifications for pandemic containment both during and after. 2 Materials and methods 2.1 Study area and data collection For this study, COVID-19-positive cases were obtained from the Bangkok Metropolitan Administration (BMA) and reported by the Department of Disease Control (DDC), Ministry of Public Health of Thailand. The data collection period spanned from 1 January 2020 to 31 December 2021 [27]. Daily weather data, including mean temperature (๐C), relative humidity (%), pressure (mmHg), wind speed (m/s), rainfall (mm), and PM2.5 (μg/m3), were provided by the Pollution Control Department (PCD), Ministry of Natural Resources and Environment. The PCD, which is responsible for Thailand's air quality monitoring stations, is the source of the data on weather and pollution levels the PCD operates. Fig. 1 shows the study area, monitoring locations, and ID name (02T, 05T, 11T, 10T). The hourly observation data includes PM2.5, relative humidity, wind speed, and temperature (T).Fig. 1 The location of the study area and the Air Quality Monitoring Station in BKK, The Pollution Control Department (PCD), Ministry of Natural Resources, and Environment. Fig. 1 2.2 Statistical analysis The Generalized Additive Model (GAM) accommodates the generalized additive model for parametric, nonparametric regression, and smoothing [[28], [29], [30], [31], [32]]. The effect of weather, air pollution, and health outcomes were investigated using a log-linear GAM [5]. Our fundamental model is a Gaussian response generalized additive model. The first step entailed building the fundamental models, which included both PM2.5 and meteorological parameters (temperature (T), RH, AH, WS, air pressure, and rainfall). Next, the variables were modified to consider the environment and individual characteristics, like wind speed. A GAM with a Gaussian distribution family was used in this study to assess the daily average lag effect (Lag0-7, Lag0-14, and Lag0-21) of mean weather and air pollution on daily COVID-19 confirmed cases [[33], [34], [35]]. Third, the day-of-week and the penalized smoothing spline function were included to manage the time trend, as shown in the following Eq. (1):(1) log(Yt)=α+βXt1+PM2.5+T+RH+AH+P+R+D(time,df) Where, The observation date is t is the day of the observation. Given the quantity of daily new cases in the Bangkok Metropolitan Area is 0, Yt is the cases that were confirmed each day-on-day t plus one. The intercept is α ; the regression coefficient is β; the weather factors on the day is Xt1; the temperature (T); relative humidity is RH; absolute humidity is AH; air pressure is P; rainfall is R; and D(Time,df), the date of the observation is shown as Time; the degree of freedom is df which is based on the penalized smoothing spline; s(mean) refers to the smoother. To account for any potential misalignment of a single lag day exposure, the lag effects of weather and air pollutants on daily new cases of COVID-19 were then considered using single lag days (Lag0-7, 0-14, 0-21), and the cumulative effects of average exposure over multiple days were then assessed using additional analyses (Lag0-7, Lag0-14, Lag0-21) [36]. The study employed the R programme (version 4.2.1) to conduct our analysis using GAM. Statistical significance was defined as p < 0.05 in these tests. 3 Results and discussion 3.1 Descriptive analysis The study's findings indicated air and weather pollution in Bangkok Metropolitan from the COVID-19 descriptive statistics during the period of 1 January 2020 to 31 December 2021, (731 days), and Fig. 2 . Shows daily PM2.5, meteorological parameter, and COVID-19 cases in Bangkok Metropolitan Regions. The result shows that there were 686.64 cases on average per BBK, and daily average values for temperature, relative humidity, absolute humidity, wind speed, pressure, and, rain according to the meteorological data found 29.46 ๐C, 66.08%, 19.53 g/m3, 0.87 m/s, where PM2.5 were 21.06 μg/m3, respectively (Table S1). The distribution of daily COVID-19 cases, PM2.5, and weather-related variables by year and season in BKK is shown in Fig. 3 . The average temperature in BKK in 2020 was 29.75° Celsius, there were more than 6.32 confirmed cases on average, and the mean absolute humidity was 19.85 g per cubic metre. A rise in daily COVID-19 cases has been facing the pandemic due to a coronavirus variation to Omicron during July 2021 in BKK, with the mean confirmed cases in 2021 being 1144.96, absolute humidity being 19.21 g/m3, and temperature being 29.17 ๐C. On the other hand, rest of the analyzed factors, do not demonstrate significant variations during the two years.Fig. 2 The daily of PM2.5, meteorological parameter, and COVID-19 cases in Bangkok Metropolitan Regions, from 1 January 2020 to 31 December 2021. Fig. 2 Fig. 3 Box plot distribution of yearly and season that shows daily COVID-19 cases and PM2.5, and meteorological parameters in BKK (2020–2021). Fig. 3 The findings of a two-year study conducted between 2020 and 2021 indicated that there was no statistically significant difference in PM2.5 concentrations. Over the course of two years, there were 213 days where the PM2.5 concentration exceeded the World Health Organization (WHO) threshold of a 24-h mean of 25 μg/m3 [37]. This accounted for approximately 23.14% of the days surpassing the standard. Notably, the study revealed that PM2.5 concentrations during the winter were 60.98% higher than the standard, followed by the summer at 33.52% and the rainy season at 0.98%. Moreover, the results indicated that both PM2.5 measures exceeded the WHO guidelines, which set an upper limit of 5 μg/m3 (annual mean) and 15 μg/m3 (24-h mean) [38]. It is interesting to observe from Fig. 3 that the cumulative number of COVID cases during the rainy season is higher compared to other seasons, while PM2.5 levels decrease during this period. In addition, a study in Pakistan discovered that socio-economic lockout has benefit controlled not just the levels of PM2.5 found but also other air pollutants and greenhouse gas (GHG) [39]. These reductions are mostly related to transportation-related operations, which were severely constrained during the lockdown phase [40]. It's probable that COVID-19's spread is unrelated to PM2.5 levels. 3.2 Relationship between PM2.5, meteorological factors and daily COVID-19 cases Fig. 4 shows the results of the relationship study using Spearman's Rank correlation coefficient between meteorological factors and PM2.5 with daily COVID-19 cases in BKK from 1 January 2020 to 31 December 2021. The result showed that the correlation between COVID-19 instances is substantial for PM2.5, RH, AH, pressure, and rainfall variables at level 5% and for most pairings at level 0.1%. PM2.5 was shown to be linked with RH (0.297), AH (0.255), pressure (−0.280), and rainfall (0.241) in the COVID-19 cases, for example. The correlation coefficient indicates that the COVID-19 cases were positively correlated with the RH, AH, and rainfall and negatively correlated with PM2.5 and pressure. Similarly, research by Adhikari and Yin (2020) in Queens, New York discovered a substantial inverse relationship between PM2.5 and new daily confirmed COVID-19 cases [41]. In the wake of its location, BKK is either in the tropical or equatorial rainforest. This study selected variables with statistically significant positive correlation with daily COVID-19 cases, namely RH, AH, R, to study the effect of a 1 unit increase in mean meteorological with positive daily COVID-19 cases as follows shown in Section 3.3.Fig. 4 Scatter plot of the Spearman's Rank correlation coefficient between meteorological factors and PM2.5 with daily COVID-19 cases in BKK from 1 January 2020 to 31 December 2021. Fig. 4 A study conducted in Lombardy, Italy, concluded that using PM10 as a carrier would not facilitate the airborne transmission of COVID-19 [42]. It is worth noting that the primary routes of human-to-human transmission for respiratory viruses, including SARS-CoV-2, are through direct contact and respiratory droplets [43]. Moreover, the indoor environment has been found to impede the spread of SARS-CoV-2 significantly [7]. However, in a setting when a person infected with COVID-19 is present, PM may accelerate the spread of infectious droplets and aerosols carrying the SARS-CoV-2 virus [44]. According to Marquès et al. (2022), patients with COVID-19, a severe condition, increase by 3.06% (95%CI: 1.11%–4.25%) when exposure to PM10 exceeds WHO recommendations by 1 μg/m3 [45]. Even though PM2.5 and PM10 are inversely correlated with the prevalence of COVID-19, particles smaller than 2.5 μm can enter type 2 alveolar cells [[45], [46], [47]], which contain the intracellular receptor (ACE2) for SARS-CoV-2, causing an increase in the prevalence of coronavirus infections. However, particles larger than 5 μm cannot enter type 2 alveolar cells [46]. Santurtn et al. (2022) also conducted a review and discovered the contribution of PM to the development of ACE-2 in respiratory cells in the human host and the clinical severity of the COVID-19-infected population [48]. It was shown that PM may reflect human activity in each location with a high population density and that this activity may also help SARS-CoV-2 spread [44]. Our study on meteorological factors found a statistically significant positive correlation between RH and AH, and the number of daily COVID-19 cases, which is in line with findings from a study of Singapore meteorological factors, Similar to a study in India that covered 244 days, from January to August 2020 [49], and a study conducted in Singapore revealed that the COVID-19 pandemic was linked to both RH and AH [50]. In addition, a study on Indian states discovered that relative and humidity-positive connections may significantly impact the fluctuation of COVID-19 case reproduction numbers [51]. On the other hand, our study discovered a negative correlation between daily COVID-19 instances in Bangkok and temperature and wind speed. The average temperature was discovered to be adversely correlated with the frequency of COVID-19 cases, similar to research from Bangladesh and Rio de Janeiro, Brazil [52], Jakarta, Indonesia [53], and Victoria, Mexico [54]. A different laboratory experiment discovered that coronavirus on flat surfaces could survive for more than five days at temperatures between 22 ๐C to 25 ๐C and that at higher temperatures, viral vitality quickly vanished [55]. Moreover, studies have shown that lower temperatures and higher relative humidity can enhance the aerosol transmission of respiratory viruses [56]. On the other hand, ultraviolet (UV) radiation, which possesses wavelengths capable of damaging RNA viruses, is often present in the general outdoor atmosphere [56,57]. It has been observed that SARS-CoV-2 transmission is reduced in hot climates with moderate outdoor UV exposure [58]. It is important to note that sunlight reaching the Earth's surface tends to increase temperature or heat, and this may be a scenario where warmer temperatures do not significantly impact the daily incidence of coronavirus cases. It has also been reported that COVID-19 can be transmitted through direct contact, human contact, coughing, sneezing, or aerosols containing droplet nuclei generated by an infected person [23]. Several factors can influence the increase or decrease in the number of COVID-19 cases. These include implementing of implementing lockdown measures, population movement within each country, exposure to symptomatic and asymptomatic individuals, and the progress of COVID-19 vaccination efforts. It is important to recognize that the spread of COVID-19 is not solely determined by air pollution or meteorological factors. Taking into account these various factors allows for more comprehensive discussions and enables us to draw clearer conclusions regarding the dynamics of COVID-19 transmission. 3.3 Effects of RH, RH, and rainfall on daily COVID-19 confirmed cases in BKK In Fig. 5 shown the trend line for all meteorological variables by season and verified COVID-19 cases is shown. After one year of transmission, COVID-19 cases rose, with the wet season accounting for the majority of the rise. When PM2.5 and pressure are at their lowest in the rainy season, wind speed and absolute humidity are also at their maximum levels. In addition to having the lowest temperature and maximum pressure, winter also has the highest pressure [59]. GAM was utilized to analyze the impacts of relative humidity and temperature in a prior study by Wu et al. (2020) [36]. As per the findings, a 1 ๐C increase in temperature was associated with a 3.08% (95% CI: 1.53%–4.63%) decline in daily new COVID-19 cases, and a 1% increase in relative humidity was attributed to a 0.85% (95% CI: 0.51%–1.19%) decline [32].Fig. 5 Season-wise daily pattern of COVID-19 cases and climate factors, from 1 January 2020 to 31 December 2021 in BKK. Fig. 5 To evaluate the impact of RH, AH, and rainfall, respectively, above and below the threshold, a piecewise linear regression was modified based on the results from GAMs and thresholds of 66%, 19 g/m3, and 3 mm were used. As shown in Table 1 describes the statistics of daily COVID-19 cases are affected by a 1-unit rise in mean meteorology, our study discovered that RH 66%, each 1% increase in mean RH (Lag0-7) resulted in a 1526.09 (95% CI: 970.38–2081.79) increase in the daily number of COVID-19 confirmed cases when mean RH was below 66%, with a positive effect that was largest at Lag0-7 being statistically significant. When the mean RH was above 66%, the positive effect of RH was not statistically significant. When mean AH was over 19 g/m3, each 1 g/m3 increase in mean AH (Lag0-7) resulted in a 1539.08 (95% CI: 809.09–2269.06) increase in the daily number of COVID-19 cases having a positive impact at Lag0-7. The positive effect of AH was not statistically significant when the mean AH was less than 19 g/m3. Additionally, when mean rainfall was above 3 mm, each 1 mm increase in decreased rainfall (Lag0-7) resulted in a 2925.09 (95% CI: 1689.77–4160.42) increase in the daily number of COVID-19 confirmed cases, but this increase was not statistically significant. It was statistically significant (p < 0.001) that the favorable effect is greatest at Lag0-21 (COVID-19 cases = 4162.53, 95% CI: 2218.28–6106.77). The detrimental impact of rainfall was not statistically significant when the mean rainfall was less than 3 mm.Table 1 The effect of a 1 unit increase in mean meteorological with positive daily COVID-19 cases in BKK during 2020–2021. Table 1 Mean RH ≤ 66% p-value Mean RH > 66% p-value Number of Case change 95%CI Number of Case change 95%CI Lag0-7 1526.09* 970.38–2081.79 <0.001 371.06 −224.77–966.88 0.213 Lag0-14 1493.55* 936.32–2050.78 <0.001 274.13 −319.16–867.43 0.356 Lag0-21 1345.27* 807.74–1882.80 <0.001 300.48 −317.49–918.45 0.331 Mean AH ≤ 19 g/m3 p-value Mean AH > 19 g/m3 p-value Number of Case change 95%CI Number of Case change 95%CI Lag0-7 571.42 −119.85–1262.70 0.098 1539.08* 809.09–2269.06 <0.001 Lag0-14 691.14 −37.14–1419.42 0.058 1252.49* 513.25–1991.73 0.001 Lag0-21 658.64 −22.49–1339.78 0.053 1092.91* 340.14–1845.67 0.004 Mean rainfall ≤3 mm p-value Mean rainfall >3 mm p-value Number of Case change 95%CI Number of Case change 95%CI Lag0-7 −92.45 −659.51–474.61 0.744 2925.09* 1689.77–4160.42 <0.001 Lag0-14 −14.68 −571.05–541.70 0.958 2501.63* 1312.60–3690.65 <0.001 Lag0-21 −642.59 −1396.88–111.69 0.088 4162.53* 2218.28–6106.77 <0.001 p < 0.05. The study indicated that there was a significant nonlinear connection between RH and daily COVID-19-confirmed cases (Lag0-7, Lag0-14, and Lag0-21, at p < 0.001). The relationship was specifically linear in the range of 66% and flat above 66%, showing that 66% was the single threshold of the RH effect on SARS-CoV-2. According to additional findings, areas with less humidity (<40% RH) have a higher risk of SARS-CoV-2 airborne transmission than areas with higher humidity (>90% RH) [60]. The transmission of the influenza virus was just as effective at lower RH (<35%) as it was at higher RH (60–80%) [61]. It has been hypothesized that the virus can persist in humid environments in water droplets beneath the physiological content of salt at high humidity, i.e., humidity >70%. When the humidity is between 40 and 60%, the evaporation that inactivates the viruses concentrates the salt, and when the humidity falls below 30%, the salt separates from the solution. This might permit the virus to continue to function [61,62]. Furthermore, indoor humidity dispersal is greatly reduced when relative humidity levels are lower than 40% [63]. The current investigation results indicated, the relative humidity ranged between 39% and 90% and which related to the daily incidence of COVID-19 cases. It suggested that RH might have an impact on the number of COVID patients in Bangkok. While AH will be proportional to the value of RH found that the association was roughly linear in the range of >19 g/m3 and became flat below 19 g/m3, demonstrating that the single threshold of the AH effect on COVID-19 was 19 g/m3. The relationship between AH and daily of COVID-19 cases was significant nonlinear (Lag0-7: p < 0.001; Lag0-14: p < 0.001; Lag0-21: p < 0.004). The relationship was specifically linear in the range of 66% and flat above 66%, showing that 66% was the single threshold of the RH effect on SARS-CoV-2. The association was roughly linear in the range of >19 g/m3 and became flat below 19 g/m3, demonstrating that the single threshold of the AH effect on COVID-19 was 19 g/m3. The relationship between AH and daily of COVID-19 cases was significant nonlinear (Lag0-7: p < 0.001; Lag0-14: p < 0.001; Lag0-21: p < 0.004). In addition, it was fascinating to note that this study demonstrated a statically significant nonlinear association between daily COVID-19 confirmed cases and rainfall (lag0-7, lag0-14, and lag0-21 with p < 0.001). To be more precise, the connection was somewhat linear in the range of >3 mm and went flat below 3 mm, showing that the single threshold of the rainfall influence on COVID-19 was 3 mm. Additionally, Fig. 5, demonstrates that the spine curve tended to rise in line with the rise in COVID-19 cases during the rainy season, which was influenced by meteorological conditions such as RH, AH, and rainfall. However, it was observed that the number of COVID-19 cases showed a statistically significant decline during the summer when considering meteorological parameters and adjusting for values. In both the unadjusted and adjusted models, a significant association was found between COVID-19 cases and the summer season, with the number of cases increasing by 640.27 (95% CI: 464.57 to 815.97) and declining by 635.39 (95% CI: 452.94 to 817.83) respectively (Table 2 ). The study also found that COVID-19 cases increased by 5.61 times in the adjusted model and 6.31 times in the unadjusted model in relation to rainfall. This aligns with the findings of Xie and Zhu (2020) who noted a decrease in COVID-19 cases with a warming climate [35].Table 2 Factors associated with COVID-19 cases and trend results from unadjusted and adjusted using the GAMs. Table 2Variables Unadjusted (95% CI) p-value Adjusted (95% CI) p-value Season Summer 650.03 (643.68–656.38) <0.001 656.41* (649.95–662.88) <0.001 Winter 117.87 (111.41–124.33) 0.186 125.92 (119.28–132.55) 0.170 PM2.5 (μg/m3) −2.03 (−2.17 to −1.89) 0.297 −3.36 (−3.52 to −3.21) 0.118 Wind speed (m/s) −163.30* (−167.98 to −158.62) 0.012 −143.25 (−149.20 to −137.29) 0.082 Relative humidity (%) 4.09* (3.95–4.23) <0.001 1.73 (0.70–2.76) 0.903 Absolute humidity (g/m3) 12.46 (11.92–13.01) 0.099 −5.80 (−9.55 to −2.05) 0.911 Pressure (mmHg) 6.31* (6.15–6.48) 0.006 −2.56 (−3.54 to −1.57) 0.851 Rainfall (mm) 10.33* (10.33–12.07) 0.352 5.74* (5.57–5.92) 0.019 Temperature (๐C) −9.20 (−10.09 to −8.31) 0.455 9.17 (5.55–12.79) 0.854 Model diagnosis for adjusted model Adjusted R-square 0.852 Deviance explained 85.60% AIC 10914.24 * Significant at p < 0.05; Akaike Information Criteria: AIC. Furthermore, Yin et al. (2022) documented the seasonality of COVID-19 transmission in Brazil [64]. While there was no association between COVID-19 cases and rainfall in the unadjusted model, a significant relationship was observed in the adjusted model. No other meteorological factors appeared to be significant for COVID-19 cases, except for a negative association between particulate matter (PM) and wind speed. The effects of season, relative humidity, pressure, and rainfall were found to be favorable for the occurrence of COVID-19 cases. Specifically, the summer and monsoon seasons showed a statistically significant positive association with the trend in the number of COVID-19 cases after adjusting for variables. The GAM model technique was used in this study to assess how the weather affected COVID-19 transmission in BKK. Our study shows that COVID-19 infections are significantly affected positively or negatively by variations in the weather. However, the findings don't seem to apply to both the adjusted and unadjusted models. In contrast to humidity and pressure, which are positively correlated with COVID-19 transmission in raw models and negatively correlated in adjusted models, respectively, temperature and pressure have a relationship with humidity and pressure. Wind speed was found to have a negative impact on COVID-19 transmission in both correlation and GAM models, which is consistent with other earlier studies [65]. Recent studies claimed that there was no link between wind speed and COVID-19 transmission, even though our correlation and the adjusted GAM model corroborate this finding [66,67]. Our study revealed that temperature had both direct and indirect effects on the transmissibility of COVID-19, although these effects were not statistically significant in the correlation analysis or the adjusted and unadjusted models [67,68]. Previous studies have shown a positive association between COVID-19 cases and temperature, as supported by numerous research reports [[69], [70], [71], [72]]. However, several studies have reported no significant association between temperature and the incidence of COVID-19 [73], or even an inverse relationship similar to what we observed in our unadjusted model, which aligns with the findings of our GAM model [36]. Due to oxidative stress and airway inflammation that is brought on by prolonged exposure to high PM levels and poor air quality, which encourages viral transmission and adverse respiratory effects, the host becomes more vulnerable to respiratory infections [13]. We discovered a negative link between PM2.5 and COVID-19 case counts, similar to research that established a negative relationship between PM2.5 and PM10 and COVID-19 death counts in Wuhan, China [74]. Increased air pollution is strongly correlated with a higher incidence of verified COVID-19 cases, according to numerous studies, which have either indicated the opposite or a positive correlation [65,66]. This study revealed that meteorological parameters like season and rainfall influenced the trend of COVID-19 cases in the Bangkok Metropolitan Region. This is further supported by Omicron spreading and infecting more quickly than earlier strains that affected the 2021 pandemic [75,76]. Findings from Mei et al. (2020) [77], Oran and Topol (2020) [78], and WHO (2021) [9] suggest that many of these illnesses have both symptomatic and asymptomatic signs. Additionally, SARS-CoV-2 mutations and person-to-person interaction in densely populated or indoor places may be to blame for the sharp increase in the daily number of cases. 3.4 Limitations of the study The interpretation of daily COVID-19 confirmed cases can be influenced by various factors, leading to potential underreporting. Factors such as government handling of cases, effectiveness of restrictions, the occurrence of festivals, implementation of lockdowns, population immunity, social behavior, tourist seasons, and population migrations can all contribute to the misinterpretation of daily case numbers. Additionally, rainfall can reduce particulate matter (PM) levels in the air by washing away dust particles. However, it is important to exercise caution when interpreting such data, as other significant factors may increase the number of cases during the rainy season, independent of weather conditions. The reduction of PM levels in the air during the rainy season may not necessarily translate into a decrease in COVID-19 cases. In fact, there might have been an increase in COVID-19 cases due to the government's measurement practices during rain in Thailand. Consequently, a negative association between PM2.5 and COVID-19 cases was observed. PM in the air can weaken the immune system of individuals, potentially prolonging the presence of SARS-CoV-2 particles in the air and facilitating their transmission from infected individuals to healthy individuals. Despite initial assumptions of a positive correlation, future research should consider these factors when assessing the long-term effects of integrated meteorological variables, lockdown measures, population mobility, immunization efforts, and COVID-19 cases in specific regions such as Bangkok. 4 Conclusion Our research revealed a linear relationship between mean RH, AH, and rainfall, and the daily number of confirmed COVID-19 cases when the RH is below 66%, and AH and rainfall are over 19 g/m3 and 3 mm respectively. Summer and monsoon were shown to impacted trend results via adjusted GAMs and COVID-19 instances. The model diagnostic for adjusted R-square was 0.852 with deviance explained at 85.60% and was statistically significant at p < 0.05. A few complicating factors, including patients with asymptomatic symptoms of COVID-19 infection, immunization, socioeconomic factors, SARS-CoV-2 variations, and topographic considerations, should also be addressed herein. Further research is needed to investigate the potential role of meteorological conditions, vaccination efforts, air pollutants, and medical histories of individuals infected with SARS-CoV-2 in preventing COVID-19 during and after the pandemic. These findings will provide valuable insights for policymakers to make informed decisions to prevent future outbreaks and mitigate the risk of respiratory infections. Moreover, the information obtained from such research can be utilized to develop strategies for planning future seasonal COVID-19 vaccinations. Funding No funding. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Supplementary data The following is the Supplementary data to this article:Multimedia component 1 Multimedia component 1 Data availability Data will be made available on request. Acknowledgments The authors would like to thank the Air Quality and Noise Management Bureau, Pollution Control Department (PCD), and Ministry of Natural Resources and Environment for the secondary data of air pollutants and meteorological parameter data. The images and graphical abstract were created with BioRender.com. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.cscee.2023.100410. ==== Refs References 1 Ahmed F. Islam M.A. Kumar M. Hossain M. Bhattacharya P. Islam M.T. Hossen F. Hossain S. Islam S. Uddin M. Islam N. Bahadur N.M. Didar-Ul-Alam M. Reza H.M. Jakariya M. First detection of SARS-CoV-2 genetic material in the vicinity of COVID-19 isolation Centre in Bangladesh: variation along the sewer network Sci. Total Environ. 776 2021 145724 2 Islam M.A. Haque M.A. Rahman M.A. Hossen F. Reza M. Barua A. Marzan A.A. Das T. Baral S.K. He C. Ahmed F. Bhattacharya P. Jakariya M. A review on measures to rejuvenate immune system: Natural mode of protection against coronavirus infection Front. Immunol. 13 2022 837290 3 Ju M.J. Oh J. Choi Y.-H. Changes in air pollution levels after COVID-19 outbreak in Korea Sci. Total Environ. 750 2021 141521 4 Sakib M.M.H. Nishat A.A. Islam M.T. Raihan Uddin M.A. Iqbal M.S. Bin Hossen F.F. Ahmed M.I. Bashir S. Hossain T. Tohura U.S. Saif S.I. Jui N.R. Alam M. Islam A. Hasan M. Sufian A. Ali A. Islam R. Hossain M.A. Halim M.A. Computational screening of 645 antiviral peptides against the receptor-binding domain of the spike protein in SARS-CoV-2, Comput Biol. Med. 136 2021 104759 5 Zhu Y. Xie J. Huang F. Cao L. Association between short-term exposure to air pollution and COVID-19 infection: evidence from China Sci. Total Environ. 727 2020 138704 6 Lauer S.A. Grantz K.H. Bi Q. Jones F.K. Zheng Q. Meredith H.R. Azman A.S. Reich N.G. Lessler J. The incubation period of coronavirus disease 2019 (COVID-19) from publicly reported confirmed cases: estimation and application Ann. Intern. Med. 172 9 2020 577 582 32150748 7 Piscitelli P. Miani A. Setti L. De Gennaro G. Rodo X. Artinano B. Vara E. Rancan L. Arias J. Passarini F. Barbieri P. Pallavicini A. Parente A. D'Oro E.C. Maio C.D. Saladino F. Borelli M. Colicino E. Gonçalves L.M.G. Tanna G.D. Colao A. Leonardi G.S. Baccarelli A. Dominici F. Ioannidis J.P.A. Domingo J.L. On behalf ofthe RESCOP Commission established by Environmental Research (Elsevier), the role of outdoor and indoor air quality in the spread of SARS-CoV-2: overview and recommendations by the research group on COVID-19 and particulate matter (RESCOP commission) Environ. Res. 211 2022 113038 8 Islam A. Hossen F. Rahman A. Sultana K.F. Hasan M.N. Haque A. Sosa- Hernández J.E. Oyervides-Muñoz M.A. Parra-Saldívar R. Ahmed T. Islam T. Dhama K. Sangkham S. Bahadur N.M. Reza H.M. Jakariya M. Marzan A.A. Bhattacharya P. Sonne C. Ahmed F. An opinion on wastewater-based epidemiological monitoring (WBEM) with clinical diagnostic test (CDT) for detecting high-prevalence areas of community COVID-19 infections Curr. Opin. Environ. Sci. Health 31 2022 100396 9 WHO. Classification of Omicron (B.1.1.529): SARS-CoV-2 Variant of Concern 2021 https://www.who.int/news/item/26-11-2021-classification-of-omicron-(b.1.1.529)- sars-cov-2-variant-of-concern 10 WHO COVID-19 Weekly Epidemiological Update, 12 January 2021 2021 https://apps.who.int/iris/handle/10665/338703 11 WHO Coronavirus Disease (COVID-19) Weekly Epidemiological Update and Weekly Operational Ppdate 2021 https://apps.who.int/iris/handle/10665/338703 12 DDC https://data.go.th/dataset/covid-19-daily. 2021 13 Domingo J.L. Rovira J. Effects of air pollutants on the transmission and severity of respiratory viral infections Environ. Res. 187 2020 109650 14 Hyman S. Zhang J. Andersen Z.J. Cruickshank S. Møller P. Daras K. Williams R. Topping D. Lim Y.-H. Long-term exposure to air pollution and COVID-19 severity: a cohort study in Greater Manchester, United Kingdom Environ. Pollut. 327 2023 121594 15 Hasan M.N. Islam M.A. Sangkham S. Werkneh A.A. Hossen F. Haque M.A. Alam M.M. Rahman M.A. Mukharjee S.K. Chowdhury T.A. Sosa-Hernández J.E. Jakariya M. Ahmed F. Bhattacharya P. Sarkodie S.A. Insight into vaccination and meteorological factors on daily COVID-19 cases and mortality in Bangladesh Groundw. Sustain. Dev. 21 2023 100932 16 Wu X. Nethery R.C. Sabath M.B. Braun D. Dominici F. Air pollution and COVID- 19 mortality in the United States: strengths and limitations of an ecological regression analysis Sci. Adv. 6 45 2020 eabd4049 17 Coker E.S. Cavalli L. Fabrizi E. Guastella G. Lippo E. Parisi M.L. Pontarollo N. Rizzati M. Varacca A. Vergalli S. The effects of air pollution on COVID-19 related mortality in Northern Italy Environ. Resour. Econ. 76 4 2020 611 634 18 Mehmood K. Saifullah Iqbal M. Abrar M.M. Can exposure to PM2.5 particles increase the incidence of coronavirus disease 2019 (COVID-19)? Sci. Total Environ. 741 2020 140441 19 Hoang T. Tran T.T.A. Ambient air pollution, meteorology, and COVID-19 infection in Korea J. Med. Virol. 93 2 2021 878 885 32691877 20 Cai Q Q.-C. Lu J. Xu Q.-F. Guo Q. Xu D.-Z. Sun Q.-W. Yang H. Zhao G.-M. Jiang Q.-W. Influence of meteorological factors and air pollution on the outbreak of severe acute respiratory syndrome Publ. Health 121 4 2007 258 265 21 Srivastava A. COVID-19 and air pollution and meteorology-an intricate relationship: a review Chemosphere 263 2021 128297 22 Li H. Xu X.-L. Dai D.-W. Huang Z.-Y. Ma Z. Guan Y.-J. Air pollution and temperature are associated with increased COVID-19 incidence: a time series study Int. J. Infect. Dis. 97 2020 278 282 32502664 23 Iqbal M.M. Abid I. Hussain S. Shahzad N. Waqas M.S. Iqbal M.J. The effects of regional climatic condition on the spread of COVID-19 at global scale Sci. Total Environ. 739 2020 140101 24 Marquès M. Domingo J.L. Positive association between outdoor air pollution and the incidence and severity of COVID-19. A review of the recent scientific evidences Environ. Res. 203 2022 111930 25 Sangkham S. Thongtip S. Vongruang P. Influence of air pollution and meteorological factors on the spread of COVID-19 in the Bangkok Metropolitan Region and air quality during the outbreak Environ. Res. 197 2021 111104 26 Lolli S. Chen Y.-C. Wang S.-H. Vivone G. Impact of meteorological conditions and air pollution on COVID-19 pandemic transmission in Italy Sci. Rep. 10 1 2020 16213 27 DDC https://data.go.th/dataset/covid-19-daily. 2021 2022 28 Liu J. Zhou J. Yao J. Zhang X. Li L. Xu X. He X. Wang B. Fu S. Niu T. Yan J. Shi Y. Ren X. Niu J. Zhu W. Li S. Luo B. Zhang K. Impact of meteorological factors on the COVID-19 transmission: a multi-city study in China Sci. Total Environ. 726 2020 138513 29 Prata D.N. Rodrigues W. Bermejo P.H. Temperature significantly changes COVID- 19 transmission in (sub)tropical cities of Brazil Sci. Total Environ. 729 2020 138862 30 Wu X. Lang L. Ma W. Song T. Kang M. He J. Zhang Y. Lu L. Lin H. Ling L. Non-linear effects of mean temperature and relative humidity on dengue incidence in Guangzhou, China Sci. Total Environ. 628–629 2018 766 771 31 Islam M.A. Hasan M.N. Tiwari A. Raju M.A.W. Jannat F. Sangkham S. Shammas M.I. Sharma P. Bhattacharya P. Kumar M. Correlation of Dengue and meteorological factors in Bangladesh: a public health concern Int. J. Environ. Res. Publ. Health 20 6 2023 5152 32 Islam M.A. Sangkham S. Tiwari A. Vadiati M. Hasan M.N. Noor S.T.A. Mumin J. Bhattacharya P. Sherchan S.P. Association between global Monkeypox cases and meteorological factors Int. J. Environ. Res. Publ. Health 19 23 2022 15638 33 T.J. Hastie, Generalized additive models. Statistical Models in S2017. p. 249-307. 34 Liu K. Hou X. Ren Z. Lowe R. Wang Y. Li R. Liu X. Sun J. Lu L. Song X. Wu H. Wang J. Yao W. Zhang C. Sang S. Gao Y. Li J. Li J. Xu L. Liu Q. Climate factors and the East Asian summer monsoon may drive large outbreaks of dengue in China Environ. Res. 183 2020 109190 35 Xie J. Zhu Y. Association between ambient temperature and COVID-19 infection in 122 cities from China Sci. Total Environ. 724 2020 138201 36 Wu Y. Jing W. Liu J. Ma Q. Yuan J. Wang Y. Du M. Liu M. Effects of temperature and humidity on the daily new cases and new deaths of COVID-19 in 166 countries Sci. Total Environ. 729 2020 139051 37 WHO Ambient (Outdoor) Air Pollution 2018 https://www.who.int/news-room/fact-sheets/detail/ambient-(outdoor)-air-quality-and-health 38 WHO Ambient (Outdoor) Air Pollution 2021 https://www.who.int/news-room/fact-sheets/detail/ambient-(outdoor)-air-quality-and-health 39 Ali S.M. Malik F. Anjum M.S. Siddiqui G.F. Anwar M.N. Lam S.S. Nizami A.S. Khokhar M.F. Exploring the linkage between PM2.5 levels and COVID-19 spread and its implications for socio-economic circles Environ. Res. 193 2021 110421 40 Arregocés H.A. Rojano R. Restrepo G. Impact of lockdown on particulate matter concentrations in Colombia during the COVID-19 pandemic Sci. Total Environ. 764 2021 142874 41 Adhikari A. Yin J. Short-term effects of ambient ozone, PM2.5, and meteorological factors on COVID-19 confirmed cases and deaths in Queens, New York Int. J. Environ. Res. Publ. Health 17 11 2020 4047 42 Bontempi E. First data analysis about possible COVID-19 virus airborne diffusion due to air particulate matter (PM): the case of Lombardy (Italy) Environ. Res. 186 2020 109639 43 Domingo J.L. Marquès M. Rovira J. Influence of airborne transmission of SARS- CoV-2 on COVID-19 pandemic. A review Environ. Res. 188 2020 109861 44 Ishmatov A. “SARS-CoV-2 is transmitted by particulate air pollution”: misinterpretations of statistical data, skewed citation practices, and misuse of specific terminology spreading the misconception Environ. Res. 204 2022 112116 45 Marquès M. Correig E. Ibarretxe D. Anoro E. Antonio Arroyo J. Jericó C. Borrallo R.M. Miret M.L. Näf S. Pardo A. Perea V. Pérez-Bernalte R. Ramírez- Montesinos R. Royuela M. Soler C. Urquizu-Padilla M. Zamora A. Pedro-Botet J. STACOV-XULA research groupMasana L. Domingo J.L. Long-term exposure to PM10 above WHO guidelines exacerbates COVID-19 severity and mortality Environ. Int. 158 2022 106930 46 Copat C. Cristaldi A. Fiore M. Grasso A. Zuccarello P. Signorelli S.S. Conti G.O. Ferrante M. The role of air pollution (PM and NO2) in COVID-19 spread and lethality: a systematic review Environ. Res. 191 2020 110129 47 Comunian S. Dongo D. Milani C. Palestini P. Air pollution and COVID-19: the role of particulate matter in the spread and increase of COVID-19's morbidity and mortality Int. J. Environ. Res. Publ. Health 17 12 2020 4487 48 Santurtún A. Colom M.L. Fdez-Arroyabe P. Real Ád Fernández-Olmo I. Zarrabeitia M.T. Exposure to particulate matter: direct and indirect role in the COVID-19 pandemic Environ. Res. 206 2022 112261 49 Gautam S. Samuel C. Gautam A.S. Kumar S. Strong link between coronavirus count and bad air: a case study of India Environ. Dev. Sustain. 23 11 2021 16632 16645 33841040 50 Pani S.K. Lin N.-H. RavindraBabu S. Association of COVID-19 pandemic with meteorological parameters over Singapore Sci. Total Environ. 740 2020 140112 51 Manik S. Mandal M. Pal S. Patra S. Acharya S. Impact of climate on COVID-19 transmission: a study over Indian states Environ. Res. 211 2022 113110 52 Rosario D.K.A. Mutz Y.S. Bernardes P.C. Conte-Junior C.A. Relationship between COVID-19 and weather: case study in a tropical country Int. J. Hyg Environ. Health 229 2020 113587 53 Rendana M. Impact of the wind conditions on COVID-19 pandemic: a new insight for direction of the spread of the virus Urban Clim. 34 2020 100680 54 Tello-Leal E. Macías-Hernández B.A. Association of environmental and meteorological factors on the spread of COVID-19 in Victoria, Mexico, and air quality during the lockdown Environ. Res. 196 2021 110442 55 Chan K.H. Peiris J.S.M. Lam S.Y. Poon L.L.M. Yuen K.Y. Seto W.H. The effects of temperature and relative humidity on the viability of the SARS coronavirus Adv. Virol. 2011 2011 734690 56 Audi A. AlIbrahim M. Kaddoura M. Hijazi G. Yassine H.M. Zaraket H. Seasonality of respiratory viral infections: will COVID-19 follow suit? Front. Public Health 8 2020 567184 57 Lytle C.D. Sagripanti J.-L. Predicted inactivation of viruses of relevance to biodefense by solar radiation J. Virol. 79 22 2005 14244 14252 16254359 58 Xu R. Rahmandad H. Gupta M. DiGennaro C. Ghaffarzadegan N. Amini H. Jalali M.S. Weather, air pollution, and SARS-CoV-2 transmission: a global analysis Lancet Planet. Health 5 10 2021 e671 e680 34627471 59 Vadiati M. Ghasemi L. Samani S. Islam M.A. Ahmadi A. Khaleghi S. A sustainable trend in COVID-19 research: an environmental perspective Front. Environ. Sci. 2023 11 60 Ahlawat A. Wiedensohler A. Mishra S.K. An overview on the role of relative humidity in airborne transmission of SARS-CoV-2 in indoor environments Aerosol Air Qual. Res. 20 9 2020 1856 1861 61 Yuan S. Jiang S.-C. Li Z.-L. Do humidity and temperature impact the spread of the novel coronavirus? Front. Public Health 8 2020 240 32574306 62 Yang W. Elankumaran S. Marr L.C. Relationship between humidity and influenza A viability in droplets and implications for influenza's seasonality PLoS One 7 10 2012 e46789 63 Skanata A. Spagnolo F. Metz M. Smyth D.S. Dennehy J.J. Humidity reduces rapid and distant airborne dispersal of viable viral particles in classroom settings Environ. Sci. Technol. Lett. 9 7 2022 632 637 35937034 64 Yin C. Zhao W. Pereira P. Meteorological factors' effects on COVID-19 show seasonality and spatiality in Brazil Environ. Res. 208 2022 112690 65 Zoran M.A. Savastru R.S. Savastru D.M. Tautan M.N. Assessing the relationship between surface levels of PM2.5 and PM10 particulate matter impact on COVID-19 in Milan, Italy Sci. Total Environ. 738 2020 139825 66 Bashir M.F. Ma B. Bilal Komal B. Bashir M.A. Tan D. Bashir M. Correlation between climate indicators and COVID-19 pandemic in New York, USA Sci. Total Environ. 728 2020 138835 67 Menebo M.M. Temperature and precipitation associate with Covid-19 new daily cases: a correlation study between weather and Covid-19 pandemic in Oslo, Norway Sci. Total Environ. 737 2020 139659 68 Gupta S. Raghuwanshi G.S. Chanda A. Effect of weather on COVID-19 spread in the US: a prediction model for India in 2020 Sci. Total Environ. 728 2020 138860 69 Islam N. Shabnam S. Erzurumluoglu A.M. Temperature, Humidity, and Wind Speed Are Associated with Lower Covid-19 Incidence, medRxiv 2020 2020 03.27.20045658 70 Nottmeyer L.N. Sera F. Influence of temperature, and of relative and absolute humidity on COVID-19 incidence in England - a multi-city time-series study Environ. Res. 196 2021 110977 71 Shi P. Dong Y. Yan H. Li X. Zhao C. Liu W. He M. Tang S. Xi S. The impact of temperature and absolute humidity on the coronavirus disease 2019 (COVID-19) outbreak - evidence from China medRxiv 2020 2020 03.22.20038919 72 Tong L. Ji L. Li D. Xu H. The occurrence of COVID-19 is associated with air quality and relative humidity J. Med. Virol. 94 3 2022 965 970 34647628 73 Poirier C. Luo W. Majumder M.S. Liu D. Mandl K.D. Mooring T.A. Santillana M. The role of environmental factors on transmission rates of the COVID-19 outbreak: an initial assessment in two spatial scales Sci. Rep. 10 1 2020 17002 74 Ma Y. Zhao Y. Liu J. He X. Wang B. Fu S. Yan J. Niu J. Zhou J. Luo B. Effects of temperature variation and humidity on the death of COVID-19 in Wuhan, China Sci. Total Environ. 724 2020 138226 75 Callaway E. Why does the Omicron sub-variant spread faster than the original? Nature 602 7898 2022 556 557 35177843 76 Kannan S. Shaik Syed Ali P. Sheeza A. Omicron (B.1.1.529) - variant of concern - molecular profile and epidemiology: a mini review Eur. Rev. Med. Pharmacol. Sci. 25 24 2021 8019 8022 34982466 77 Mei X. Zhang Y. Zhu H. Ling Y. Zou Y. Zhang Z. Guo H. Liu Y. Cheng X. Liu M. Huang W. Wang J. Yi Z. Qian Z. Lu H. Observations about symptomatic and asymptomatic infections of 494 patients with COVID-19 in Shanghai, China Am. J. Infect. Control 48 9 2020 1045 1050 32645475 78 Oran D.P. Topol E.J. Prevalence of asymptomatic SARS-CoV-2 infection Ann. Intern. Med. 173 5 2020 362 367 32491919
PMC010xxxxxx/PMC10286574.txt	==== Front Spectrochim Acta A Mol Biomol Spectrosc Spectrochim Acta A Mol Biomol Spectrosc Spectrochimica Acta. Part A, Molecular and Biomolecular Spectroscopy 1386-1425 1873-3557 Elsevier B.V. S1386-1425(23)00751-5 10.1016/j.saa.2023.123066 123066 Article SARS CoV-2 infection screening via the exhaled breath fingerprint obtained by FTIR spectroscopic gas-phase analysis. A proof of concept Glöckler Johannes a1 Mizaikoff Boris ab Díaz de León-Martínez Lorena a1⁎ a Institute of Analytical and Bioanalytical Chemistry, Ulm University, Albert-Einstein-Allee 11, 89081 Ulm, Germany b Hahn-Schickard Institute for Microanalysis Systems, Sedanstrasse 14, 89077 Ulm, Germany ⁎ Corresponding author. 1 Both authors equally contributed to this work. 22 6 2023 22 6 2023 1230662 2 2023 30 5 2023 20 6 2023 © 2023 Elsevier B.V. All rights reserved. 2023 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Graphical abstract The COVID-19 pandemic remains a global challenge now with the long-COVID arising. Mitigation measures focused on case counting, assessment and determination of variants and their likely targets of infection and transmission, the pursuit of drug treatments, use and enhancement of masks, social distancing, vaccination, post-infection rehabilitation, and mass screening. The latter is of utmost importance given the current scenario of infections, reinfections, and long-term health effects. Research on screening platforms has been developed to provide more sensitive, specific, and reliable tests that are accessible to the entire population and can be used to assess the prognosis of the disease as well as the subsequent health follow-up of patients with sequelae of COVID-19. Therefore, the aim of the present study was the simulation of exhaled breath of COVID-19 patients by evaluation of three identified COVID-19 indicator breath biomarkers (acetone (ACE), acetaldehyde (ACH) and nitric oxide (NO)) by gas-phase infrared spectroscopy as a proof-of-concept principle for the detection of infected patients’ exhaled breath fingerprint and subsequent follow-up. The specific fingerprints of each of the compounds and the overall fingerprint were obtained. The synthetic exhaled breath evaluation concept revealed a linearity of r=0.99 for all compounds, and LODs of 6.42, 13.81, 9.22 ppm, and LOQs of 42.26, 52.57, 69.23 ppm for NO, ACE, and ACH, respectively. This study proves the fundamental feasibility of gas-phase infrared spectroscopy for fingerprinting lung damage biomarkers in exhaled breath of patients with COVID-19. This analysis would allow faster and cheaper screening and follow-up of infected individuals, which could improve mass screening in POC settings. Keywords Gas-phase infrared spectroscopy substrate-integrated hollow waveguide iHWG Exhaled breath VOC COVID-19 screening ==== Body pmc1 INTRODUCTION The ongoing pandemic of COVID-19 has affected all countries. Efforts by several research groups focused on addressing the mitigation strategies established by health agencies within each country. Strategies include counting cases, evaluating and determining the variants, and their likely targets of infection and transmission, searching for pharmacological treatments, the use and improvement of masks, social distancing, vaccination, post-infection rehabilitation, and mass screening [1], [2]. The latter being of utmost importance given the present scenario of infections, re-infections, and long-term health effects now so-called the ‘long COVID’. Among the technologies described for screening COVID-19 cases and post-infection monitoring are RTq-PCR tests as the gold standard [3] and rapid antigen and antibody tests [4], some have been evaluated and accepted by the Centers for Disease Control and Prevention (CDC), and the Food and Drug Administration (FDA) as acceptance tests for case detection, depending on the temporality of infection [5]. However, these tests are only indicative of latent SARS-CoV-2 virus infection or in the case of antibody tests of past recent infection. In this context, research on screening platforms has been developed to offer more sensitive, specific, reliable tests that are accessible to the entire population and can be used to evaluate the disease prognosis as well as the subsequent monitoring of patients with long-COVID. Despite SARS-CoV-2 being considered a respiratory disease, it has been shown that the virus can use the olfactory nerve to infiltrate and damage the nervous system by activating T lymphocytes which, in turn, activate microglia and inflammatory mediators. Damage to the nervous system has been associated with disease-specific metabolites and unusual chemical biomarkers that are emitted into the exhaled breath from soft tissue and body cavities, such as the lungs or nasal cavity, which can be detected in a person's breath [6]. These biomarkers are volatile organic compounds (VOCs) and represent important metabolic endpoints that can be evaluated for the clinical and non-clinical status of an individual, and thus provide relevant health status information. VOCs have been studied in different infectious and non-infectious lung diseases including COVID-19 [7], [8], [9], [10]. Gould et al. state that the virus-host interaction alters cellular metabolism at different levels, triggering oxidative stress processes, that result in changes in the composition and concentration of certain VOCs in exhaled breath and other biological matrices [11]. On April 2022, the FDA approved the first COVID-19 diagnostic test using exhaled breath samples, this test is a portable GC-MS and is based on the detection of five VOCs from the ketone and aldehyde families associated with SARS-CoV-2 infection [12]. Nevertheless, these techniques require complex sample treatment schemes [13]. The need for sensitive, specific, portable, and low-cost techniques for the determination of VOCs in exhaled breath that can be applied in point-of-care (POC) settings has led researchers to propose emerging analytical technologies such as gas-phase infrared spectroscopy (IR). Among all the IR regions, the mid-IR region (λ= 400 to 4000 cm-1) has been applied to biological studies because the range of λ= 900 to 1800 cm-1 is the 'fingerprint region’ of biological samples [14], [15]. Several biomarkers have been assessed in exhaled breath for disease detection by this technique, however further studies are required in the development of technologies that can detect differences between exhaled breath fingerprints of patients with infectious diseases. Previous studies of our research group have developed strategies for the detection of compounds in exhaled breath via this technology [13], [16], thus demonstrating its usefulness for the detection of chemical fingerprints of VOCs in human exhaled breath. Some of the most reported VOCs in exhaled breath of COVID-19-infected individuals are ketones, aldehydes, alcohols, and nitric oxide. Studies in Edinburgh and Dortmund indicated concentrations of acetone and methanol as biomarkers that could discriminate between patients infected with SARS-CoV-2 and non-infected [17]. Acetone is generated from the decarboxylation of acetoacetate in hepatocytes, which derives from lipolysis or lipid peroxidation, and it has been associated with liver damage because of hypoxia. The liver anomalies caused by COVID-19 are also likely to cause hypoxia and therefore, acetone in breath could be possibly related to the infection. Regarding acetaldehyde, a recent study reported a simultaneous increase in acetaldehyde and acetone determined in exhaled breath of COVID-19-positive patients through GC-MS [18]. Also, in another study on exhaled breath of children with COVID-19, they reported that the breath abundance of acetaldehyde, propanal and n-propyl acetate increased during acute infection and decreased as infection resolved [19]. On the other hand, Nitric oxide (NO) indicates oxidative stress in our body, the imbalance of NO leads to more than a dozen pathophysiological conditions like hypertension, stroke, cardiac failure, CNS disorder, diabetes mellitus, and many others. The origin of NO is found in the airway epithelium, and its increased level can determine several lung diseases [20]. Analysis of peripheral blood indicated that viral infection often caused a significant increase in pro-inflammatory cytokines and chemokines and developed into a strong cytokine storm. When high inflammation persists for a long time, it causes damage to multiple tissues and organs. In addition, high inflammation causes a severe imbalance of NO/ROS in the body, which in turn leads to oxidative stress [21]. The identity of the biomarker compounds is consistent with COVID-19 derangement of breath-biochemistry by ketosis, gastrointestinal effects, and inflammatory processes. In this regard, breath analysis is attractive because it offers: i) POC location; ii) rapid results (< 10 min) without dependency on reagents; iii) non-invasive sampling with a low biosecurity burden; iv) and v) usability in a worldwide range of scenarios, including low-resourced environments such as community or primary care settings. Development and validation of technologies that address disease-specific biomarkers are of paramount importance since this approach may allow rapid diagnosis of COVID-19 in the coming endemic flu seasons. Therefore, the objective of the present study was a preliminary evaluation of VOCs reported in the breath of patients with COVID-19 (i.e., acetone (ACE), acetaldehyde (ACH), and nitric oxide (NO)) by gas-phase infrared spectroscopy via synthetic exhaled breath as a proof-of-concept principle study for future detection and monitoring of COVID-19 cases and long-COVID as a perspective. 2 Materials and methods 2.1 Calibration curves and Validation The choice of representative biomarkers in exhaled breath of people infected by the SARS-CoV-2 virus was made based on the existing literature up to the time of the beginning of the study, acetone (ACE), acetaldehyde (ACH) and nitric oxide (NO). For each of the selected biomarkers, validation was performed by establishing calibration curves, first separately for each compound and then for the three compounds in the mixture. Exhaled human breath was simulated by the addition of 4% CO2 (4-6% of CO2 is reported in exhaled breath) [22] . Additionally, for ACH a high humidity background was also added (58% r.h. @ 36°C) to simulate the real exhaled breath fingerprint and to show that these measurements are also possible with humid gases. The method validation was based on the Guide for the Validation of Analytical Methods for the Determination of Organic Compounds in trace levels AOAC/FAO/IAEA/IUPAC [23], evaluating the following parameters: limit of detection (LOD) and limit of quantification (LOQ), linearity (r), sensitivity (m), and precision (repeatability and reproducibility). For linearity, the five-point calibration curve for ACE and ACH was plotted at concentrations of 15 to 100 ppm and for NO at three points from 250 to 750 ppm. For sensitivity, the slope of the curve was evaluated. The limit of detection (LOD) and limit of quantification (LOQ) were calculated based on the standard curve according to the blank method. The precision was measured by the coefficient of variation. The repeatability was measured by three calibration curves that were obtained on the same day. Reproducibility was measured by the coefficient of variation of seven calibration curves performed on three different days. 2.2 Measurement Procedure The experimental setup is schematically shown in Fig. 1 . For the gas supply, a system of Bronkhorst and Vögtlin mass flow controllers (MFCs) (A) with different flow capacities for each of the gases was used. The MFCs were operated automatically with the use of a self-written LabVIEW program to accurately generate the calculated flows and mixtures. The total gas flow was kept at 500 mL/min.Figure 1 Measurement setup with MFCs (A), valves (B), iHWG (C), infrared spectrometer (D), OAPM (E), and infrared detector (F). All measurements were performed in a stopped-flow mode with a software-controlled valve (B) on each side of the custom 14 cm long substrate-integrated hollow waveguide (iHWG) (C) containing a 12.5 cm light guiding channel with a quadratic cross-section of 4 mm, which was closed on both sides with infrared transparent zinc selenide windows. As a light source, a Bruker Alpha II (Bruker Optics GmbH, Ettlingen, Germany) FTIR spectrometer (D) was used. The infrared light was directed by a 1'' off-axis parabolic mirror (OAPM) (E) and focused onto the gas channel of the iHWG. After interaction with the molecules in the light pathway, the infrared signal was sensed by the infrared detector (MIP-10-1M-F-M4, VIGO Systems S.A., Poznanska, Poland) (F). Before each measurement, the tubes and the iHWG, which were used as a gas cell, were purged for 3 minutes using synthetic air (20.5% O2, 79.5% N2, MTI IndustrieGase AG, Neu-Ulm, Germany) at a flow rate of 500 mL/min. After closing both valves, an IR background spectrum was recorded. Immediately after finishing the measurement, the valves opened again and a defined gas mixture from the gas mixing system was purged through the iHWG for 3 minutes. Then the valves closed again, and repeated sample measurements started. After recording five IR spectra of the sample gas, the gas flow was instantly turned only to synthetic air to get to the purging step again. 2.3 Data Acquisition The IR data acquisition was based on previous studies [16]. For IR data acquisition and processing, OPUS 8.5 (Bruker Optik GmbH, Ettlingen, Germany) software package was used. Each IR spectrum was recorded in the spectral range of 4000 to 700 cm-1 with a spectral resolution of 2 cm-1 by using a Blackman–Harris 3-term apodization function averaging 64 scans. For IR spectra evaluation of acetone, acetaldehyde, and nitric oxide, an integration method was developed with selected spectral regions. 3 Results 3.1 Validation of the analytical method The validation parameters for each one of the VOCs are presented in Table 1 . For ACE from the IR spectrum obtained, a single characteristic peak was selected for integration. The wavelength range chosen for this integration was 1397-1327 cm-1. The calibration curves were performed at concentrations of 100, 75, 50, 50, 25, and 15 ppm (Fig. 2 ).Table 1 Validation parameters for each of the investigated biomarkers. Parameter Values NO ACE ACH LOD (ppm) 6.42 13.81 9.22 LOQ (ppm) 42.26 52.57 69.23 Linearity (r) 0.998 0.999 0.999 Sensitivity (CI 95%) 0.0001±0.01 0.001±0.001 0.002±0.001 Precision Repeatability (%RSD) 3.9-10.4 2.1-17-6 3.4-13.4 Reproducibility (%RSD) 8.6-10.6 3.1-11.8 3.1-12.4 CI 95 %: Confidence Interval at 95 %; %RSD: Relative Standard Deviation Figure 2 A representative IR spectrum for establishing the calibration function for acetone. For ACH, the validation parameters are presented in Table 1. From the IR spectrum obtained, a single characteristic peak was selected for integration; the wavelengths used for integration were 1239-1137 cm-1. The calibration curves were established in the concentrations range of 100, 75, 50, 50, 25, and 15 ppm (Fig. 3 ).Figure 3 A representative IR spectrum for establishing the calibration function for acetaldehyde. For NO, from the IR spectrum obtained, a characteristic double peak was selected for integration; the wavelength regime for this integration was 1638.5-1577 cm-1. The calibration curves were established via the analysis of concentrations in the range 750, 500, and 250 ppm (Table 1) (Fig. 4 ).Figure 4 A representative IR spectrum for establishing the calibration function for nitric oxide. 3.2 VOC mixtures For a better understanding and representation of the exhaled breath biological matrix, the CO2 concentration equivalent to the concentration found in naturally exhaled breath (4%) was added. Thirty-one measurements of the mixture containing different concentrations of each of the gases were performed in duplicate (Fig. 5 a), to accurately quantify the biomarkers in the mixture and to establish the least variability between measurements. Representative spectra of two different concentrations of the biomarkers are shown in Fig. 5b and c. It is important to note that the quantification of the complex gas mixture was successfully performed, even with the presence of CO2. This demonstrates the capacity of the applied technology for the determination and quantification of gases in a corresponding mixture.Figure 5 5a. A representative IR spectrum of the VOC mixture fingerprint including CO2. CO2 at 4%, acetone at 100 ppm, acetaldehyde at 25 ppm and nitric oxide at 250 ppm. 5b. A representative IR spectrum of the VOC mixture fingerprint with CO2. CO2 at 4%, acetone at 25 ppm and acetaldehyde at 75 ppm. 5c. A representative IR spectrum of the VOC mixture fingerprint with CO2. CO2 at 4%, acetone at 25 ppm, acetaldehyde at 100 ppm and NO 250 ppm. 3.3 COVID-19-infected exhaled breath simulation For the representation of the synthetic exhaled breath of people with lung damage caused by SARS-CoV-2 infection, a calibration curve with acetaldehyde, 4% CO2, and 58% relative humidity at a temperature of 35°C was performed. Only the quantification of acetaldehyde was possible due to the contribution of H2O, which dominates in the mixture and overlaps with the other biomarkers. Nevertheless, as a perspective from these results the contribution from the humidity of the sample will be reduced. Fig. 6 a shows a representative IR spectrum of the analysis, where the peaks obtained for CO2, H2O, and the peak quantified for acetaldehyde are shown. The calibration curve was established using the following concentrations of acetaldehyde: 100, 75, 50, 25, and 15 ppm (Fig. 6b).Figure 6 6a. A representative IR spectrum of synthetic exhaled breath with 95.5 ppm ACH, 4% CO2 and 58% relative humidity at 35°C. 6b. IR spectra used for the acetaldehyde calibration curve with 4% CO2 and 58% relative humidity at 35°C. 4 Discussion In the present study, a proof-of-concept principle, for the evaluation of specific VOC fingerprints indicative of SARS-CoV-2 virus infection, in synthetic exhaled breath by application of gas-phase infrared spectroscopy is presented. The public health system's capacity relies on detecting, testing, contact tracing, and isolating those who are or might be sick or have been exposed to known or suspected COVID-19 cases. The principal challenge lies in the identification of the cases, currently, RTq-PCR remains the gold standard in the detection of those infected, in the same way, the application of antigen and antibody tests for mass screening has been established. These technologies have certain disadvantages such as low specificity, high percentages of false positives and negatives, and invasiveness in the sample collection, which decreases its acceptance, but also, they are not accessible to all people due to the high economic cost [24], [25]. With the continuous occurrence of new SARS-CoV-2 virus variants, and the fact that they behave differently in the host and among the same populations, it is necessary to develop and apply sensitive techniques that can establish and determine specific chemical fingerprints based on the metabolism of the virus interaction with the host and discriminate between other infections, thus providing an efficient, accurate and rapid result for case counting, also considering the long-COVID scenario. It is of outmost importance to develop techniques which could discriminate between long-COVID, COVID-19, other lung diseases and infections, and could serve as a continuous monitoring of these patients. It is hypothesized that the breath fingerprint will change in the course of the infection and therefore in the course of long-COVID development and subsequent rehabilitation. The study of volatile organic compounds in exhaled breath is of great relevance in the present context. This has been performed by different technologies; however, these represent high costs, complex and time-consuming sample preparation processes, sensitivity and specificity issued as well as the need for a highly qualified analysts for their performance and the subsequent analysis of the results. In addition, they are difficult to transport to POC settings and remote locations, with the need for a laboratory and certain conditions for the analysis and the analyst [26], [27]. There is strong evidence that the analysis of trace components in exhaled breath could provide a complementary approach to non-invasive monitoring of inflammation, oxidative stress, and other processes in the airways and lungs [26]. Optical analytical techniques with inherently high sensitivity and specificity are ideally suited to fill this gap. A significant advantage of gas-phase infrared spectroscopy is the ability to perform online measurements next to exquisite molecular selectivity [28], [29]. In this context exhaled breath is analyzed during exhalation in real-time, whereas offline techniques rely on collecting breath samples in a bag or a sorbent trap for subsequent analysis. The potential issues of offline methods, like reproducibility of breath sample collection, contamination during sample storage, and the inability to allow for instantaneous feedback can be avoided using online methods. Additionally, information on the concentration during different exhalation phases is directly accessible via fast online studies, whereas offline methods integrate across a complete exhalation cycle or require an extra effort to separate exhaled gas coming from the lungs from gas that originates from the upper airways (i.e., dead space air) [30], [31], [32]. There are several clinical applications where real-time breath analysis would have an immediate application [32], [33] such as monitoring the progress of the infection and the long-term effects in long-COVID scenarios. Spectroscopy in the mid-infrared spectral region, therefore, appears most advantageous, since most of the gaseous compounds of biomedical interest are molecular gases that have characteristic and pronounced absorption bands in this spectral region. Addressing these fingerprint spectra, therefore, allows sensitive, specific and rapid monitoring of gas mixtures [34], [35], [36], [37]. This technology has been applied in COVID-19 screening scenario, Liang et al., 2023, presented an approach of breath analysis by ultra-sensitive broadband laser spectroscopy, which in turn works in the same range -MIR-; with a total of 170 individual breath samples they reported great discrimination with an area under the receiver-operating-characteristics curve of 0.849(4), nevertheless in their approach they still transport and process samples offline [38]. This demonstrates the feasibility of application of gas-phase IR spectroscopy in disease screening with breath sampling. The VOCs selected in the present study have been reported as relevant biomarkers in the breath of people with pulmonary conditions such as COPD, lung cancer, and COVID-19. In this context, exhaled ACH reflects aspects of oxidative stress and metabolism, the relationship of oxygenated compounds such as ketones and aldehydes with oxidative stress and the regulation of cell proliferation is well established. In the case of acute COVID-19, an inflammatory process occur due to the hyperactivity of the immune system which involves a state of cellular hyperinflammation [39], but also this state may remain in the long-COVID stage until the patient's rehabilitation [40]. ACH is derived, along with hydrocarbons, from lipid peroxidation and inflammatory processes and has been reported widely in a range of respiratory conditions [11], [17], [41]. In this context, oxidative stress in COVID-19 patients may be caused when the organism interacts with the virus, generating local inflammation and thus releasing the compound in the exhaled breath. On the other hand, ACE is considered an oxygen-containing compound produced from glucose metabolism, and it has been described as a biomarker of lung health in different pathologies such as lung cancer [42], cystic fibrosis [43], asthma [44], and – in the present context – has been described as an indicator VOC in the breath of patients with COVID-19 [17], [27], [41]. On the other hand, it has been shown that in the acute phase of COVID-19, a hypermetabolic phase occur and it demands energy consumption [45]. In a study by Ruskiewicz et al., they report that the identity of the marker compounds identified in this study are consistent with a combination of extrapulmonary metabolic, and gastrointestinal manifestations of COVID-19 within the body as well as airway inflammatory responses [17]. Nitric oxide (NO) plays a major role in pulmonary and cardiovascular physiology. Is a reactive oxygen species (ROS) continually produced by epithelial cells of the paranasal sinuses and nasopharynx via NO synthase (NOS) enzymes [46], [47], [48]. NO diffuses to the bronchi and lungs, where it induces vasodilatory and bronchodilatory effects. This biomarker is involved in several biological roles such as activation of ciliary movements, mucus secretion, antimicrobial effects, and virus inactivation [49], [50], [51]. Concerning COVID-19, a distinctive feature of endothelial dysfunction and thrombotic episodes may be associated with the suppression of endothelial nitric oxide synthase (eNOS) with a concomitant deficiency of NO. In healthy vessels, nitric oxide is released to the endothelium as a vasodilator and antithrombotic factor; on the other hand, in injured vessels, NO is impaired, contributing to hypertension and thrombus formation [52]. Table 2 presents a comparative study of biomarkers in the different pulmonary conditions. At the time of the study, no information was found about the exact concentrations of biomarkers in patients infected with the SARS-CoV-2 virus, however, these concentrations are usually in the ppb to ppm range, so among the limitations considered in the present study are the high concentrations used in this proof of concept; it remains as a research perspective to test and improve the technology at real-life scenarios and also to improve detection limits (sensitivity). Nevertheless, the proposed technology provides a complete fingerprint of the VOCs present in exhaled breath, which expands upon the information provided by addressing only three biomarkers. Hence, another perspective is the analysis of actual exhaled breath in patients infected with the SARS-CoV-2 virus. In this context, to discriminate COVID-19 from other respiratory diseases including long-COVID is not sufficient for clinical test application because other respiratory diseases have similar symptoms and biochemical backgrounds, which could confound differential diagnosis. Therefore, the value of a new test is not only to distinguish COVID-19 patients from healthy people but also to identify patients with other lung diseases. Even though spectroscopic techniques offer sensitivities in the ppb to ppt concentration range, there are still several improvements required to render MIR spectroscopy a useful clinical tool in routine breath gas monitoring. Nevertheless, several studies have shown that multiple rather than individual VOCs give rise to pattern changes across the entire IR spectra, which may be used to be more reliably associated with a particular pathological condition. Also, if this technique is coupled with other orthogonal techniques such as VOCs sensors, it could improve and enhance the obtained information to complement breath fingerprints and generate more sensitive and specific POC devices.Table 2 Comparison of exhaled breath biomarkers, concentrations, and assessment techniques. Biomarker Disease Concentration Analytical methodology Reference Nitric Oxide Asthma >15 ppb Chemoluminescence analyzer [53] Control 4-9 ppb Control 13 ppb GC-MS [54] COPD 9.8 ± 0.7 ppb Chemoluminescence analyzer [55] Smokers 10.3 ± 1.0 ppb Control 5.5 ± 0.4 ppb Women 1.6-21.5 ppb Chemoluminescence analyzer [56] Men 2.6-28.8 ppb Asbestosis 20 ppb Chemoluminescence analyzer [57] Pulmonary arterial hypertension 13.9 ± 6.8 ppb Chemoluminescence analyzer [58] Pulmonary Hypertension for COPD 15.0 ± 9.3ppb Pulmonary Hypertension for Heart disease 6.7 ± 2.0 ppb COPD 157.3 ppb Electrochemical voltammetric sensors [59] COPD 10 ppb Electrochemical sensor-based device [60] Acetone Lung Cancer 500 ppb CO2 laser photoacoustic spectrometer [42] Lung Disease 350 ppb Control 300 ppb Lung Cancer 250->2500 ppb Portable gas chromatograph-In2O3 thick film-type gas sensor [61] Pulmonary arterial hypertension 16.54 ± 14.7 nMol/L Microreactor-FTIR ion cyclotron resonance mass [62] Lung Cancer 6.97 ± 7.02 nMol/L ARDS >300 ppb IMR-MS [63] Lung Cancer 4.41-96.3 ppb GC-MS [64] Acetaldehyde Lung Cancer 0-55 nMol/L GC-MS [65] Lung Cancer 25 ppb SIFT-MS [66] Controls < 3 ppb SIFT-MS [67] Ventilated patients 13 ± 5 ppb MCC–IMS [68] All techniques presented in Table 2 are highly sensitive, however, involve very high instrumental costs, require highly trained personnel, and are laboratory-based tools. An advantage of gas-phase infrared spectroscopy is that the analysis can be performed online and on-site, which provides the user with a technology that can be adapted for POC settings enabling ease of application, accessibility, and rapid and specific analysis that allows for effective monitoring and case tracking. This virus is not the first and will not be the last to cause a pandemic. In the past, mankind has witnessed many virus-derived pandemics, and the potential for new pandemics in the future due to the emergence of new viruses with mutational and lethal infection capacity cannot be ignored. The lessons learned from this situation include that it is necessary to invest in widespread monitoring for predicting pandemic risks to take early action in avoidance of global pandemic scenarios. The emerging area of IR-spectroscopic exhaled breath sensing is likely to continue to yield promising results as a complementary tool for testing for SARS-CoV-2 infections with the benefits of potential applications for large-scale field screening studies. In this context and with the obtained results, the capability of gas-phase infrared spectroscopy for the detection of global VOC fingerprints in exhaled breath has been demonstrated, even though significantly more work is needed to push the technology into practice. 5 Conclusion In the ongoing COVID-19 pandemic and its sequelae, the development of new platforms for screening large populations is of vital importance. Encouraging data is presented using IR-spectroscopic gas sensing concepts for future faster and cheaper screening of infected individuals. Gas-phase infrared spectroscopy could potentially improve SARS-CoV-2 screening in POC settings, where the use of exhalomics could assist in large-scale diagnosis and further differentiation between lung diseases with the establishment of disease-specific molecular IR fingerprinting providing rapid information on the health status of persons (i.e., infected vs. not infected) and facilitating monitoring and follow-up of cases. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available on request. Acknowledgements This study was supported by the Ministerium für Wissenschaft, Forschung und Kunst (MWK) in Baden-Württemberg, Germany within the program “Sonderförderlinie COVID-19”. AUTHOR CONTRIBUTIONS JG: Conceptualization, Device design, Writing and Editing; BM: Conceptualization, Funding, Writing and Editing; LDLM: Conceptualization, Experimental, Writing and Editing. ==== Refs References 1 Girum T. Lentiro K. Geremew M. Migora B. Shewamare S. Global strategies and effectiveness for COVID-19 prevention through contact tracing, screening, quarantine, and isolation: a systematic review Trop. Med. Health. 48 2020 1 15 10.1186/s41182-020-00285-w 31920458 2 Walker P.G.T. Whittaker C. Watson O.J. Baguelin M. Winskill P. Hamlet A. Djafaara B.A. Cucunubá Z. Olivera Mesa D. Green W. The impact of COVID-19 and strategies for mitigation and suppression in low-and middle-income countries Science. 369 2020 413 422 10.1126/science.abc0035 32532802 3 Chaimayo C. Kaewnaphan B. Tanlieng N. Athipanyasilp N. Sirijatuphat R. Chayakulkeeree M. Angkasekwinai N. Sutthent R. Puangpunngam N. Tharmviboonsri T. Pongranweewan O. Chuthapisith S. Sirivatanauksorn Y. Kantakamalakul W. Horthongkham N. Rapid SARS-CoV-2 antigen detection assay in comparison with real-time RT-PCR assay for laboratory diagnosis of COVID-19 in Thailand Virol. J. 17 2020 177 10.1186/s12985-020-01452-5 33187528 4 Augustine R. Das S. Hasan A. Abdul Salam S. Augustine P. Dalvi Y.B. Varghese R. Primavera R. Yassine H.M. Thakor A.S. Rapid antibody-based COVID-19 mass surveillance: relevance, challenges, and prospects in a pandemic and post-pandemic world J Clin. Med. 9 2020 3372 10.3390/jcm9103372 33096742 5 Ravi N. Cortade D.L. Ng E. Wang S.X. Diagnostics for SARS-CoV-2 detection: A comprehensive review of the FDA-EUA COVID-19 testing landscape Biosens. Bioelectron. 165 2020 112454 10.1016/j.bios.2020.112454 6 T.C. Miller, S.D. Morgera, S.E. Saddow, A. Takshi, M. Palm, Electronic nose with detection method for alcohol, acetone, and carbon monoxide in coronavirus disease 2019 breath simulation model, IEEE Sens. J. 21 (2021) 15935–15943. https://doi.org/10.1109/JSEN.2021.3076102. 7 Dragonieri S. Schot R. Mertens B.J.A. Le Cessie S. Gauw S.A. Spanevello A. Resta O. Willard N.P. Vink T.J. Rabe K.F. An electronic nose in the discrimination of patients with asthma and controls J. Allergy Clin. Immunol. 120 2007 856 862 10.1016/j.jaci.2007.05.043 17658592 8 Tirzīte M. Bukovskis M. Strazda G. Jurka N. Taivans I. Detection of lung cancer in exhaled breath with an electronic nose using support vector machine analysis J. Breath Res. 11 2017 36009 10.1088/1752-7163/aa7799 9 de Vries R. Vigeveno R.M. Mulder S. Farzan N. Vintges D.R. Goeman J.J. Bruisten S. van den Corput B. Geelhoed J.J.M. Visser L.G. Ruling out SARS-CoV-2 infection using exhaled breath analysis by electronic nose in a public health setting MedRxiv. 2021 10.1101/2021.02.14.21251712 10 Grassin-Delyle S. Roquencourt C. Moine P. Saffroy G. Carn S. Heming N. Fleuriet J. Salvator H. Naline E. Couderc L.-J. Metabolomics of exhaled breath in critically ill COVID-19 patients: A pilot study EBioMedicine. 63 2021 103154 10.1016/j.ebiom.2020.103154 11 Gould O. Ratcliffe N. Król E. de Lacy Costello B. Breath analysis for detection of viral infection, the current position of the field J. Breath Res. 14 2020 41001 12 FDA, InspectIR COVID-19 Breathalyzer (for use on PNY-1000), 2022. https://doi.org/10.1088/1752-7163/ab9c32. 13 Selvaraj R. Vasa N.J. Nagendra S.M.S. Mizaikoff B. Advances in mid-infrared spectroscopy-based sensing techniques for exhaled breath diagnostics Molecules. 25 2020 2227 10.3390/molecules25092227 32397389 14 Sitole L. Steffens F. Krüger T.P.J. Meyer D. Mid-ATR-FTIR spectroscopic profiling of HIV/AIDS sera for novel systems diagnostics in global health OMICS J. Integr. Biol. 18 2014 513 523 10.1089/omi.2013.0157 15 T. Jin, J. Zhou, P.T. Lin, Mid-infrared waveguides for volatile organic compounds detection, in: Opt. Photonics Sens. Environ., Optical Society of America, 2019: p. EW2A. 2. https://opg.optica.org/abstract.cfm?URI=ES-2019-EW2A.2 16 Glöckler J. Jaeschke C. Kocaöz Y. Kokoric V. Tütüncü E. Mitrovics J. Mizaikoff B. iHWG-MOX: a hybrid breath analysis system via the combination of substrate-integrated hollow waveguide infrared spectroscopy with metal oxide gas sensors ACS Sensors. 5 2020 1033 1039 10.1021/acssensors.9b02554 32189494 17 Ruszkiewicz D.M. Sanders D. O’Brien R. Hempel F. Reed M.J. Riepe A.C. Bailie K. Brodrick E. Darnley K. Ellerkmann R. Diagnosis of COVID-19 by analysis of breath with gas chromatography-ion mobility spectrometry-a feasibility study EClinicalMedicine. 29 2020 100609 10.1016/j.eclinm.2020.100609 18 Török Z.-M. Blaser A.F. Kavianynejad K. de Torrella C.G.M.G. Nsubuga L. Mishra Y.K. Rubahn H.-G. de Oliveira Hansen R. Breath Biomarkers as Disease Indicators: Sensing Techniques Approach for Detecting Breath Gas and COVID-19 Chemosensors. 10 2022 167 10.3390/chemosensors10050167 19 Berna A.Z. Akaho E.H. Harris R.M. Congdon M. Korn E. Neher S. M’Farrej M. Burns J. Odom John A.R. Reproducible Breath Metabolite Changes in Children with SARS-CoV-2 Infection, ACS Infect. Dis. 7 2021 2596 2603 10.1021/acsinfecdis.1c00248 20 Maurya M.R. Onthath H. Morsy H. Riyaz N.-U.-S. Ibrahim M. Ahmed A.E. Abuznad R. Alruwaili A. Alsaedi F. Kasak P. Sadasivuni K.K. Colorimetry-Based Detection of Nitric Oxide from Exhaled Breath for Quantification of Oxidative Stress in Human Body Healthcare. 9 2021 1055 10.3390/healthcare9081055 34442192 21 Fang W. Jiang J. Su L. Shu T. Liu H. Lai S. Ghiladi R.A. Wang J. The role of NO in COVID-19 and potential therapeutic strategies Free Radic. Biol. Med. 163 2021 153 162 10.1016/j.freeradbiomed.2020.12.008 33347987 22 M.A. Anghel, F. Iacobescu, The Influence of Temperature and CO2 in Exhaled Breath, in: 16th Int. Congr. Metrol., EDP Sciences, 2013: p. 10012. https://doi.org/10.1051/metrology/201310012 23 Fajgelj A. Ambrus A. Guidelines for single-laboratory validation of analytical methods for trace-level concentrations of organic chemicals in: Principles and practices of method validation 2000 179 252 10.1039/9781847551757-00179 24 Dramé M. Teguo M.T. Proye E. Hequet F. Hentzien M. Kanagaratnam L. Godaert L. Should RT-PCR be considered a gold standard in the diagnosis of COVID-19? J Med Virol. 92 11 2020 Nov 2312 2313 10.1002/jmv.25996 32383182 25 Roy S. Physicians’ dilemma of false-positive RT-PCR for COVID-19: a case report SN Compr. Clin. Med. 3 2021 255 258 10.1007/s42399-020-00655-9 33426468 26 Banik G.D. Mizaikoff B. Exhaled breath analysis using cavity-enhanced optical techniques: a review J. Breath Res. 14 2020 43001 10.1088/1752-7163/abaf07 27 Rodríguez-Aguilar M. Díaz de León-Martínez L. Zamora-Mendoza B.N. Comas-García A. Guerra Palomares S.E. García-Sepúlveda C.A. Alcántara-Quintana L.E. Díaz-Barriga F. Flores-Ramírez R. Comparative analysis of chemical breath-prints through olfactory technology for the discrimination between SARS-CoV-2 infected patients and controls Clin. Chim. Acta. 519 2021 126 132 10.1016/j.cca.2021.04.015 33901429 28 Henderson B. Khodabakhsh A. Metsälä M. Ventrillard I. Schmidt F.M. Romanini D. Ritchie G.A.D. te Lintel Hekkert S. Briot R. Risby T. Marzcin N. Harren F.J.M. Critescu S.M. Laser spectroscopy for breath analysis: towards clinical implementation Appl. Phys. B. 124 2018 161 10.1007/s00340-018-7030-x 30956412 29 Bruderer T. Gaisl T. Gaugg M.T. Nowak N. Streckenbach B. Müller S. Moeller A. Kohler M. Zenobi R. On-line analysis of exhaled breath: focus review Chem. Rev. 119 2019 10803 10828 10.1021/acs.chemrev.9b00005 31594311 30 Herbig J. Müller M. Schallhart S. Titzmann T. Graus M. Hansel A. On-line breath analysis with PTR-TOF J. Breath Res. 3 2009 027004 10.1088/1752-7155/3/2/027004 31 Wang C. Sahay P. Breath Analysis Using Laser Spectroscopic Techniques: Breath Biomarkers Spectral Fingerprints, and Detection Limits, Sensors. 9 2009 8230 8262 10.3390/s91008230 22408503 32 Boshier P.R. Cushnir J.R. Mistry V. Knaggs A. Španěl P. Smith D. Hanna G.B. On-line, real time monitoring of exhaled trace gases by SIFT-MS in the perioperative setting: a feasibility study Analyst. 136 2011 3233 3237 10.1039/C1AN15356K 21717028 33 Metsälä M. Optical techniques for breath analysis: From single to multi-species detection J. Breath Res. 12 2018 27104 34 Kim S.-S. Young C. Vidakovic B. Gabram-Mendola S.G.A. Bayer C.W. Mizaikoff B. Potential and challenges for mid-infrared sensors in breath diagnostics IEEE Sens. J. 10 2009 145 158 10.1088/1752-7163/aa8a31 35 Straume T. Loftus D.J. Li J. Coleman M.A. Davis C.E. McMonigal K.A. Piccini M. Singh A.K. Biomarker-detection technologies for comprehensive medical diagnosis during deep-space missions Recent Patents Sp. Technol. 3 2013 13 23 https://www.ingentaconnect.com/content/ben/rptst/2013/00000003/00000001/art00004 36 L. Bodiou, J. Lemaitre, A. Gutierrez, W. El Ayed, Y. Dumeige, I. Hardy, J. Charrier, F. Starecki, E. Baudet, R. Chahal, V. Nazabal, J.L. Doualan, A. Braud, P. Camy and P. Nemec, Rare-earth doped selenides active waveguides for integrated mid-infrared sensing applications, in: 19th Eur. Conf. Integr. Opt. (ECIO 2017), 2017. https://www.ecio-conference.org/wp-content/uploads/2017/03/ECIO_2017_T7.5.pdf 37 Tütüncü E. Mizaikoff B. Cascade laser sensing concepts for advanced breath diagnostics Anal. Bioanal. Chem. 411 2019 1679 1686 10.1007/s00216-018-1509-5 30565171 38 Liang Q. Chan Y.-C. Toscano J. Bjorkman K.K. Leinwand L.A. Parker R. Nozik E.S. Nesbitt D.J. Ye J. Breath analysis by ultra-sensitive broadband laser spectroscopy detects SARS-CoV-2 infection J. Breath Res. 17 2023 36001 10.1088/1752-7163/acc6e4 39 Vabret N. Britton G.J. Gruber C. Hegde S. Kim J. Kuksin M. Levantovsky R. Malle L. Moreira A. Park M.D. Immunology of COVID-19: current state of the science Immunity. 52 2020 910 941 10.1016/j.immuni.2020.05.002 32505227 40 Afrin L.B. Weinstock L.B. Molderings G.J. COVID-19 hyperinflammation and post-COVID-19 illness may be rooted in mast cell activation syndrome Int. J. Infect. Dis. 100 2020 327 332 10.1016/j.ijid.2020.09.016 32920235 41 Chen H. Qi X. Ma J. Zhang C. Feng H. Yao M. Breath-borne VOC biomarkers for COVID-19 J. Breath Res. 15 2021 047104 10.1088/1752-7163/ac2e57 42 Mitrayana D.K. Apriyanto M. Satriawan, CO2 Laser Photoacoustic Spectrometer for Measuring Acetone in the Breath of Lung Cancer Patients Biosensors. 10 2020 55 10.3390/bios10060055 32471087 43 Barker M. Hengst M. Schmid J. Buers H.-J. Mittermaier B. Klemp D. Koppmann R. Volatile organic compounds in the exhaled breath of young patients with cystic fibrosis Eur. Respir. J. 27 2006 929 936 10.1183/09031936.06.00085105 16455833 44 Smolinska A. Klaassen E.M.M. Dallinga J.W. Van De Kant K.D.G. Jobsis Q. Moonen E.J.C. Van Schayck O.C.P. Dompeling E. Van Schooten F.J. Profiling of volatile organic compounds in exhaled breath as a strategy to find early predictive signatures of asthma in children PLoS One. 9 2014 e95668 24752575 45 Lakenman P.L.M. Van der Hoven B. Schuijs J.M. Eveleens R.D. van Bommel J. Olieman J.F. Joosten K.F.M. Energy expenditure and feeding practices and tolerance during the acute and late phase of critically ill COVID-19 patients Clin. Nutr. ESPEN. 43 2021 383 389 10.1016/j.clnesp.2021.03.019 34024544 46 P. Sogni, P. Garnier, A. Gadano, R. Moreau, J. Dall’Ava-Santucci, A.T. Dinh-Xuan, D. Lebrec, Endogenous pulmonary nitric oxide production measured from exhaled air is increased in patients with severe cirrhosis, J. Hepatol. 23 (1995) 471–473. https://doi.org/10.1016/0168-8278(95)80207-X 47 M. Özkan, R.A. Dweik, Nitric oxide and airway reactivity, Clin. Pulm. Med. 8 (2001) 199–206. https://journals.lww.com/clinpulm/Fulltext/2001/07000/Nitric_Oxide_and_Airway_Reactivity.1.aspx 48 Högman M. Thornadtsson A. Liv P. Hua-Huy T. Dinh-Xuan A.-T. Tufvesson E. Dressel H. Janson C. Koskela K. Oksa P. Effects of growth and aging on the reference values of pulmonary nitric oxide dynamics in healthy subjects J. Breath Res. 11 2017 47103 10.1088/1752-7163/aa7957 49 Keyaerts E. Vijgen L. Chen L. Maes P. Hedenstierna G. Van Ranst M. Inhibition of SARS-coronavirus infection in vitro by S-nitroso-N-acetylpenicillamine, a nitric oxide donor compound Int. J. Infect. Dis. 8 2004 223 226 10.1016/j.ijid.2004.04.012 15234326 50 Åkerström S. Gunalan V. Keng C.T. Tan Y.-J. Mirazimi A. Dual effect of nitric oxide on SARS-CoV replication: viral RNA production and palmitoylation of the S protein are affected Virology. 395 2009 1 9 10.1016/j.virol.2009.09.007 19800091 51 Jung K. Gurnani A. Renukaradhya G.J. Saif L.J. Nitric oxide is elicited and inhibits viral replication in pigs infected with porcine respiratory coronavirus but not porcine reproductive and respiratory syndrome virus Vet. Immunol. Immunopathol. 136 2010 335 339 10.1016/j.vetimm.2010.03.022 20409593 52 Green K. Graziadio S. Turner P. Fanshawe T. Allen J. Molecular and antibody point-of-care tests to support the screening, diagnosis and monitoring of COVID-19 2020 Cent Evidence-Based Med https://www.cebm.net/wp-content/uploads/2020/04/POCT-Covid19.pdf 53 Hansel T.T. Kharitonov S.A. Donnelly L.E. Erin E.M. Currie M.G. Moore W.M. Manning P.T. Recker D.P. Barnes P.J. A selective inhibitor of inducible nitric oxide synthase inhibits exhaled breath nitric oxide in healthy volunteers and asthmatics FASEB J. 17 2003 1298 1300 10.1096/fj.02-0633fje 12738811 54 Leone A.M. Gustafsson L.E. Francis P.L. Persson M.G. Wiklund N.P. Moncada S. Nitric oxide is present in exhaled breath in humans: direct GC-MS confirmation Biochem. Biophys. Res. Commun. 201 1994 883 887 10.1006/bbrc.1994.1784 8003027 55 Liu J. Sandrini A. Thurston M.C. Yates D.H. Thomas P.S. Nitric Oxide and Exhaled Breath Nitrite/Nitrates in Chronic Obstructive Pulmonary Disease Patients Respiration. 74 2007 617 623 10.1159/000106379 17643055 56 Olivieri M. Talamini G. Corradi M. Perbellini L. Mutti A. Tantucci C. Malerba M. Reference values for exhaled nitric oxide (reveno) study Respir. Res. 7 2006 94 10.1186/1465-9921-7-94 16813647 57 Lehtonen H. Oksa P. Lehtimäki L. Sepponen A. Nieminen R. Kankaanranta H. Saarelainen S. Järvenpää R. Uitti J. Moilanen E. Increased alveolar nitric oxide concentration and high levels of leukotriene B4 and 8-isoprostane in exhaled breath condensate in patients with asbestosis Thorax. 62 2007 602 607 10.1136/thx.2006.067868 17251310 58 Carpagnano G.E. Radaeli A. Lacedonia D. Correale M. Carpagnano G. Palmiotti A. Barbaro M.P.F. Di Biase M. Brunetti N. Scioscia G. Malerba M. Exhaled nitric oxide and exhaled breath temperature as potential biomarkers in patients with pulmonary hypertension, Biomed Res Int. 2018 2018 7292075 10.1155/2018/7292045 59 Sun Y. Wang X.-M. Chen Y.-H. Zhu R. Liao C.-C. Exhaled hydrogen sulfide in patients with chronic obstructive pulmonary disease and its correlation with exhaled nitric oxide Chin. Med. J. (Engl) 126 2013 3240 3244 10.3760/cma.j.issn.0366-6999.20123064 24033943 60 Shrestha S.K. Shrestha S. Sharma L. Pant S. Neopane A. Comparison of fractional exhaled nitric oxide levels in chronic obstructive pulmonary disease, bronchial asthma and healthy subjects of Nepal J. Breath Res. 11 2017 047101 10.1088/1752-7163/aa7e63 61 Tanda N. Hoshikawa Y. Sato T. Takahashi N. Koseki T. Exhaled acetone and isoprene in perioperative lung cancer patients under intensive oral care: possible indicators of inflammatory responses and metabolic changes Biomed. Res. 40 2019 29 36 10.2220/biomedres.40.29 30787261 62 Jalil B.A. Xie Z. Li Q. Fu X.-A. El-Kersh K. Volatile Organic Compounds in Exhaled Breath of Patients with Pulmonary Arterial Hypertension: A Comparative Analysis, Ain B55. USE YOUR Illusion I Clin. Res 2019 PAH, American Thoracid Society A3644 A 10.1164/ajrccm-conference.2019.199.1_MeetingAbstracts.A3644 63 Meidert A.S. Choukèr A. Praun S. Schelling G. Dolch M.E. Exhaled Breath and Oxygenator Sweep Gas Propionaldehyde in Acute Respiratory Distress Syndrome Molecules. 26 2020 145 10.3390/molecules26010145 33396196 64 Koureas M. Kirgou P. Amoutzias G. Hadjichristodoulou C. Gourgoulianis K. Tsakalof A. Target Analysis of Volatile Organic Compounds in Exhaled Breath for Lung Cancer Discrimination from Other Pulmonary Diseases and Healthy Persons Metabolites. 10 2020 317 10.3390/metabo10080317 32756521 65 Fuchs P. Loeseken C. Schubert J.K. Miekisch W. Breath gas aldehydes as biomarkers of lung cancer Int. J. Cancer. 126 2010 2663 2670 10.1002/ijc.24970 19839051 66 Španěl P. Smith D. Quantification of volatile metabolites in exhaled breath by selected ion flow tube mass spectrometry, SIFT-MS Clin. Mass Spectrom. 16 2020 18 24 10.1016/j.clinms.2020.02.001 34820516 67 Huang J. Kumar S. Hanna G.B. Investigation of C3–C10 aldehydes in the exhaled breath of healthy subjects using selected ion flow tube-mass spectrometry (SIFT-MS) J. Breath Res. 8 2014 37104 10.1088/1752-7155/8/3/037104 68 Müller-Wirtz L.M. Kiefer D. Ruffing S. Brausch T. Hüppe T. Sessler D.I. Volk T. Fink T. Kreuer S. Maurer F. Quantification of Volatile Aldehydes Deriving from In Vitro Lipid Peroxidation in the Breath of Ventilated Patients Molecules. 26 2021 3089 10.3390/molecules26113089 34064214
PMC010xxxxxx/PMC10286899.txt	==== Front Language and Health 2949-9038 2949-9038 The Author(s). Published by Elsevier B.V. on behalf of Shandong University. S2949-9038(23)00001-5 10.1016/j.laheal.2023.05.001 Article Long COVID: The impact on language and cognition Cummings Louise ⁎1 Department of English and Communication, The Hong Kong Polytechnic University, Hong Kong Special Adminitrative Region of China ⁎ Correspondence to: Department of English and Communication, The Hong Kong Polytechnic University, 11 Yuk Choi Rd, Hung Hom, Kowloon, Hong Kong Special Adminitrative Region of China. 1 ORCID ID: https://orcid.org/0000-0002-8947-8195 22 6 2023 22 6 2023 10 3 2023 3 5 2023 16 5 2023 © 2023 The Authors 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. COVID-19 continues to have profound health and economic consequences around the world. Aside from the large number of deaths from this viral infection, there is a growing population of individuals who have not made a good recovery from their COVID illnesses. These children and adults continue to experience COVID symptoms for months and even years after the onset of their illness. One group of symptoms that can be particularly troubling are language and cognitive difficulties. These difficulties can compromise learning and academic attainment and prevent a return to employment in adults. The author has examined the language skills of 110 adults who reported experiencing Long COVID. Among these individuals, 99 adults reported significant cognitive-linguistic difficulties as part of their ongoing COVID symptoms. This article examines these difficulties in detail. It proposes that these cognition-based language difficulties should be included in the class of cognitive-communication disorders. These disorders are typically assessed and treated by speech-language pathologists who manage communication difficulties in clients with traumatic brain injury, right-hemisphere damage, and neurodegeneration. Keywords Brain fog Cognitive-communication disorder COVID-19 Language disorder Long COVID Speech-language pathology ==== Body pmc1 Introduction The COVID-19 pandemic has left its mark on human health in ways that extend beyond a large and tragic number of global deaths. For many individuals who survived their COVID infections, and had mild to moderate illnesses initially, the virus has posed a serious and persistent threat to health in the form of Long COVID. It is not an exaggeration to say that this debilitating condition has destroyed the lives of those who have developed what the World Health Organization calls the “post COVID-19 condition”. Apart from ongoing physical symptoms, it is now widely recognized that cognitive dysfunction is part of Long COVID. This manifests in all manner of problems with memory, planning, reasoning, and attention. Importantly, language is also compromised, leading to significant communication problems that impact on daily functioning in areas such as employment, social relationships, and family role. In previous work, the author has characterized these language difficulties and their impact on functioning in a study of 92 adults with Long COVID (Cummings, 2023a) and in an online survey of 973 adults with self-reported Long COVID and “brain fog” (Cummings, 2023b). In this article, the focus of discussion will be on specific linguistic impairments that these adults experience, with illustration provided through data recorded and analyzed by the author. The article unfolds as follows. In Section 2, Long COVID is defined by examining the clinical case definition of the post COVID-19 condition that is adopted by the World Health Organization. The language difficulties that are the focus of this discussion are part of a wider group of cognitive-linguistic difficulties that have been labelled as “brain fog” by people with Long COVID. Section 2 also examines the types of difficulties that constitute “brain fog” as well as the prevalence of these difficulties. In Section 3, specific linguistic difficulties are examined and illustrated by using data from adults with Long COVID as they perform a range of language tasks. These tasks include activities like narrative production and confrontation naming. It will be argued that some apparently “normal” scores in these tasks mask considerable linguistic difficulties that might go unnoticed if performance is based solely on language test scores. The health professionals who assess and treat language difficulties are speech-language pathologists. How cognitive-linguistic difficulties in Long COVID may best be conceptualized within the type of diagnostic categories that are recognized by these health professionals is addressed in Section 4. 2 What is Long COVID? Terms such as Long COVID and COVID long haulers emerged early in the COVID-19 pandemic to describe people for whom the symptoms of COVID illness persisted well beyond the point at which viral illnesses normally resolve. Although these terms are widely used by patient groups and media outlets, in most scientific and medical contexts the terminology of the World Health Organization (2021) is adopted. WHO has developed the following clinical case definition of what it calls “post COVID-19 condition”:“Post COVID-19 condition occurs in individuals with a history of probable or confirmed SARS CoV-2 infection, usually 3 months from the onset of COVID-19 with symptoms and that last for at least 2 months and cannot be explained by an alternative diagnosis. Common symptoms include fatigue, shortness of breath, cognitive dysfunction but also others and generally have an impact on everyday functioning. Symptoms may be new onset following initial recovery from an acute COVID-19 episode or persist from the initial illness. Symptoms may also fluctuate or relapse over time.” For present purposes, two features of this definition are noteworthy. The first is that cognitive dysfunction is now a widely recognized feature of the post COVID-19 condition alongside physical symptoms. This was not always the case. Many adults who presented to medical professionals with COVID-related cognitive issues in the early months of the COVID-19 pandemic had their symptoms dismissed as depression, anxiety, and even post-traumatic stress disorder (PTSD). The second noteworthy feature concerns the timing of cognitive symptoms. Although there is no empirical evidence to support this claim, it appears likely that cognitive issues in Long COVID persist from the initial illness. Many people who participated in the author’s study withdrew from work and social activities during the acute stage of their infections. In the absence of the types of cognitive demands that these activities place on an individual, it may have appeared to many that their cognition was not affected by their illness. However, as physical symptoms lessened and people attempted to resume occupational and social activities, the extent of their cognitive difficulties became apparent for the first time. Several studies have examined the prevalence of cognitive dysfunction, or “brain fog”, in adults with Long COVID. In a meta-analysis of 19 studies involving 11,324 patients with post COVID-19 syndrome, Premraj et al. (2022) reported brain fog in 32% of patients. Additional cognitive symptoms in these patients included memory issues (27%) and attention disorder (22%). Davis et al. (2021) reported cognitive dysfunction in 88% of 3,762 participants with confirmed or suspected COVID-19 with illness lasting over 28 days. Cognitive dysfunction was one of three symptoms that were most commonly experienced after 6 months of illness (fatigue and post-exertional malaise were the other two symptoms). Children also develop Long COVID. Like adults, they can experience cognitive dysfunction. Lopez-Leon et al. (2022) conducted a systematic review and meta-analysis of 21 studies that examined 80,071 children and adolescents with COVID-19. The prevalence of Long COVID was 25.24%. Cognitive symptoms, which included reduced concentration, learning difficulties, confusion, and memory loss, had a prevalence of 6.27% in these children and adolescents. Although the term “brain fog” has descriptive benefits for people with Long COVID, it has limited scientific value when discussing cognitive difficulties in the post COVID-19 condition. This is because the term captures a rather loose group of cognitive and language problems, ranging from word-finding difficulty and issues with reading and writing to memory loss and concentration problems. To gain some insight into the nature and prevalence of these difficulties, the author conducted an online survey of 973 adults with Long COVID (Cummings, 2023b). These adults were asked to indicate if they experienced a range of cognitive-linguistic difficulties (see Table 1). The most commonly reported symptom was word-finding difficulty (93.1%). Of 11 language difficulties examined, nine occurred in over 50% of survey respondents. Clearly, there is a considerable burden of language and communication problems in people with Long COVID. What this burden looks like in terms of the responses of individual participants to tasks in the COVID-19 language study reported in Cummings (2023a) is examined in Section 3.Table 1 Language and communication problems in 973 adults with Long COVID. Table 1Language problem Frequency I struggle to find words 93.1% I forget what I wanted to say 90.9% I lose concentration easily when talking to others 89.6% I mix words up and produce incorrect words 72.4% I cannot recall what has been said earlier in conversation 65.4% I find reading difficult 61.7% I cannot recall what has been said in conversation after it has taken place 60.6% I find writing difficult 51.2% I veer off topic in conversation and cannot get back 50.8% I struggle to produce utterances and sentences 46.6% I struggle to understand what people are saying 38.2% 3 Language in adults with Long COVID The author’s study of language in adults with Long COVID revealed several cognitive-linguistic difficulties. Participants with Long COVID had significantly poorer performance than healthy participants in the study in the following areas: delayed and immediate verbal recall; informativeness of spoken discourse; letter fluency; and category fluency for animals (Cummings, 2023a). This is not the place to rehearse the findings of this study in detail. Instead, the focus of this paper is on exploring some of the individual responses of participants to the tasks used in the study. This will allow readers to understand the extent to which language is disrupted in Long COVID, an issue that is not properly appreciated if only language test scores are considered. Also, it will help readers to understand the cognitive processing difficulties that give rise to language and communication problems in Long COVID. This will prepare the ground for the classification of language problems in Long COVID as a cognitive-communication disorder in Section 4. 3.1 Discourse informativeness Adults with Long COVID in the study produced spoken discourse which was less informative than discourse produced by healthy participants in the study. This finding was consistent across all three discourse production tasks used in the study: the Cookie Theft picture description task from the Boston Diagnostic Aphasia Examination (Goodglass et al., 2001); Flowerpot Incident narration (storytelling based on a sequence of six, black-and-white line drawings); and Cinderella narration (telling the story of Cinderella after viewing pictures in a wordless picture book). This finding was not solely related to memory problems; reduced informativeness was also a feature of spoken discourse production in response to pictures that the speaker viewed throughout his or her performance of the task. It was argued in Cummings (2023a) that the reduced informativeness of adults with Long COVID on these tasks was related to higher-level discourse processes involved in the management of information. These processes make extensive use of cognitive skills, particularly executive functions such as planning and organization. Specifically, adults with Long COVID were able to produce grammatically well-formed, meaningful discourse which was nevertheless not particularly informative. Below, several ways in which adults with Long COVID produced spoken discourse with reduced informativeness are examined. Memory problems in adults with Long COVID led to reduced informativeness of the stories that participants produced during immediate and delayed verbal recall. Considerable information was omitted during these tasks, resulting in scores that fell well below the mean performance of healthy participants on the same tasks. Below, a 49-year-old woman called VT is recalling the 100-word Sam and Fred story that was used to examine immediate and delayed recall in the study. Prior to her COVID illness, VT worked part-time as a French teacher in a secondary school. She was tested at 211 days (7 months) after the onset of her COVID illness. At this stage, VT was still on sick leave from work; subsequently, she had to leave work on account of Long COVID. VT’s scores on both immediate recall (5/14) and delayed recall (3.5/14) placed her between 2 and 3 standard deviations below the mean scores of healthy participants (9.73/14 and 9.38/14, respectively). Her use of “Tam” is a simple mishearing of the author’s speech, possibly related to the author’s accent and/or the administration of the task online: Immediate recall “Tam and Fred are two farmers (2.86) they’re working some fields (1.51) something about the barn door and um (.) there’s a bad, bad weather storm coming and they’re very distressed and maybe somebody helps them” Delayed recall “two farmers I’m tempted to say Fred and Tam um (1.68) barn door open ah disastrous weather um can’t recollect if somebody helped them or not um very atmospheric scene setting ah struggling with the sequence of events I’m retaining two farmers worked very hard um some kind of unmitigated disaster can’t I think it was the weather-related um and that’s it” During immediate recall, VT omitted considerable information. She did not recall that crops were washed away, that sheep and cows escaped from the barn, and that the animals were returned to the barn by nightfall. There was also no mention of the fact that the farmers were brothers and that they had worked on the farm for thirty years. The same information is omitted on delayed recall. Additionally, VT is unsure on delayed recall if the farmers received assistance – she reported that they did receive help on immediate recall – and she omits a further point of information, namely, that the farmers were working the fields. There are also several remarks during delayed recall that allude to VT’s cognitive difficulties (e.g., can’t recollect, struggling with) as well as language that masks or conceals the omission of information during recall (viz., very atmospheric scene setting). VT’s lexical choices also contribute to the reduced informativeness of her responses on these recall tasks. The use of the indefinite pronouns something and somebody takes the place of more specific (and, hence, more informative) lexical forms such as the barn door blew open and the villagers helped the farmers. There is little information conveyed through the indefinite noun phrase some kind of. VT’s lexical choices compound her omission of information and reduce yet further the informativeness of her responses to these recall tasks. On these recall tasks at least, reduced informativeness seems to be a direct consequence of VT’s post-COVID memory problems. But reduced discourse informativeness in adults with Long COVID cannot always be so easily explained in terms of memory difficulties. On occasions when memory was not taxed during discourse production, reduced informativeness was also a feature of the language used by adults with Long COVID in the study. Below, a 36-year-old man called DB is describing the Cookie Theft picture from the Boston Diagnostic Aphasia Examination (Goodglass et al., 2001). DB contracted SARS-CoV-2 through his work as a hospital physiotherapist. He developed acute COVID-19 in March 2020 and returned to work after 10 days. DB worked full-time for 6 months. However, in September 2020 his fatigue became so severe that he had to cease working. At the time of testing, DB had been on sick leave for 2 months. DB obtained a score of 4/12 on this picture description task. This score was between 2 and 3 standard deviations below the mean score of 7.79/12 for healthy participants in the study: Cookie theft “I see (.) uh two (.) children that may look like a boy and a girl er (.) looking for cookies in a cookie jar in a top cupboard and a (.) woman who (.) may be their mother (.) erm washing dishes at the sink the sink is (.) overflowing uhm (.) and she’s in front of a window with the curtains open” DB’s description omits considerable information. He does not describe how the boy is standing on a stool and that the stool is about to tip over. DB does mention the girl but fails to describe any actions that she undertakes. He does not state, for example, that the girl has her arm outstretched to receive cookies from the boy and that she is gesturing to the boy to be quiet so that their mother does not become aware that they are stealing cookies from the cupboard. The mother is standing in a puddle of water. But this, too, is omitted from DB’s description. The reduced informativeness of DB’s description cannot be explained in terms of memory difficulties as the Cookie Theft picture is displayed throughout the task. A more plausible explanation is that DB failed to scan the scene adequately, reporting only the two most salient features of the picture – the boy is taking cookies and the woman is washing dishes. The cognitive processes that might explain DB’s behaviour include a failure to attend to all events in the picture. Also, DB may not have been able to achieve sufficient suppression or inhibition of his response to the two most salient features of the scene, so much so that these features came to dominate his description of the picture to the exclusion of other details. Attention and inhibition belong to a group of cognitive skills called executive functions. There is evidence of executive function impairments in adults with Long COVID (Miskowiak et al., 2021; Krishnan et al., 2022). Also, DB’s performance on letter fluency (a measure of executive function) was between 1 and 2 standard deviations below the mean score of healthy participants on this task in the study, suggesting that DB had some difficulties in the area of executive function. Although impaired executive function is a cognitively plausible explanation of DB’s picture description problems, it is difficult to say with certainty if it was the cause of the reduced informativeness of his discourse. The Flowerpot Incident narration created further challenges for adults with Long COVID. This task involves a sequence of six, black-and-white line drawings which the narrator must develop into a coherent story. Unlike the single scene in the Cookie Theft picture, this narrative task requires the speaker to integrate information by drawing inferences that connect events depicted in the individual pictures. This task was particularly challenging for a 49-year-old female participant (‘ST’) with Long COVID. Prior to her COVID illness, ST was a police inspector. She was tested at 519 days (17.3 months) after the onset of her COVID illness. ST obtained an informativeness score of 7/20 on this task. This score placed her between 2 and 3 standard deviations below the mean score of healthy participants (13.85/20) on the same task: Flowerpot incident “a plant falls on a man’s head he’s angry (1.94) and then goes inside (1.31) the house (5.25) then he’s knocking a door which I can’t work out if he’s in the house maybe a flat um (2.79) then the lady (.) or a lady ah gives his dog a bone and he kisses her hand (1.88) I think” ST’s omission of information was extensive. She omitted certain propositions in their entirety such as ‘The man and his dog are walking along the street’ and ‘A woman answered the door’. There is also partial omission of propositions, e.g., that the plant falls from a balcony. But most problematic of all – and the most significant cause of the reduced informativeness of ST’s narrative – is ST’s failure to embed the actions of the characters within a network of concepts that give coherence to the events in the story. We are not told the man’s purpose in going inside the building, viz., to remonstrate with the owner of the apartment from which the plant fell. We are also not told the consequence of the plant striking the man on the head (he developed a painful lump) or the reason why the woman gave the man’s dog a bone (to make amends for the injury he sustained). Concepts such as purpose, consequence, and reason create coherence between events in the story; their omission makes the characters’ actions appear disconnected and unmotivated. There is also some confusion on ST’s part about where the man is. This suggests that ST is unable to draw an inference to the effect that the door the man is knocking on is the door to the apartment from which the plant fell. Finally, ST uses a definite noun phrase to introduce the house into the story. This assumes that there has been some prior mention of the house when this is not the case (a listener may reasonably ask what house?). In this instance, it is the combination of omitted information, a lack of unifying concepts and inferences, and difficulty introducing new information that leads to the reduced informativeness of ST’s narrative. The most challenging discourse production task in the study was Cinderella narration. The cognitive demands of this task include the retention of a large amount of information in memory alongside the need to plan and organize many different events and characters across several episodes in the story. Given this level of cognitive complexity, it is not surprising that some of the most marked reductions in informativeness for the adults with Long COVID in the study occurred in this task. The narrative below was produced by a 42-year-old female participant (‘PB’) with Long COVID. PB also worked as a police officer prior to contracting SARS-CoV-2. She was tested at 456 days (15.2 months) after the onset of her COVID illness. PB achieved an informativeness score of 19.5/50 on this task. This score was between 2 and 3 standard deviations below the mean score (32.1/50) of healthy participants on the same task: Cinderella “Cinderella’s (1.77) mother dies and her father (1.74) marries (2.22) a lady with two (.) step dau [daughters] so two daughters (1.14) and (16.49) and (1.62) her father I think is the king (1.71) and his son is looking for a prince (2.00) princess sorry, sorry princess and she so they have (1.77) a (.) ball arranged and (4.04) Cinderella wants to go but thee stepsisters won’t let her go and she has to stay behind so thee fairy godmother (1.35) cast a spell and she had a beautiful dress and carriage (1.88) and went to the ball where she (4.09) danced with a prince and (1.15) at midnight she had to be home so she ran and to left one of her glass slippers behind and he found the slipper and wanted to find (2.30) who is the woman was that owned the slipper and (1.56) thee stepmother had a locked (.) her in her room but they managed to get her out and the prince (1.59) she tried the shoe on and it fit and a prince married her” Like other adults with Long COVID, PB omitted considerable information from her narrative (e.g., Cinderella was made to do all the household chores). Additionally, she produced some incorrect information such as when she stated that Cinderella’s father was the king. PB’s informativeness was further reduced by the omission of parts of propositions. We are not told that the ball was arranged to take place at the palace, that Cinderella’s dress and carriage came from her rags and a pumpkin, and that she ran away from the palace and towards her waiting carriage. PB also does not state why Cinderella had to be home at midnight (the spell would be broken at midnight). As well as these informational difficulties, PB did not use pronominal reference effectively. She occasionally used pronouns in the absence of a prior referent. This occurs towards the end of her narrative when she stated they managed to get her out; there is no clear referent of the pronoun they in the preceding discourse context (the referent in this case is the mice). On a further occasion, PB used the pronoun she to refer to Cinderella when the proximal referent is actually the fairy godmother; although on this occasion at least, the hearer’s prior knowledge of the story would have guided the correct assignment of a referent. PB also used the indefinite noun phrase a prince at the very end of her narrative. This form is more appropriate for the first mention of a character when the prince had in fact already been mentioned earlier in the story. This is further evidence of PB’s difficulties in managing information in discourse. In this section, we have examined the language of adults with Long COVID in a range of discourse contexts. For the most part, this language has been grammatically well-formed and meaningful. However, it has not been as informative as the language that is used by healthy participants on the same tasks. In trying to establish why this is the case, we have looked to cognitive processes such as executive functions to understand the informational difficulties of adults with Long COVID. In short, these adults retain knowledge of language form and meaning, but they are unable to leverage this knowledge to produce informative discourse on account of post-COVID cognitive difficulties. This results in well-formed and meaningful discourse that is nonetheless markedly under-informative. In the next section, we examine how COVID-related cognitive difficulties compromise language in a different context, namely, the access and retrieval of words from the mental lexicon during confrontation naming. 3.2 Word-finding difficulty The confrontation naming performance of adults with Long COVID studied by Cummings (2023a) did not differ from the naming performance of healthy participants in the study. However, this normal test performance belies considerable word-finding difficulty in these adults. Among the 973 adults who completed the online Long COVID survey, word-finding difficulty was the most commonly reported language symptom, with some 93% of respondents stating that they experienced this problem (Cummings, 2023b). The extent of these difficulties becomes apparent when the naming performance of individual participants in the study is examined. This reveals considerable cognitive inefficiency during naming. The following data is from a 49-year-old woman called TC. Before her COVID infection, TC was healthy and worked as a university lecturer. She was recorded at 337 days (11.2 months) after the onset of her infection. TC obtained a score of 18/20 on confrontation naming. This score was comparable to the mean performance of healthy participants on the same task (mean = 17.62; standard deviation = 2.08). TC’s responses to five pictures are shown in Table 2:Table 2 TC’s responses during confrontation naming. Table 2Target word TC’s response seahorse “it’s a one of those sea things urm urm urm urm they’re very small erm seahorse” doorknob “you open the door with it erm it’s uhm uhm eer I wanna say door handle but there’s a better word than that erm erm because you’d find it in a kitchen on a on a on a press erm er I know what it is it’s a doorknob” windmill “it’s a erm er (.) things go around with the wind it’s a windmill” thimble “you sew with it and you put it on your (.) finger no you don’t you put it on your thumb it is a thimble” lobster “the pinchy guys it’s a er it’s a lobster” To achieve naming of each of these pictures, TC undertook extensive circumlocution. She effectively talked around each target word as a means of eliciting its production. TC used circumlocution to name all 20 target words in the confrontation naming task. It was clear that post-COVID, TC’s lexical access and retrieval displayed cognitive inefficiencies, and that circumlocution was a compensatory strategy that she had adopted to facilitate her naming of pictures. NV is a 53-year-old woman. Prior to her COVID infection, NV was a healthy secondary school teacher. She was studied at 795 days (26.5 months) after the onset of her COVID illness. Although NV achieved a maximum score of 20/20 on confrontation naming, her naming performance exhibited considerable struggle. Pauses, dysfluency, and circumlocutions (e.g., a thing you use to do wood) were used extensively before the target word was produced. Table 3 shows how NV proceeded to name three items during the confrontation naming task:Table 3 NV’s responses during confrontation naming. Table 3Target word NV’s response chisel “siss, siss, screw not a screwdriver it’s um (1.39) a thing you use te, te, do wood um (5.97) ah, juh, juh, te is it tu, tu (1.08) chuh (1.47) chisel” artichoke “it’s a vegetable um (2.32) hum (4.71) it’s a lot of effort to eat it you have it with butter (1.21) ah (3.87) ah (3.34) ah, ah (.) ar, artichoke” cannon “ah (3.87) it’s a um (2.53) big gun um (5.87) xxx (unintelligible) cannon” These cases demonstrate how test scores alone are not always a reliable indicator of an individual’s linguistic performance. Like other participants with Long COVID in the study, TC and NV had essentially normal naming based on the scores that they obtained. However, as the above data demonstrates, naming was a cognitively inefficient process for both participants, a finding that is consistent with self-reported word-finding difficulties in speakers with Long COVID. 3.3 Verbal fluency In the study, verbal fluency was examined in adults with Long COVID through the use of letter fluency and category fluency tasks. During letter fluency, participants were asked to produce as many words as possible beginning with the letters F-A-S in 60 s. They were instructed to avoid proper names like Francis and Sally and morphological variants of the same word (e.g., fishes, fished, fishing). Letter fluency is viewed as a measure of executive function, although this has been challenged in some studies (e.g., Whiteside et al. (2016) reported that in their study letter fluency loaded exclusively onto a language factor and not executive function). Participants undertook two category fluency tasks in the study. They were asked to produce the names of as many animals and vegetables as possible in 60 s. Category fluency examines lexical generation; participants must access a specific semantic field in their mental lexicon and produce as many exemplars of the field as possible. On all these verbal fluency tasks with the exception of category fluency for vegetables, adults with Long COVID had significantly poorer performance than healthy adults in the study. Below, some of the difficulties that adults with Long COVID had on these tasks are examined. Explanations of these difficulties in terms of cognitive processing problems are considered. For many adults with Long COVID and poor letter fluency, reduced speed of processing appeared to compromise the production of words beginning with F-A-S. This was evident both in the production of a small number of words in 60 s, and in the presence of lengthy timed pauses between each word that was produced. Below are the ‘A′ words that were produced by a 49-year-old woman (‘KS’) with Long COVID. KS worked as a doctor in general practice before contracting SARS-CoV-2. She was tested 449 days (15 months) post COVID onset and had been on sick leave for some time at this point. KS’s combined letter fluency score was 29 words. This score placed her between 1 and 2 standard deviations below the mean score (48 words) of healthy participants on the same task.“aim (1.38) able (3.11) ambivalent (2.40) and (7.32) ask (1.81) answer (5.61) aunty (14.75) artichoke” KS produced only 8 words beginning with ‘A′ in 60 s (one word every 7.5 s on average). This compares to a mean score of 11.9 words beginning with ‘A′ in a study of verbal fluency in 1,300 healthy participants (Tombaugh et al., 1999). Even more noteworthy, however, are the lengthy timed pauses throughout KS’s response. It took over 7 s for KS to produce ask and over 14 s to produce artichoke. KS was able to produce words according to the search criteria of the task (i.e., words beginning with ‘A′ that are not proper names, etc.). She was also able to avoid repetition of words. However, her speed of processing was reduced, with each word taking longer to access and produce than is typically the case in healthy participants. A quite different difficulty is on display in the following ‘S′ words produced by a 58-year-old woman (‘OL’) with Long COVID. OL was tested at 533 days (17.7 months) post COVID onset. She was not working on account of her COVID illness. After excluding the proper name Skype and derivational forms of the same word (e.g., sink-sinkhole) – these words do not satisfy the criteria of the task – OL produced a total of 16 words beginning with ‘S′:“sugar, spice, saturated, smelly, socks, um sink, sinkhole, um southern, south (.) ah sausages (.) ah sensations (1.30) ah silly um (1.79) saucer (.) ah socks (2.78) stilts, stiletto heels my favourite um (1.33) sugar (3.21) spice (4.38) Skype (1.29) um screwdriver ah (2.51) secateurs, scribe (.) ah um (2.11) saus [sausage] no I said that one” The number of ‘S′ words produced by OL is similar to the mean score of healthy participants on the same letter fluency task (Tombaugh et al., 1999). What is noteworthy about OL’s response is not the overall number of words she produces but the number of times she repeats words. This occurs on four occasions: socks-sugar-spice-sausage. Only on the last of these words does OL realize that she has already produced the word sausage and she initiates a correction. OL’s repetition of words suggests some difficulty with the monitoring of her verbal output. As well as reduced self-monitoring, OL also has difficulty inhibiting words (e.g., Skype) that do not satisfy the criteria for the task. On this occasion, OL has normal speed of processing; her difficulties lie in the monitoring of her verbal output and the suppression or inhibition of words that do not satisfy the task criteria. Category fluency for animals was another area of relatively poor verbal fluency performance for adults with Long COVID in the study. This task assesses an individual’s ability to access the mental lexicon, locate the target category of animals, and produce as many instances of the category as possible in 60 s. Animals of all types (mammals, reptiles, birds, etc.) can be included. Examinees can produce general words (e.g., dog) and specific words (e.g., spaniel). However, in terms of scoring, one mark is awarded when these terms are produced consecutively (e.g., dog-spaniel) and when animal names represent different developmental stages (e.g., lamb-sheep). Like letter fluency, the respondent must avoid repetition of words. When healthy individuals undertake category fluency tasks, they typically employ two strategies called clustering and switching. Clustering involves the generation of consecutive words belonging to the same subcategory; in switching, respondents generate words consecutively that belong to different subcategories. Clustering facilitates the generation of words by allowing respondents to undertake a search of a single subcategory and, hence, a relatively small lexical space. Because switching involves conducting a search across different subcategories, it can be a less efficient strategy for word generation. Some adults with Long COVID made effective use of clustering during the category fluency tasks in the study. Unsurprisingly, these adults obtained scores for animal naming that are consistent with the performance of healthy participants on the same task. A 56-year-old woman (‘LL’) with Long COVID produced the following words during animal naming. LL was tested 250 days (8.3 months) after the onset of her COVID illness. At the time of testing, LL had not returned to her pre-COVID employment as a primary school teacher on account of her COVID illness:“donkey, cow, sheep (1.21) chicken, hen (1.29) um horse (1.21) um fly (1.37) a dog, cat (1.14) ah mouse, rat (1.93) ah um (1.62) what other animals are there (1.96) hens think I’ve already said hens (3.95) ah giraffe (1.10) lions, tiger um rhino, orangutan, monkey (1.51) um parrots (5.25) penguin, whale, fish” LL produced a total of 20 animal names (a hen is a female chicken and so chicken and hen was given one score). This score is similar to mean scores of 20.1 words (Tombaugh et al., 1999) and 18.4 words (Acevedo et al., 2000) obtained by healthy participants the same age as LL in large normative studies. Clustering was used to good effect by LL and facilitated her generation of animal names. There are two large clusters in LL’s response. The first cluster can be broadly categorized as farm animals, at least in a UK context: donkey-cow-sheep-chicken-hen-horse. The second cluster is native African animals and contains the instances giraffe-lions-tiger-rhino-orangutan-monkey. There are also two smaller clusters, one inside and one outside a larger cluster. The cluster primates occurs inside the native African animals cluster and contains the two instances orangutan and monkey. The other cluster is rodents and contains the instances mouse and rat. LL used clustering effectively to generate as many animal names as possible in 60 s. However, many other adults with Long COVID in the study appeared unable to use a clustering strategy to any significant extent during animal naming. This resulted in category fluency scores that were well below the mean scores of healthy participants in large normative studies. A 58-year-old woman (‘OL’) with Long COVID produced the following words during the animal category fluency test in the study. OL was tested at 533 days (17.7 months) following the onset of her COVID illness. OL worked as a project manager, and she owned her own jewelry business before contracting SARS-CoV-2. However, she had not been able to resume work on account of her ongoing COVID illness:“elephant, ape, ostrich um (1.01) dog, cat, mouse um (2.05) penguin um (1.13) tiger (1.28) ah (5.14) monkey um (3.92) xxx (unintelligible) dolphin um (3.04) sparrow (8.70) I can’t really think (1.82) this is not good um (2.22) rat (8.83) uh bad” OL produced a total of 12 animal names. This score placed her between 1 and 2 standard deviations below the mean score of healthy participants in studies by Tombaugh et al. (1999) and Acevedo et al. (2000). There is little clustering on display. OL produces three bird names – ostrich, penguin, and sparrow – but they are interspersed by other words. The two rodent names mouse and rat, which were produced consecutively by LL, are separated by five other animal names in OL’s response. Words in the native African animals category – elephant, ape, tiger, and monkey – are also produced in two groups of two words. OL makes extensive use of switching. In the sequence monkey-dolphin-sparrow-rat, OL switches between the subcategories of primate, marine animal, bird, and rodent. That OL’s lack of clustering and use of switching was a less cognitively efficient lexical generation strategy is not only evident in the total number of words that OL produced, but also in the duration of pauses between words. LL and OL produced a similar number of pauses during their naming (12 and 10 pauses, respectively). But whereas these pauses had an average duration of 1.96 s during LL’s animal naming, they had an average duration of 3.03 s during OL’s generation of animal names. OL’s use of switching as her dominant lexical generation strategy was cognitively less efficient than LL’s use of clustering. The additional cognitive processing required to switch between lexical subcategories during OL’s generation of animal names resulted in a lower number of words with more time required to produce each word. 4 Management of cognitive-linguistic difficulties in Long COVID The cognitive-linguistic difficulties examined in Section 3 raise a number of important questions. The first question concerns how these difficulties should be assessed and diagnosed by health professionals. These professionals include speech-language pathologists and neuropsychologists, although this section will focus on the former group of clinicians. The part of this question that relates to diagnosis depends on how we conceive of cognitive-linguistic difficulties in Long COVID. Should these difficulties sit among aphasias or are they cognitive-communication disorders? Or do these difficulties form a novel category of communication disorder that has the potential to expand the current nosology of language disorder? The second question concerns the impact of cognitive-linguistic difficulties on the daily functioning of adults with Long COVID. We will see below that the impact is significant, with COVID-related language difficulties compromising employment, social relationships, and family roles, among other domains. The third question is related to the second question in that it addresses how the impact of COVID-related language difficulties on employment may be best addressed by occupational health teams. Speech-language pathologists are ideally positioned to make employers and occupational health professionals aware of how cognitive-linguistic difficulties in Long COVID can serve as a barrier to work reintegration in much the same way that physical symptoms of COVID-19 (e.g., respiratory difficulties) can compromise a return to work. This takes the professional role of speech-language pathologists beyond the assessment and diagnosis of cognitive-linguistic difficulties in adults with Long COVID to include important education and advocacy work on behalf of these clients. When the cognitive-linguistic difficulties of adults with Long COVID were examined in Section 3, a point that was only briefly addressed but which is particularly relevant to the diagnosis of these difficulties is that the structural language skills of these adults are in relatively good condition. For the most part, adults with Long COVID use well-formed, meaningful language that contains an appropriate range of vocabulary. The language of these adults displays few of the grammatical, lexical, and semantic difficulties of adults with aphasia, even as it is produced with considerable hesitancy and non-fluency related to problems with memory and planning. These adults also have relatively good auditory verbal comprehension. They are able to follow complex language that is used to convey task instructions, although in many cases these instructions have to be repeated to facilitate language comprehension. In short, the linguistic knowledge of adults with Long COVID remains intact, even as they find it difficult on account of cognitive processing problems to leverage this knowledge to produce discourse and perform other language tasks. The language problems of these adults are cognitive in nature, with knowledge of linguistic rules (if that is how we want to conceive of linguistic knowledge) still largely intact. Given the cognitive origin of language difficulties in adults with Long COVID, we need to conceive of these difficulties, less in terms of a primary language impairment like aphasia and more in terms of a cognitive-communication disorder. (It should be noted that aphasia can occur in COVID-19, particularly in severe disease; see Priftis (2023) for a review.) Cognitive-communication disorders are familiar to speech-language pathologists, particularly those clinicians who assess and treat adults with conditions like traumatic brain injury (TBI), right-hemisphere damage (RHD), and neurodegenerative disease (e.g., Alzheimer’s disease). Adults with these conditions have been noted to communicate less adequately than their structural language skills in phonology, syntax and semantics would predict. Sentence-level syntax is relatively intact in these adults even as they struggle to tell a story or give an informative description of a scene. These discourse-level difficulties arise from cognitive deficits in areas such as executive function (see Cummings (2017) for further discussion). Within a nosology of language disorder, there are grounds for including the communication difficulties of adults with Long COVID in the class of cognitive-communication disorders, albeit the physiological basis of these difficulties is less well understood than in other cognitive-communication disorders. In order to diagnose cognitive-communication difficulties in Long COVID, speech-language pathologists must know how best to assess these difficulties. One of the most important lessons to emerge from clients with TBI and RHD who have cognitive-communication disorders is that these disorders are not revealed on standardized language assessments of the type used to assess adults with aphasia. The predominantly word- and sentence-level formats of these assessments are not sensitive to the effects of cognitive dysfunction on language. This was confirmed in Cummings (2023a); there was no significant difference in test scores between adults with Long COVID and healthy participants on a sentence generation task and a confrontation naming task used in the study. This was despite the fact that the picture naming of these adults displayed considerable cognitive inefficiencies (see Section 3.2). The tasks that were most sensitive to the cognitive-linguistic difficulties of adults with Long COVID in Cummings (2023a) were discourse production tasks. As the cognitive demand of these tasks increased, from picture description (Cookie Theft) at the simplest to production of a fictional narrative (Cinderella) at the most complex, the performance of adults with Long COVID showed a stepwise decrease. Inefficiencies in cognitive skills (e.g., executive function) that are used to plan a complex narrative like the Cinderella story are exposed by discourse production tasks but remain hidden from view in test formats that are based on individual words and sentences. Speech-language pathologists are encouraged to view discourse production tasks as an indispensable tool in the assessment of Long COVID adults with cognitive-communication difficulties in much the same way that they recognize the benefits of these assessments for clients with TBI and RHD. It is not an exaggeration to say that Long COVID has a devastating impact on the lives of those who are affected by the condition. The impact of this debilitating illness and its accompanying cognitive-communication difficulties is felt across all life domains, but particularly on employment, social relationships, and family role. In Cummings (2023b), the extent of that impact was quantified in an online survey of 973 adults with Long COVID and self-reported “brain fog”. In terms of employment, the impact was most significant. Some 67.9% of respondents were in full-time employment prior to developing COVID-19. This dropped to only 24.6% after the onset of their COVID illness. There was an equally significant increase in the number of people who were not working due to disability; this rose from just 2.4% before developing COVID-19 to 32.5% after the onset of COVID illness. Most remarkable of all is that only 22.8% of respondents agreed with the statement ‘I meet the communication needs of my job or college’. The impact on employment is even more keenly felt given that the average age of respondents to the survey was just 47.4 years (Cummings, 2023b). This is a much younger population of adults than those typically assessed and treated by speech-language pathologists (e.g., adults with strokes and neurodegenerative diseases). Long COVID had brought the careers of many of these middle-aged adults to an abrupt end. Many others had substantially reduced the number of hours that they worked or had changed the types of roles that they could perform. With communication skills significantly compromised by Long COVID, many respondents to the survey reported a loss of friendships and other social relationships as a consequence (Cummings, 2023b). Respondents often avoided social interactions because they felt embarrassed (54.9%) or frustrated (83.2%) by their poor communication skills post COVID-19. Some 65.8% of respondents reported that they had less desire to communicate with others. Many social relationships that had been forged with work colleagues deteriorated when respondents were on long periods of sick leave. Adults with Long COVID were often either too tired to engage in the communication that was required to maintain these relationships or felt that their health and communication problems were dismissed by work colleagues. Statements such as “You look so well” and “I often struggle to find words also” were a source of considerable frustration for adults with Long COVID. There was a feeling among these adults that they had been abandoned by friends and family members because they could no longer serve their needs. One 44-year-old woman with Long COVID remarked: “People do not have the same contact with me. I rarely hear from many of my friends or certain members of the family. Maybe it’s because with my communication problems I no longer fulfil their needs.” COVID-related cognitive-communication difficulties also had the effect of altering the roles of many individuals within the family. Most respondents to the online survey and participants in the study were women in their forties who had school-age children (Cummings, 2023a, Cummings, 2023b). Prior to Long COVID, they had assisted their children with homework and regularly attended parent-teacher evenings. The pandemic placed additional educational burdens on parents who became responsible for their children’s home schooling. The cognitive-communication difficulties of adults with Long COVID left them struggling to help their children with homework, perform home schooling, and discuss their children’s academic progress with teachers. The daily conversations that parents have with their children about events at school were also compromised. One 45-year-old woman with Long COVID reported: “[I] find I have ‘tuned out’ of conversations, e.g., in the car when [my] child [is] telling me about the day at school.” Once a context for conversation about events that had taken place during the day, family meals were a struggle for parents with Long COVID. The digestion of food prompted such a marked deterioration in the communication abilities of one woman with Long COVID that she was unable to talk to her family after an evening meal:“Of all the things Long COVID has taken away from me, it’s the ability to speak properly after an evening meal. An evening meal has always been a family time for us to share what’s happened in the day. The energy involved in digestion affects my cognitive ability and language. Therefore, I am unable to speak after we’ve eaten food.” While speech-language pathologists await the research that is needed to devise effective interventions for adults with Long COVID, they can undertake a vital education and advocacy role in relation to these clients. Many adults with Long COVID and cognitive-communication difficulties try unsuccessfully to receive ill health retirement from work. Others undertake phased returns to work that are unrealistic given the type of difficulties they are experiencing. More often than not, such phased returns end in failure. Many adults with Long COVID could resume some form of economic productivity if employers undertook appropriate adjustments to address their difficulties. However, for this occur, employers and occupational health teams must have a proper understanding of cognitive-communication difficulties in Long COVID. Also, they must be aware that Long COVID does not only involve physical symptoms such as respiratory difficulties and post-exertional malaise but also difficulties with language and communication that may outlive physical problems and prevent a return to work. Speech-language pathologists have the knowledge and skills that are needed to articulate the communication needs of adults with Long COVID to employers and occupational health teams. They also have a professional duty to advocate for these clients who, on account of their communication challenges, cannot present their individual circumstances clearly and fully in speech and writing to employers. Although it is now three years since the start of the COVID-19 pandemic, it is still the case that many adults with Long COVID are trying to manage debilitating physical and cognitive symptoms with little in the way of effective healthcare to support them. Long COVID clinics are struggling to keep pace with demand, at least in the UK. Many adults with Long COVID have had no access to these clinics or have received an initial assessment with little or no follow-up. Adults with Long COVID frequently report feeling abandoned and dismissed by health providers. They have been forced to manage their own recovery and navigate their way through complex care pathways that are not always responsive to their needs. As one participant in the study by Cummings (2023a) remarked, “it really feels like it’s down to me to make a diagnosis and request treatment”. Notwithstanding these various challenges, there are also some positive developments in this Long COVID journey. At a time when health services were under considerable strain because of the pandemic, patients with Long COVID began organizing themselves into online groups. These groups are now well established and developed and are providing essential support and information for people with Long COVID. The Long Covid Nurses & Midwives UK group is a case in point (https://teamlcnmuk.wixsite.com/lcnmuk). The bodies that represent health professionals in the UK are assessing the type of clinical services that might best address the needs of people with Long COVID. At the time of writing, the author is contributing to the development of clinical guidelines for Long COVID by the Royal College of Speech and Language Therapists (RCSLT) in the UK. There is still much work to be done. The Royal College of Speech and Language Therapists (2022) reported that only 13.8% of therapists who provide speech and language therapy to people with Long COVID work within a specially commissioned/funded or dedicated Long COVID service (note: this is 13.8% of 111 responses within a wider group of 676 respondents to an online survey). While professional bodies continue to advocate for better clinical services for people with Long COVID, they are also providing patients and clinicians with vital resources on Long COVID (e.g., RCSLT (2023) Long COVID podcast). These efforts must continue apace if we are to assist those who already have Long COVID, and if we are to have services in place for those who will develop this condition following future waves of SARS-CoV-2 infection. 5 Summary This article has examined cognitive-linguistic difficulties in adults who do not make a full recovery from SARS-CoV-2 infection. So-called Long COVID or post COVID-19 condition affects a significant number of adults and children who contract SARS-CoV-2, with many continuing to experience debilitating physical and cognitive symptoms several months and even years after the onset of infection. The cognitive-linguistic difficulties of adults with Long COVID in three areas were examined: discourse informativeness; confrontation naming; and verbal fluency. Language difficulties in each of these domains were related to cognitive inefficiencies following COVID-19 illness. To reflect the cognitive basis of these difficulties, they were categorised as a cognitive-communication disorder, a diagnostic label that is already familiar to speech-language pathologists. These difficulties have a profound impact on daily functioning, compromising employment, social relationships, and family role, among other domains. The article examined the role of speech-language pathology within the management of adults with Long COVID and cognitive-communication difficulties. It also emphasized the need for specialist clinical communication services for this new and growing population of clients. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data Availability The authors do not have permission to share data. ==== Refs References Acevedo A. Loewenstein D.A. Barker W.W. Harwood D.G. Luis C. Bravo M. Hurwitz D.A. Aguero H. Greenfield L. Duara R. Category fluency test: Normative data for English- and Spanish-speaking elderly Journal of the International Neuropsychological Society 6 7 2000 760 769 10.1017/s1355617700677032 11105466 Cummings L. Cognitive aspects of pragmatic disorders Cummings L. Research in clinical pragmatics. Perspectives in Pragmatics, Philosophy & Psychology Vol. 11 2017 Springer International Publishing AG 587 616 Doi: 10.1007/978-3-319-47489-2_22 https://link.springer.com/chapter/10.1007/978-3-319-47489-2_22 Cummings L. Cognitive-linguistic difficulties in adults with Long COVID Cummings L. COVID-19 and speech-language pathology 2023 Routledge 72 95 10.4324/9781003257318-5 https://tandfbis.s3.us-west-2.amazonaws.com/rt-files/docs/Open+Access+Chapters/9781003257318_10.4324_9781003257318-5.pdf https://library.oapen.org/handle/20.500.12657/58162 Cummings L. Communication-related quality of life in adults with Long COVID Cummings L. COVID-19 and speech-language pathology 2023 Routledge 96 129 Doi: 10.4324/9781003257318-6. https://library.oapen.org/handle/20.500.12657/58163 Davis H.E. Assaf G.S. McCorkell L. Wei H. Low R.J. Re'em Y. Redfield S. Austin J.P. Akrami A. Characterizing long COVID in an international cohort: 7 months of symptoms and their impact EClinicalMedicine 38 2021 101019 10.1016/j.eclinm.2021.101019 Goodglass, H., Kaplan, E., & Barresi, B. (2001). Boston diagnostic aphasia examination. Third Edition. Lippincott Williams & Wilkins. Krishnan K. Miller A.K. Reiter K. Bonner-Jackson A. Neurocognitive profiles in patients with persisting cognitive symptoms associated with COVID-19 Archives of Clinical Neuropsychology 37 4 2022 729 737 10.1093/arclin/acac004 35136912 Lopez-Leon S. Wegman-Ostrosky T. Ayuzo Del Valle N.C. Perelman C. Sepulveda R. Rebolledo P.A. Cuapio A. Villapol S. Long-COVID in children and adolescents: A systematic review and meta-analyses Scientific Reports 12 1 2022 9950 10.1038/s41598-022-13495-5 35739136 Miskowiak K.W. Johnsen S. Sattler S.M. Nielsen S. Kunalan K. Rungby J. Lapperre T. Porsberg C.M. Cognitive impairments four months after COVID-19 hospital discharge: Pattern, severity and association with illness variables European Neuropsychopharmacology 46 2021 39 48 10.1016/j.euroneuro.2021.03.019 33823427 Premraj L. Kannapadi N.V. Briggs J. Seal S.M. Battaglini D. Fanning J. Suen J. Robba C. Fraser J. Cho S.M. Mid and long-term neurological and neuropsychiatric manifestations of post-COVID-19 syndrome: A meta-analysis Journal of the Neurological Sciences 434 2022 120162 10.1016/j.jns.2022.120162 Priftis K. Neurolinguistic deficits and other cognitive disorders in adults with severe COVID-19 infection Cummings L. COVID-19 and speech-language pathology 2023 Routledge 53 71 10.4324/9781003257318-4 Royal College of Speech and Language Therapists Understanding the need for and provision of speech and language therapy services for individuals with post-COVID syndrome in the UK 2022 London: Author. https://www.rcslt.org/wp-content/uploads/2022/01/Post-COVID-syndrome-report-RCSLT-January-2022.pdf Tombaugh T.N. Kozakb J. Reesc L. Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming Archives of Clinical Neuropsychology 14 2 1999 167 177 https://academic.oup.com/acn/article/14/2/167/1787 14590600 Whiteside D.M. Kealey T. Semla M. Luu H. Rice L. Basso M.R. Roper B. Verbal fluency: Language or executive function measure? Applied Neuropsychology Adult 23 1 2016 29 34 10.1080/23279095.2015.1004574 26111011
PMC010xxxxxx/PMC10287098.txt	==== Front Lancet Infect Dis Lancet Infect Dis The Lancet. Infectious Diseases 1473-3099 1474-4457 Elsevier Ltd. S1473-3099(23)00418-8 10.1016/S1473-3099(23)00418-8 Media Watch A tale of two pandemics Warren Sylvia 22 6 2023 22 6 2023 Curlee Lynn The other pandemic2023Charlesbridge PublishingUS978-1623543501 176£15·16© 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcIn 2020, according to WHO, over 1·53 million people died in the COVID-19 pandemic; the uncertainty, fear, and fervent hope for a vaccine or cure are now almost a memory. In the same year, 680 000 people died from AIDS-related illnesses, and 37·7 million people were living with HIV. With the advent of pre-exposure prophylaxis and developments in highly active antiretroviral therapy, a diagnosis of HIV is no longer considered a death sentence, and for many who were born in the USA after the turn of the millennium, the AIDS pandemic may feel like a piece of history, akin to the 1918 flu pandemic. However, for HIV and AIDS there is “still no vaccine, still no cure.” There are major differences between SARS-CoV-2 and HIV: the former is transmissible by air, the latter by body fluids or blood, yet in The Other Pandemic: An AIDS Memoir, artist and children's author Lynn Curlee draws out some of the parallels between navigating life when the shadow of viral infection is cast over everyday life. Curlee's book is principally a deeply personal memoir, a raw and evocative howl of grief at the senselessness of the deaths of many of his lovers and personal friends, and the decimation of the gay community in the face of political and medical ambivalence towards what was often seen as a ‘degenerate’ part of society. As Curlee says, “When my list of the dead reached forty-one, I stopped keeping count. It seemed pointless”. Little of the memoir portion of his book touches on parallels with the COVID-19 pandemic, but as a social document it is important. It is richly illustrated with photographs of his acquaintances, and his artist's eye for the visual complements the narrative urgency with which he describes both the joy and possibility of gay life, and the fear that descended upon every man he knew. He works with his intended young adult audience, trusting them with the ability to draw parallels without making them explicit, talking with a frankness about sex and drug use that does not feel infantilising. His discussion about the difficulties around palliative care and negotiating end-of-life care in the hinterlands before the death-rattle might well give comfort to an audience who have also lived through a pandemic where they were unable to grieve with others. Curlee documents a period in history that will be unrecognisable to young people in the USA today; rent in Manhattan, NY at US$112 a month being one of the elements that differ markedly from today's reality. His descriptions of the underground gay community before AIDS are exhilarating and set up a sense of deep freedom and happiness that makes the ultimate devastation of AIDS even more poignant. Curlee does recognise in the epilogue that access to healthcare, including financial barriers, has a serious effect on the ability to live with HIV—he also fails to mention the labour rights work of Cleve Johnson when discussing the AIDS quilt, one of the largest piece of community art in history. It is in the late acknowledgement of these issues that the potential for this book to teach queer history to young adults feels like a missed opportunity; it is in no doubt that gay and bisexual men who have sex with men were more likely to contract HIV when the cause of transmission was still unknown. However, the author glosses over several key points that are vital to his argument. The first is his lack of acknowledgment to the lesbian community for being caretakers of those dying from AIDS in the US. He mentions wards being, “staffed by volunteers”, yet neglects to say that when the medical establishment was enforcing separate wards in hospitals it was largely lesbian societies such as the Blood Sisters who stepped in (neither lesbian or bisexual appear in the index of this book). Curlee also mentions in his first chapter, “Civil Rights, Black Power, Women's Lib, and the idea of Gay Liberation were all related. It was about equality.” He mentions the gay community's anger at price-gouging life-saving medication but falls shy of criticising the system as it stands, and never says that his Black ex-boyfriends had to negotiate a homophobic world whilst also facing racism. The other pandemic seems as though it was written in two halves. As a memoir, it is perfectly pitched to teenagers, queer or otherwise, who should know about the ways that politics, medicine, and prejudice intersect. The science is delivered in a neutral way, with a quiet critique of institutional biases that can physically affect the way people exist within the world, if lacking some of the bite that is required. In the end, Curlee's book is an exploration of grief, and a remarkably honest one. Bringing this back to politics, the reviewer expects this will be banned in several US states. It should not be, and it is an indictment that children are being kept from accessing books that discuss recent lived history.
PMC010xxxxxx/PMC10287099.txt	==== Front Lancet Infect Dis Lancet Infect Dis The Lancet. Infectious Diseases 1473-3099 1474-4457 Elsevier Ltd. S1473-3099(23)00417-6 10.1016/S1473-3099(23)00417-6 Media Watch Sharing the road during the COVID-19 pandemic Dunsmuir Henrietta 22 6 2023 22 6 2023 Accardi Millicent Borges Quarantine highway2022FlowerSong Press978-1953447357 106£12·00© 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcWritten in the early period of the COVID-19 pandemic, Millicent Borges Accardi's fourth poetry collection Quarantine highway examines her experience of isolation. The COVID-19 pandemic opened up a new reality, as life as we knew it was radically transformed. A ‘normal’ task such as going to a food shop became a minefield of ‘Clorox’ wipes and hand gel, and surgical masks left operating theatres and entered our daily wardrobe. Writing during a time of radical changes in our daily life, Borges Accardi uses poetry to try to navigate the disorientating and uncertain early months of the pandemic. Her poems focus on how we sought to escape from reality, whilst longing for the pre-pandemic days. In this anthology, Borges Accardi examines her own quarantine experience. She prefaces her collection with a description of quarantine in 14th century Venice written by authors from the US Centers for Disease Control. When quarantine was first implemented in the Italian city, ships were required to anchor for forty days and those aboard were completely isolated. The contemporary poems that follow consider a markedly different quarantine experience, with an emphasis on the quarantine's unknown future, and the presence of technology in homes. Initially, Borges Accardi reminds us of those days when we were consigned inside: ‘television glued / as news rolled by’ (p 5). There are yearnings for pre-pandemic days (Yes it's difficult) in which life was ‘easy’ (p 23). Whilst trapped inside, childhood memories are brought to the surface and Borges Accardi vividly recalls her experiences of previous infections. Graphic details of headlice and chickenpox are recounted in I told my friend to rub her lice against my hair. Similarly, in I made up a story for myself once, the speaker retreats inward and admits ‘where you/ close your eyes and send you/ back to the safe place’ (p 72), demonstrating how our imaginations became a distraction from ongoing anxieties. Fiction, too, became a solace for us at this time. Our ‘comfort’ lay in ‘movies’ (p 24) as we took refuge in the imaginary, and readers are reminded of Netflix shows such as Tiger King and Love is Blind which peaked in popularity during this period (p 19). In What we Call Time, we even live vicariously through fiction as we ‘watch old/ movies with longing’ (p 27). As Borges Accardi illustrates, we lived our lives through screens. Yet, as a result of this over-consumption, our idling worlds degenerated as we did ‘the nonsense/ we knew we shouldn’t:/ over-drinking, board games,/ chanting curses at each other’ (p 5). In her final poems, and with the benefit of hindsight, Borges Accardi recalls how vapid these ‘board games’ then seemed; they lacked the ‘richness of reality,/ embodied with a generous sadness’ (p 85). The struggle to find a sense of purpose in this time is repeatedly underlined. As this idling drifts on, Borges Accardi makes us painfully aware of our clumsy, self-conscious attempts to do the right thing. In Let your eyes slide over the estuary, her frustrations scream ‘I find myself spraying Lysol on the paper/ as if I mean, as if the fuck we don’t/ know’ (p 24). In the early pandemic days, we performed these rituals in an attempt to keep ourselves safe. But their effectiveness was unclear. The uncertainty of this time is reflected in Borges Accardi's stilted and jolted phrasing. By the end of the collection, the hopelessness of how to navigate this new normal is made clear: ‘does anyone really know what is coming next’ (p 79). Many of the poems in this collection were written as a result of discussions Borges Accardi had over Zoom with fellow writers. Others were written as part of Juan Morales’ writing challenge for CantoMundo fellows. The effect of this creates a resounding number of voices which filter in, influence, and inspired Borges Accardi's work. Many poems are ‘inspired’ or ‘from a line by’ fellow poets—authors such as Elizabeth Acevedo, Pablo Neruda and Javier Zamora. Borges Accardi describes how poetry ‘drew us close, reunited our spirits and held our souls safely within our own and each other's isolation’ (p 91). The book is a credit to a synthesis of a learning experience where the poets shaped and re-shaped each other's interpretations of a world dominated by COVID-19. This array of themes experienced by us all is underscored by the self-evident need for connection, companionship, and as Accardi found, a kind of love. In We still are not breathing, she tells us that ‘Love is not a currency, neither is it an assignment’ (p 14). Instead, it is a visceral need, like ‘the drinking of water for thirst’ (p 14). However we may recall the peak of quarantine, this collection of poems reminds us of how fiction and friendships helped so many of us through uncertain times. Quarantine highway gathers almost seventy poems, penned by Borges Accardi and influenced and shaped by a group of writers who drove down this highway together.
PMC010xxxxxx/PMC10287100.txt	==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(23)01125-X 10.1016/S0140-6736(23)01125-X Correspondence Loneliness in the time of COVID-19: an alarming rise Priya Giri Shakshi a Dubey Anubhuti a a Department of Psychology, Deen Dayal Upadhyaya Gorakhpur University, Gorakhpur, Uttar Pradesh 273004, India 22 6 2023 24-30 June 2023 22 6 2023 401 10394 21072108 © 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcIt has been 3 years since WHO declared COVID-19 a pandemic. Globally, WHO reports 764 474 387 confirmed cases of COVID-19 and more than 6·9 million deaths at the time of writing this letter. The pandemic has brought about a plethora of health challenges, both physical and mental. Among these challenges, the sense of isolation and loneliness experienced by many people is particularly noteworthy. Consider a scenario where a person was away from home for work purposes but became stranded due to the pandemic. Unfortunately, during this period, the person's mother passed away due to COVID-19. Regrettably, pandemic-related travel restrictions prevented the individual from returning home to pay their last respects to their deceased mother. The person had no one to confide in personally. Months later, when the restrictions eased, the person returned to work but struggled to connect with their colleagues and chose to isolate themself from others. The experience left the person feeling lonely and isolated. The emotional toll of this traumatic event is unimaginable.1 This person is not the only one who feels the weight of loneliness, as many adults have been hit hard by this feeling during the COVID-19 pandemic.2 Unfortunately, these feelings are likely to continue and experts are concerned about the potential for loneliness to become a chronic issue. If left unaddressed, chronic loneliness can lead to various severe mental and physical health conditions.3 It is crucial for the medical community to recognise the impact of loneliness on individuals and take steps to address it. Loneliness is a painful, subjective experience characterised by a feeling of insufficient or unsatisfactory desired social connections. Loneliness can result in unhealthy behaviours, such as poor sleeping patterns, lack of exercise, and unhealthy dietary habits, which can contribute to an increased risk of premature mortality by 26% if not appropriately dealt with.4 Moreover, loneliness is believed to be associated with the adverse effects of chronic stress on the body, including inflammation, weakened immune function, and an elevated risk of cardiovascular disease.4 Health-care providers can incorporate screening for social isolation and loneliness into routine assessments and develop care plans that address these issues to avoid any further mental health-related concerns due to any novel illnesses. ??? For the WHO COVID-19 Dashboard see https://covid19.who.int/ We declare no competing interests. ==== Refs References 1 Dsouza DD Quadros S Hyderabadwala ZJ Mamun MA Aggregated COVID-19 suicide incidences in India: fear of COVID-19 infection is the prominent causative factor Psychiatry Res 290 2020 113145 2 Ernst M Niederer D Werner AM Loneliness before and during the COVID-19 pandemic: a systematic review with meta-analysis Am Psychol 77 2022 660 677 35533109 3 Keller FM Derksen C Kötting L Dahmen A Lippke S Distress, loneliness, and mental health during the COVID-19 pandemic: test of the extension of the Evolutionary Theory of Loneliness Appl Psychol Health Well-Being 15 2023 24 48 35266309 4 Cacioppo JT Cacioppo S Capitanio JP Cole SW The neuroendocrinology of social isolation Annu Rev Psychol 66 2015 733 767 25148851
PMC010xxxxxx/PMC10287101.txt	==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(23)01121-2 10.1016/S0140-6736(23)01121-2 Correspondence Disrespectful language about patients with long COVID Pelosi Anthony a a Priory Hospital Glasgow, Glasgow GH1 3DW, UK 22 6 2023 24-30 June 2023 22 6 2023 401 10394 21082108 © 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcIn the March Editorial, The Lancet stated that “Because of long COVID's [also known as post-COVID-19 condition] diverse symptomatology, reliance on self-reported symptoms, and a lack of diagnostic tests and consensus definition, many patients struggle to obtain a definitive diagnosis. As a result, long COVID is often easily dismissed as a psychosomatic condition. Given what we now know about the effects of long COVID and its biological basis, it must be taken seriously.”1 Although this was surely not the intention, this statement is deeply offensive. Why would any clinician dismiss patients whose bodily symptoms are caused or worsened by psychological factors but do their best to help those with similarly severe symptoms that are a direct result of infective, inflammatory, or metabolic abnormalities? It is difficult to think of any serious illness in which symptoms and disability could not be exacerbated by psychological factors or social adversity (or both). The Editorial expressed concern about the “excruciatingly slow” progress in providing proper clinical services for people with prolonged ill health following COVID-19 infection. There is no chance of decent care for this heterogeneous group of patients if psychosocial factors are not taken seriously when considering differential diagnoses, clinical formulation, and appropriately individualised management plans. The Lancet's careless discussion of psychosomatic conditions mirrors disrespectful language that can be encountered in medical, surgical, and even psychiatric clinics throughout the UK. The Editors could make amends by using their writing skills to try to improve doctor–patient communication about biological, psychological, and social influences on long COVID and essentially every other serious human illness. I declare no competing interests. ==== Refs Reference 1 The Lancet Long COVID: 3 years in Lancet 401 2023 795 36906338
PMC010xxxxxx/PMC10287169.txt	==== Front JHEP Rep JHEP Rep JHEP Reports 2589-5559 Elsevier S2589-5559(23)00154-4 10.1016/S2589-5559(23)00154-4 100823 Article Contents 22 6 2023 7 2023 22 6 2023 5 7 100823100823 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc
PMC010xxxxxx/PMC10287176.txt	==== Front Sci Total Environ Sci Total Environ The Science of the Total Environment 0048-9697 1879-1026 The Authors. Published by Elsevier B.V. S0048-9697(23)03686-0 10.1016/j.scitotenv.2023.165063 165063 Article Life-Cycle Assessment of the thermal and catalytic pyrolysis over sepiolite of face masks Garcia-Garcia Guillermo a Martín-Lara Mª. Ángeles b⁎ Calero Mónica b⁎ Ortega Francisco b Blázquez Gabriel b a Department of Agrifood Chain Economics, Institute of Agricultural and Fisheries Research and Training (IFAPA), Centre ‘Camino de Purchil’, 18080 Granada, Spain b Department of Chemical Engineering, Faculty of Sciences, University of Granada, 18071 Granada, Spain ⁎ Corresponding authors. 22 6 2023 22 6 2023 1650636 3 2023 16 6 2023 20 6 2023 © 2023 The Authors. Published by Elsevier B.V. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Since the start of the global COVID-19 pandemic, extensive quantities of face masks have been used and discarded. Most of these masks end up in landfills, causing a high environmental impact and no benefits. However, there are alternative ways to deal with this waste in a more sustainable way. For example, valorisation of face masks through pyrolysis has received special attention because it offers efficient application to produce a liquid oil that can be used as a diesel substitute and a solid char that can be used as an activated carbon substitute after activation. In this context, this study applies the Life-Cycle Assessment methodology to quantify and analyse the environmental impacts of different treatment scenarios based on the pyrolysis of surgical masks and FFP2 masks. It also compares their environmental performance with the conventional practice of landfilling. The scenarios studied include both thermal and catalytic pyrolysis by using sepiolite, a low-cost material abundant in Spain. Data on the pyrolysis process were obtained from laboratory experiments. It was found that the use of the produced oil as a diesel substitute very significantly reduces the environmental impact in all pyrolysis scenarios. Consequently, the pyrolysis of face masks can reduce the environmental impact caused by the treatment of this waste material. Furthermore, the thermal pyrolysis performs environmentally better than the catalytic pyrolysis. In all scenarios, freshwater ecotoxicity and marine ecotoxicity are the environmental impact categories that cause the highest environmental impact overall. Graphical abstract Unlabelled Image Keywords Pyrolysis Face masks Life-Cycle Assessment LCA Environmental impact Sustainability Waste management Editor: Deyi Hou ==== Body pmcData availability No data was used for the research described in the article.
PMC010xxxxxx/PMC10287177.txt	==== Front Sci Total Environ Sci Total Environ The Science of the Total Environment 0048-9697 1879-1026 Published by Elsevier B.V. S0048-9697(23)03718-X 10.1016/j.scitotenv.2023.165095 165095 Article A comparative analysis of the partitioning behaviour of SARS-CoV-2 RNA in liquid and solid fractions of wastewater Breadner Patrick R. a Dhiyebi Hadi A. a Fattahi Azar a Srikanthan Nivetha a Hayat Samina a Aucoin Marc G. b Boegel Scott J. b Bragg Leslie M. a Craig Paul M. a Xie Yuwei cd Giesy John P. de Servos Mark R. a⁎ a Department of Biology, University of Waterloo, 200 University Ave W, Waterloo, Ontario N2L 3G1, Canada b Department of Chemical Engineering, University of Waterloo, 200 University Ave W, Waterloo, Ontario N2L 3G1, Canada c Key Laboratory of Pesticide Environmental Assessment and Pollution Control, Nanjing Institute of Environmental Sciences, Ministry of Ecology and Environment, Nanjing 210042, China d Toxicology Centre, University of Saskatchewan, 44 Campus Dr, Saskatoon, Saskatchewan S7N 5B3, Canada e Department of Environmental Science, Baylor University, One Bear Place, Waco, TX 76798, USA ⁎ Corresponding author. 23 6 2023 23 6 2023 1650959 4 2023 30 5 2023 21 6 2023 © 2023 Published by Elsevier B.V. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. As fragments of SARS-CoV-2 RNA can be quantified and measured temporally in wastewater, surveillance of concentrations of SARS-CoV-2 in wastewater has become a vital resource for tracking the spread of COVID-19 in and among communities. However, the absence of standardized methods has affected the interpretation of data for public health efforts. In particular, analyzing either the liquid or solid fraction has implications for the interpretation of how viral RNA is quantified. Characterizing how SARS-CoV-2 or its RNA fragments partition in wastewater is a central part of understanding fate and behaviour in wastewater. In this study, partitioning of SARS-CoV-2 was investigated by use of centrifugation with varied durations of spin and centrifugal force, polyethylene glycol (PEG) precipitation followed by centrifugation, and ultrafiltration of wastewater. Partitioning of the endogenous pepper mild mottled virus (PMMoV), used to normalize the SARS-CoV-2 signal for fecal load in trend analysis, was also examined. Additionally, two surrogates for coronavirus, human coronavirus 229E and murine hepatitis virus, were analyzed as process controls. Even though SARS-CoV-2 has an affinity for solids, the total RNA copies of SARS-CoV-2 per wastewater sample, after centrifugation (12,000 g, 1.5 h, no brake), were partitioned evenly between the liquid and solid fractions. Centrifugation at greater speeds for longer durations resulted in a shift in partitioning for all viruses toward the solid fraction except for PMMoV, which remained mostly in the liquid fraction. The surrogates more closely reflected the partitioning of SARS-CoV-2 under high centrifugation speed and duration while PMMoV did not. Interestingly, ultrafiltration devices were inconsistent in estimating RNA copies in wastewater, which can influence the interpretation of partitioning. Developing a better understanding of the fate of SARS-CoV-2 in wastewater and creating a foundation of best practices is the key to supporting the current pandemic response and preparing for future potential infectious diseases. Graphical abstract Unlabelled Image Keywords COVID-19 SARS-CoV-2 Wastewater-based surveillance Viral partitioning PEG precipitation Ultrafiltration device Editor: Kyle Bibby ==== Body pmcData availability The raw data that supports the conclusions of this article has been deposited in the Federated Research Data Repository (FRDR) and is available at: doi:10.20383/102.0747. Additional questions regarding the data can be made to MS.
PMC010xxxxxx/PMC10287178.txt	==== Front Health Policy Health Policy Health Policy (Amsterdam, Netherlands) 0168-8510 1872-6054 The Authors. Published by Elsevier B.V. S0168-8510(23)00145-8 10.1016/j.healthpol.2023.104860 104860 Article Illustrating the impact of commercial determinants of health on the global COVID-19 pandemic: Thematic analysis of 16 country case studies Freeman Toby 1⁎ Baum Fran 1 Musolino Connie 1 Flavel Joanne 1 McKee Martin 2 Chi Chunhuei 3 Giugliani Camila 4 Falcão Matheus Zuliane 5 De Ceukelaire Wim 6 Howden-Chapman Philippa 7 Huong Nguyen Thanh 8 Serag Hani 9 Kim Sun 10 Dardet Carlos Alvarez 11 Gesesew Hailay Abrha 12 London Leslie 13 Popay Jennie 14 Paremoer Lauren 15 Tangcharoensathien Viroj 16 Sundararaman T 17 Nandi Sulakshana 17 Villar Eugenio 18 1 Stretton Health Equity, University of Adelaide, Adelaide, SA, 5005 AUSTRALIA. 2 London School of Hygiene & Tropical Medicine, London WC1H 9SH, UNITED KINGDOM 3 Center for Global Health, Oregon State University, Corvallis, OR 7331, U.S.A 4 Universidade Federal do Rio Grande do Sul, Rua Ramiro Barcellos, 2400 CEP 90035-003, Porto Alegre, BRAZIL 5 University of São Paulo, Brazil, Av. Dr. Arnaldo, 715 - 211 - Cerqueira César, São Paulo - SP, 01246-904, BRAZIL 6 Médecine pour le Peuple, Brussels 1000, BELGIUM 7 Department of Public Health, University of Otago, Wellington, 6042, NEW ZEALAND 8 Faculty of Social Science and Behavior, Hanoi University of Public Health, 1A Duc Thang Road, Duc Thang Ward, North Tu Liem District, Hanoi, VIETNAM 9 University of Texas Medical Branch (UTMB), 301 University Blvd., Galveston, Texas, 77555, USA. 10 People's Health Institute, 36 Sadang-ro 13-gil, Dongjak-gu, Seoul 07004, SOUTH KOREA. 11 CIBERESP, Center for Research in Epidemiology and Public Health, University of Alicante, 03560 SPAIN 12 Research Centre for Public Health, Equity and Human Flourishing, Torrens University Australia, Adelaide, SA, 5000 AUSTRALIA & College of Health Sciences, Mekelle University, Mekelle, 231 ETHIOPIA 13 School of Public Health, University of Cape Town, SOUTH AFRICA. 14 Division of Health Research, Faculty of Health & Medicine, Lancaster University, Bailrigg, Lancaster LA1 4YW, UNITED KINGDOM 15 Political Studies, University of Cape Town, Cape Town, SOUTH AFRICA 16 International Health Policy Programme, Ministry of Public Health, Nonthaburi, THAILAND 17 People's Health Movement, Delhi, INDIA 18 Universidad Peruana Cayetano Heredio, San Martín de Porres 15102, PERU ⁎ Corresponding author, Ph: + 61 406 320 334, Stretton Health Equity, University of Adelaide, Adelaide, SA, 5005 AUSTRALIA. 23 6 2023 23 6 2023 1048601 12 2022 12 6 2023 15 6 2023 © 2023 The Authors. Published by Elsevier B.V. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Previous research on commercial determinants of health has primarily focused on their impact on non-communicable diseases. However, they also impact on infectious diseases and on the broader preconditions for health. We describe, through case studies in 16 countries, how commercial determinants of health were visible during the COVID-19 pandemic, and how they may have influenced national responses and health outcomes. We use a comparative qualitative case study design in selected low- middle- and high-income countries that performed differently in COVID-19 health outcomes, and for which we had country experts to lead local analysis. We created a data collection framework and developed detailed case studies, including extensive grey and peer-reviewed literature. Themes were identified and explored using iterative rapid literature reviews. We found evidence of the influence of commercial determinants of health in the spread of COVID-19. This occurred through working conditions that exacerbated spread, including precarious, low-paid employment, use of migrant workers, procurement practices that limited the availability of protective goods and services such as personal protective equipment, and commercial actors lobbying against public health measures. Commercial determinants also influenced health outcomes by influencing vaccine availability and the health system response to COVID-19. Our findings contribute to determining the appropriate role of governments in governing for health, wellbeing, and equity, and regulating and addressing negative commercial determinants of health. Keywords Social determinants of health health equity COVID-19 privatisation comparative study ==== Body pmcBackground The ‘commercial determinants of health’ framework describes the ways that for-profit actors, through their activities and in pursuit of their interests, contribute to shaping the health of populations and the social and economic structures that they act within [1], [2], [3], [4]. While commercial actors can make a positive contribution to health, for example by providing decent employment (and thus income), paying taxes, and creating and developing health-promoting products, such as medicines, they can also harm health. It is, however, difficult for those responsible for protecting and promoting public health to counter these harms because of the large power asymmetry, with some transnational corporations having wealth that far exceeds many nations [5], [6], [7], [8], [9]. Particularly in countries with neoliberal approaches to public policy, there is great concern about the undue influence of commercial actors. Many have extensive access to decision-making processes and can thus shape public policy to serve their interests, in some cases capturing what were once core functions of the state [10], [11], [12]. These once private functions that have been privatised in some countries include care for people in some of the most disadvantaged circumstances in society, such as children removed from their families [13], and asylum seekers and refugees [14]. Thus, following Diderichsen, Dahlgren, and Whitehead [15], and Gilmore and colleagues [4], we include the privatisation of previously state-provided services as a critical commercial determinant of health, intertwined as it is with the political interests driving the privatisation, and the potential for detrimental impacts on the health and wellbeing of the population, and on equity [16]. There is now a large body of literature investigating commercial determinants of non-communicable diseases, including creation of unhealthy food environments, in particular affecting children [17], the marketing and sale of unhealthy commodities such as alcohol, tobacco, gambling products, and fossil fuels, and lobbying against greater public health regulations and increased tax on these commodities. Although this literature has paid less attention to infectious diseases [1,18,19] some issues have been investigated. These include the neglect of antimicrobials by pharmaceutical companies and the role of producers of food, iron ore, palm oil, and biofuel crops in the spread of Ebola in West Africa [20,21]. The COVID-19 pandemic provides an opportunity to revisit these issues. Commercial determinants of health may have been a factor in the emergence of COVID-19. There have long been warnings of the risks of catastrophic pandemics caused by intensive agriculture, attendant deforestation and industrial animal farming [22]. Zhang [22] traces the potential origin of COVID-19 in the Huanan Seafood Wholesale Market to market-oriented reforms in China and industrialisation in food production, drawing parallels with the emergence of SARS in 2003. Thoradeniya and Jayasinghe [23] include corporate interests and neoliberal capitalism as two drivers of the context in which COVID-19 emerged, exacerbating deforestation and climate change, disrupting biodiversity, and accelerating industrialised animal factory farms. The United Nations argues that “deforestation, and habitat encroachment are primary pathways of transmission for emerging infectious diseases, including COVID-19” [24] and commercial interests are major drivers of deforestation and habitat encroachment [25]. The role of political and social determinants of health in the magnitude of COVID-19 cases and deaths is now clear [26]. Countries whose governments were able to act rapidly and decisively, which had created socially and racially inclusive societies, and which had established strong social welfare safety nets, achieved better outcomes [26,27]. Evidence from many countries highlighted how social inequities on dimensions such as gender, ethnicity, citizenship/migration status, housing, wealth, employment and working conditions, and racism and discrimination explained patterns of disease [26,[28], [29], [30]]. In contrast, relatively little has been written on how commercial determinants of health shaped countries’ pandemic discourses, responses and outcomes. Dall'Alba and Rocha [31] provide a brief overview of the situation in Brazil, which saw corporate lobbying to promote ineffective therapies for COVID-19 and to undermine the quality of Brazilian primary health care, focusing instead on hospitals and medicines that offer greater potential for profits with significant cases, particularly in the private health insurance sector. These commercial interests were supported by Federal government strategy [32,33]. Maani and colleagues [34] have articulated some of the commercial determinants of health impacting COVID-19 in the United States. Some were positive, through the development of vaccines (albeit with considerable public funding and while making large profits [35,36]) and contributing to public health responses. Other impacts were negative, notably, commercial activities that have contributed to severe COVID-19 illness risk factors (such as cardiovascular disease, diabetes, and obesity) through the sale and marketing of unhealthy products, privatisation of government services reducing public capacity, and commercial capture of institutions such as the Centers for Disease Control and Prevention [34]. Gilmore and colleagues [4] provide some examples of how commercial determinants affected COVID-19 responses, including political lobbying by unhealthy commodity companies, and pharmaceutical companies’ use of intellectual property rights which increased the cost of vaccines and so restricted vaccine access. Barlow and colleagues have also examined how the pandemic impacted on international trade and, thus, on commercial actors, and their responses to it [37] and van Schalkwyk and colleagues have outlined the ways in which these actors may exploit the pandemic and associated recovery [38]. However, no articles were found that provide an overarching consideration of the ways in which different commercial determinants of health impacted COVID-19 outcomes. This paper aims to fill this gap through consideration of 16 country case studies. Our research question was ‘what evidence is there for the different ways commercial determinants of health impacted on countries’ COVID-19 outcomes? Methods This research followed a comparative qualitative case study design [39], analysing 16 case study countries (see Table 1 ). This design was chosen to allow detailed understanding of the complexity of factors that would have interacted to determine COVID-19 outcomes in each country, taking into account its specific historical, geographic, and political context, characteristics over which researchers had no control [39]. We also saw the empirical findings for the case studies as an opportunity to elaborate theory [40]. We know generally about the ways commercial determinants of health may positively or negatively influence health outcomes [5,7,18]; through these case studies we sought to provide detailed understanding of the different ways commercial determinants affected COVID-19 outcomes.Table 1 Case study countries’ COVID-19 outcomes (deaths, cases, vaccine coverage) in 2021 and 2022. Table 1 Income Cumulative Excess deaths per 100,000 January 2021 Cumulative Excess deaths per 100,000 January 2022 Cumulative reported deaths per million January 2021 Cumulative reported deaths per million January 2022 Cumulative cases per million January 2021 Cumulative cases per million January 2022 Vaccine coverage July 2021 Vaccine coverage Jan 2022 Australia High 3.73 (3.72-3.73) 15.55 (15.54-15.55) 35.25 148.71 1117.49 100060.66 14.99% 2 dose 32.36% 1 dose 78.37% 2 dose 84.08% 1 dose Ethiopia Low 49.83 (31.59-76.52) 174.24 (110.44-267.54) 17.76 62.24 1167.75 3946.16 N/A1 N/A1 New Zealand High 0.74 (0.73-0.74) 1.60 (1.6-1.6) 4.88 10.34 449.45 3242.1 14.36% 2 dose 22.43% 1 dose 76.59% 2 dose 81.40% 1 dose Nigeria Low 40.95 (28.01-55.93) 78.95 (54-107.83) 7.50 14.83 620.82 1197.64 0.67% 2 dose 1.21% 1 dose 2.58% 2 dose 7.02% 1 dose Taiwan High 0.08 (0.08-0.08) 3.7 (3.7-3.7) 0.34 35.67 38.18 787.51 1.60% 2 dose 32.31% 1 dose 73.11% 2 dose 80.02% 1 dose South Korea High 2.77 (2.68-3.59) 13.12 (12.7-17.02) 27.77 131.99 1530.22 16841.22 14.23% 2 dose 37.83% 1 dose 85.86% 2 dose 87.12% 1 dose Thailand Upper Middle 0.17 (0.14-0.21) 44.7 (35.06-56.29) 1.10 317.15 268.5 34995.49 5.51% 2 dose 19.50% 1 dose 69.64% 2 dose 74.89% 1 dose Vietnam Lower Middle 0.14 (0.09-0.19) 75.53 (49.15-110.04) 0.36 384.82 18.51 23181.77 0.60% 2 dose 5.44% 1 dose 75.57% 2 dose 80.55% 1 dose Belgium High 199.19 (186.16-235.42) 272.28 (254.47-321.76) 1813.22 2497.52 61049.92 271247.28 59.00% 2 dose 69.57% 1 dose 76.35% 2 dose 78.72% 1 dose Brazil Upper Middle 123.62 (111.6-144.92) 339.45 (308.28-396.04) 1049.79 2932.75 43031.79 118992.1 19.21% 2 dose 48.82% 1 dose 70.06% 2 dose 79.59% 1 dose India Lower Middle 85.12 (66.9-105.82) 254.87 (199.77-316.79) 110.8 356.13 7720.35 29761.18 7.30% 2 dose 25.82% 1 dose 50.95% 2 dose 67.68% 1 dose Peru Upper Middle 459.80 (346.87-607.81) 900.83 (679.57-1190.81) 3111.86 6160.33 34120.47 96656.55 15.62% 2 dose 24.40% 1 dose 68.82% 2 dose 77.24% 1 dose South Africa Upper Middle 226.62 (175.49-297.77) 487.39 (377.34-640.26) 735.55 1583.78 24212.4 60045.01 4.72% 2 dose 9.86% 1 dose 27.69% 2 dose 32.66% 1 dose Spain High 230.18 (204.34-264.16) 344.82 (305.85-394.5) 1247.59 1994.32 58682.35 213096.76 57.62% 2 dose 68.14% 1 dose 81.92% 2 dose 87.53% 1 dose United Kingdom (UK) High 184.61 (183.4-186.94) 268.94 (266.98-272.63) 1557.61 2285.45 56017.78 254483.27 55.87% 2 dose 68.58% 1 dose 71.02% 2 dose 76.77% 1 dose United States of America (US) High 173.65 (154.18-200.02 342.50 (303.62-394.47) 1346.91 2673.81 79115.81 225564.83 51.14% 2 dose 57.83% 1 dose 64.22% 2 dose 75.53% 1 dose Sources: Global Burden of Disease Study. COVID-19 projections: Cumulative deaths. Institute for Health Metrics and Evaluation; 2022 [29 March 2022]; Available from: https://covid19.healthdata.org/global?view=cumulative-deaths&tab=trend; Our World in Data. Cumulative confirmed deaths per million; 2022 [29 March 2022] Available from: https://ourworldindata.org/covid-deaths; Our World in Data. Cumulative confirmed cases per million; 2022 [29 March 2022] Available from: https://ourworldindata.org/covid-cases; Our World in Data. Vaccine coverage; 2022 [29 March 2022] Available from: https://ourworldindata.org/covid-vaccinations 1 Our World in Data COVID vaccinations data not available for Ethiopia. COVAX data (https://covidvax.live/location/eth) estimates that as of 31 July 2021, 2,217,097 vaccine doses had been administered, enough for one dose for 1.88% of the Ethiopian population. As of 31 January 2002, 10,975,026 vaccine doses had been administered, enough for one dose for 9.31% of the population. In selecting countries, we drew from those in the top and bottom quartiles in terms of cumulative excess deaths per 100,000. We used cumulative excess deaths attributable to COVID-19 from the Global Burden of Disease [41] compiled as of July 2021 and reviewed with updated data in December 2021 (resulting in one change to selected countries). We sought a spread of low-, middle-, and high-income countries. This is important as Mialon [42] notes that most literature on the commercial determinants of health is from high income countries. Additional criteria were pragmatic: data availability and having a country expert in the research team's network who we could invite to contribute to the analysis for their country. In most cases, these experts were academic or policy researchers, who were residents and/or citizens of the case study country. Firstly, the research team created a data collection template for each country informed by insights from the literature on COVID-19 outcomes (see Supplementary material). The template included questions on private and public health sector performance, system weaknesses revealed during the pandemic, considerations of equity, political leadership, and important contextual factors. Secondly, the country experts and research team then collaborated to create a detailed case study, including extensive grey and peer-reviewed literature from each country, to find available evidence corresponding to each section of the template. The literature examined included: academic articles and books; government reports and websites; reports by non-government and international institutions; and media such as online newspaper articles. Having country experts allowed inclusion of local literature that was not in English. Non peer-reviewed literature was included because of the rapidly changing nature of the pandemic, and to understand local political and civil society viewpoints usually not included in academic literature. In addition, our data collection framework focused on issues such as political, civil society, and equity considerations because these were less researched in the peer review literature. The need to respond to each country's context and salient issues meant an overly structured approach would not be useful. Instead, a broad data collection template was used, and the approach and goals of the data collection were discussed with each country lead. This led to the production of rich and deeply knowledgeable case studies of each country. Drafts were developed iteratively with feedback from the central research team. The case studies covered the period from the start of the pandemic (January 2020) through to March 2022, when they were completed. Thirdly, from these case studies, the research team identified commercial determinants of health as a clear, central concern in understanding experiences and outcomes of the COVID-19 pandemic. Our analysis of commercial determinants was guided by Freudenberg et al.’s [18] commercial determinants of health framework. This framework highlights the need to look at both the structures (including the political and economic system, social stratification, organisation of structures and relations, governance, and norms) that shape the influence, power, and practices of commercial actors (including market-oriented practices such as supply chain management, product design, and pricing, and policy- and political-oriented practices such as lobbying and philanthropy), and how commercial determinants affect health. Freudenberg and colleagues [18] identify market-oriented practices, such as the promotion of unhealthy commodities, tax contributions by corporations, and policy- and political-oriented practices, such as lobbying governments. Wherever there was evidence of commercial market-oriented and policy- and political-oriented practices, or the social structures that shaped commercial activities, these data were coded for inclusion by the lead author, who led the analysis of commercial determinants themes, identifying themes that grouped findings across case studies. These themes were further workshopped and reviewed by co-authors, with alternative explanations discussed and additional considerations added. Lastly, we conducted iterative rapid literature searches (of Web of Science and Google Scholar, in June-August 2022) pertaining to COVID-19 to ascertain the extent of broader evidence for our commercial determinant themes in the literature. Results Below we highlight the high-level themes we identified from the comparative analysis of the country accounts, structured according to Freudenberg and colleague's framework [18]. Structural influences on commercial determinants of health Political and economic system As indicated by the framework, it was essential to interrogate the distinct historical and political economy of each country. For example, Nigeria had inherited a colonial legacy [43] in which “In essence, the Nigerian people and their land were imagined not as people with rights to exist and function as a community or even nations. They were imagined as corporate money making entities whose bodies were enslaved and lands plundered” [44]. This was seen as framing the evidence related to privatisation and corporate power in the Nigerian case study. The role of the private sector in Nigeria's response is exemplified by the government and the Nigeria Centre for Disease Control partnering through the Coalition Against Covid-19 with Nestle, KPMG, PWC, and several banks [45]. The South African case study documented widespread corruption under President Zuma [46], which set the stage for failures of procurement of COVID-19 personal protective equipment (PPE), including the assassination of a health department whistle-blower [47]. The heavily pro-business stance of leaders in countries such as the US, UK, and Australia also shaped government COVID-19 responses. The fraught political status of Taiwan, contested by its much larger neighbour China, was also central to understanding the challenges it faced in obtaining vaccines [48]. Thirty years of underinvestment in public systems for health, education, and social protection was noted as creating vulnerabilities in Peru that meant the effects of COVID-19 were severe, particularly for people experiencing disadvantage and marginalisation [49,50]. In contrast, the Vietnam case study noted that the one party communist government allowed a strong, co-ordinated, immediate response to the COVID-19 pandemic that emphasised population health over competing economic interests [51,52]. Stratification The most prominent theme relating to stratification we identified was how working conditions driven by commercial interests affected COVID-19 spread and risk. Across the globe, precarious and lower paid essential workers faced greater infection risk [53] as well as being more susceptible to the economic fallout of job losses [54]. Some workplaces were found to be especially problematic. For example, in South Korea, outbreaks in a call centre and a warehouse highlighted poor working environments, with poor ventilation, poor disinfection measures, and no capacity for social distancing [55,56]. Informal employment brings particular vulnerabilities [57]. Our Vietnamese case study noted widespread informal employment, where workers do not have access to paid leave, or recourse to claim government support. Undocumented workers in the US and in Brazil faced similar challenges [58,59]. Foreign born workers have also been found to have higher infection risk, identified as a critical issue in the case studies from Taiwan, Thailand, and South Korea [58,[60], [61], [62]]. Migrant workers are more likely to have precarious and/or informal work, low skilled work that is not amenable to remote working, and care work that puts in them at high occupational risk of contracting COVID-19, as well as potentially poorer access to health care [60,61]. Overcrowded accommodation for migrant workers was reported in India, and across Europe [58]. In India, the stringent, abruptly imposed lockdown measures stranded millions of migrant workers far from family support [63]. While corporations employing these workers were compensated in many ways for their losses, the tremendous loss in earnings and the sufferings and death of the migrant workers received very meagre or no compensation. In all these cases, shortcomings can be seen as at least partly driven by the private sector being driven to maximise profits and reduce costs, especially for labour where short-term savings are easiest. In this way, the private sector has externalised costs of their working practices to society and to governments, in terms of COVID-19 infections, and government responsibility to support the private sector during the pandemic. Commercial actors that had long argued that the role of the state should be minimal turned to it when their interests were threatened. In contrast, the South African Department of Labour and Employment was proactive in attempting to regulate and protect working populations [64], demonstrating how a state-led response could use public funds to address unemployment and adversely impacted working conditions. Organisation/Governance The key theme relating to organisation and governance was how the procurement of vaccines and other COVID-19 response goods and services were governed and organised. Vaccination procurement Much has been written about global COVID-19 vaccine inequities [65,66]. Several of our case study countries experienced difficulties in procuring COVID-19 Vaccines. In Nigeria, the 4 million vaccines received through the COVID-19 Vaccines Global Access (COVAX) scheme as of July 2021 covered less than 2% of the population [67]. Globally, COVAX failed to meet half its 2021 target of 2 billion doses [36,68]. COVAX has been criticised for avoiding sharing power amongst members, unduly depending on private philanthropy such as the Bill and Melinda Gates Foundation, socialising corporate risk at a time when pharmaceutical companies are posting large profits, and for allowing richer countries to gain more beneficial terms while poorer countries remained reliant on aid [69]. Oxfam International has noted that Pfizer, BioNTech and Moderna were making $1,000 profit every second towards the end of 2021 while the world's poorest countries remained largely unvaccinated [70]. In the US case study, it was argued that even though Pfizer and Moderna vaccines – the two most commonly used vaccines globally [71] - were developed by the private sector, the discovery, development, testing, and production of COVID-19 vaccines had been largely paid for using public money [35,36]. Wouters and colleagues [72] have tabulated how much public and non-profit funding the leading vaccine candidates have had. Despite this, most patents for COVID-19 vaccines are owned by private companies. Pfizer and Moderna were estimated to have made US$41 billion profit (above production costs) on vaccines sold to governments as of July 2021 [73]. The cost of production of two doses of COVID-19 vaccines is estimated as low as US$2.40 per person [for two doses, 73], yet the US government purchased Pfizer vaccines at $39 per two doses, and Moderna vaccines at $30 per two doses [35]. An attempt to allow countries to manufacture their own vaccines, using a Trade-Related Aspects of Intellectual Property Rights (TRIPS) waiver, stalled for almost two years because of failure to garner enough countries support at the World Trade Organization, where it was blocked by the EU, UK, and Switzerland [36,68]. The extent of the pharmaceutical industry's lobbying and threats to these countries has been documented [74]. A weaker version of the initial proposal eventually passed in July 2022 despite heavy opposition from the pharmaceutical industry [75]. The speed, lack of transparency, and process of vaccine procurement was also heavily criticised in several case studies, including South Korea, Peru, Brazil and Australia, where the slow response of the government led to the term ‘vaccine strollout’ [76]. For some countries, this was exacerbated by very powerful pharmaceutical company interests. In South Africa, vaccines were largely procured from Pfizer and Johnson and Johnson, for USD$10 per dose [77]. Once again we saw the privatisation of profit and socialisation of risk [78] as both involved transfer of risk from manufacturers to governments. The Pfizer deal was subject to indemnity and no-fault compensation requirements [77], and the Johnson & Johnson deal was subject to a government compensation fund that would indemnify Johnson & Johnson for any vaccine-related injuries [79], and the company receiving a letter from the government endorsing the local investment the company had made in Aspen Pharmaceuticals [80].In Brazil, Federal Government hesitancy delayed vaccination rollout, however a decades-old public immunization program and state-owned pharmaceutical laboratories were key to catching up with other countries [81]. . In Peru, there was a scandal involving almost 500 elites, including the President, Health Minister and a congressman receiving early access to a COVID-19 vaccine from a university conducting clinical trials [82], highlighting further potential for corruption. The U.S. government's vaccine donation to South Korea after the Korea-U.S. summit in May 2021 is hard to explain other that in the context of the long-standing political and economic alliance between the two countries, given the U.S. government's previous stance to prioritise vaccine donation to neighbouring countries such as Canada and Mexico, and the other QUAD countries (Australia, India, and Japan). During the summit, South Korean semiconductor (Samsung), battery (LG and SK), and automobile (Hyundai) companies announced their commitment to invest in the U.S., and the U.S. government reciprocated by donating vaccines to South Korea [83]. The South Korean government was finally able to defend itself against "vaccine procurement failure" attacks from the right-wing opposition party and the right-wing news media, but this undermined the public perception of the necessity of the TRIPS waiver and legitimised the South Korean government's position of not supporting it [83]. COVID-19 response goods and services As well as vaccinations, the public health response to COVID-19 required a range of goods and services, including PPE, laboratory testing, rapid antigen tests, face masks, quarantine facilities, and contact tracing. In some countries the public sector played a major role, with notable examples including Vietnam [84], Taiwan and South Korea [85], where the state had a monopoly on N95 and surgical masks [though this did involve private mask production companies, 86],. This allowed the government to support universal access to masks, and to ration and prioritise masks for health services when needed. In other countries, initial attempts to engage the private sector faltered. For example, in India, COVID-19 testing and vaccination were primarily provided by the public sector [87,88] after an initial attempt to shift the task of vaccinating those aged under 60 to the private sector, later blocked following a Supreme Court ruling accompanied by widespread public protest. The converse happened in South Africa, where because of resource inequities between private and public health care [in South Africa, the private health sector consumes 50% of health spending to serve 15% of the population, 89], COVID-19 testing per population was 4.8 times higher in the private sector in South Africa compared to the public sector [90]. This was reported to be due in part to intellectual property barriers that limited access by national laboratories to test materials at key points in the pandemic [89]. The UK has come under considerable scrutiny for outsourcing key COVID-19 public health responsibilities to private companies, with poor results, and a series of procurement scandals [91,92]. A preferred provider fast track list allowed the government to bypass the usual procurement regulations, and it is alleged that one parliamentarian received £29m from a company in which her family had an interest [92]. Mobile testing centres were outsourced to security company G4S, but reports noted how they frequently failed to turn up when needed [93]. Contact tracing was replaced with a privately run tracing system that has been heavily criticised [94], and has performed particularly badly in more disadvantaged areas [95]. The UK government issued a “ventilator challenge” that saw a failed attempt to engage an inexperienced private company to manufacture ventilators [96]. This was in contrast to a ventilator project launched by the South African Department of Trade and Industry led by a national science agency which successfully stewarded private sector support to produce ventilators - a public-private collaboration led by a state-funded research institution [97]. Among other expensive UK contracting failures, a Chinese company sold antibody tests to the UK government that didn't work [98], and a Turkish company sold PPE to the UK government, but only delivered 10% of the order, and that 10% failed to meet required standards [99]. In Australia and in some other countries, quarantine was outsourced to the private hotel sector rather than purpose-built facilities, and security to enforce quarantine was outsourced to private security companies. This was heavily criticised when outbreaks occurred in these facilities [100]. In the Australian state of Victoria, these failures led to 22 breaches, resulting in hundreds of deaths, and eight lockdowns [101]. In Belgium, the outsourcing of contact tracing to private call centres became controversial when one of the call centres allegedly defrauded the government agency by diverting some staff onto other projects [102]. Actor Influences on Commercial Determinants of Health Freudenberg and colleagues split actor influences up into market-oriented practices and policy and political-oriented practices. Market-oriented practices We observed market-oriented practices in privatisation in the health sector, and emerging concerns around private aged care and prisons. Privatisation in the health sector The health sector has an obvious role in a country's response to COVID-19. While the public health sector response is largely under the control of governments, the private sector will require appropriate incentives to contribute to a public health emergency and is, inherently, more difficult to influence. In Vietnam, the private health sector had limited engagement with the response to COVID-19, with commentators expressing frustration at being unable to mobilise it [103]. In Nigeria, government encouragement of the private health sector - in part driven by externally imposed structural adjustment programs [104,105] - has left private providers delivering approximately 60% of health services in the country. This meant that access to health care was highly dependent on socioeconomic status, with poorer people accessing unlicensed, unregulated low fee commercial health providers [104]. A myriad of shortcomings were identified in the Nigerian private sector's COVID-19 response, including failure to apply relevant medical protocols and standards, and COVID-19 testing using expired reagents [104]. In Thailand, the public sector has a dominant role in health service provision and financing, with full geographical coverage of primary health care coverage and a goal of Universal Health Coverage [106]. This supported broad access to COVID-19 related services, including for migrant workers [107]. The government applied the same terms, conditions, and payment rates to public and private sectors for providing pandemic services and imposed stringent auditing measures to prevent fraud [107]. When COVID-19 hit in India, private hospitals largely abandoned health care provision for the first few months, and when they resumed, they initially would turn away COVID-19 patients. There was evidence of price gouging (where excessively high prices are charged when demand increases), increasing profits and inappropriate care [87,108]. Attempts to contract or regulate the private sector failed, and the Indian government had to rely on the public health sector in their COVID-19 response [87,108]. Williams, Yung, and Grépin [105] reported how, in low and middle income countries, private providers faced a liquidity crisis, which caused failures of provision and led to unethical behaviour such as the price gouging, and the refusal to admit and treat COVID-19 patients seen in the Indian case study. Our case studies indicated that in high income countries such as New Zealand, Australia, Taiwan, and South Korea, the private health systems have also played a very limited role in the pandemic response. In the US, the private model of primary care proved extremely vulnerable to the reduction in routine care [109]. In the cases of New Zealand and Taiwan, universal health care and the rapid, comprehensive government and community responses meant that the public health system was able to cope with the relatively low level of COVID cases [110,111]. Assa and Calderon's [112] analysis of 147 countries found that health sector privatisation may have undermined countries’ responses to COVID-19: controlling for country income and other covariates, countries with greater private health expenditure had more COVID-19 cases and deaths (with a 10% increase in private health expenditure associated with a 4.3% increase in cases and 4.9% increase in mortality). They argued that privatisation in the health sector undermined a country's long-term pandemic preparedness. The private health sector in the United States raised particular challenges during the pandemic because of the prominent role of employer-provided health insurance [113]. More than 20 million people in the United States lost their job during the pandemic [113,114]. This resulted in over 10 million people losing their health insurance coverage [inclusive of people who lost their jobs and their dependents, 114]. Others regained insurance coverage through Medicaid, the Affordable Care Act, or self-paid insurance coverage [115] – thus these alternative coverage plans were either publicly funded or paid for from peoples’ savings. There were two areas with little current evidence of the influence of commercial determinants of health but which warrant concern for the future. These have particular relevance for high income countries: privatisation of aged care and prison facilities. Aged care COVID-19 outbreaks in aged care institutions were a global phenomenon, caused by the mix of the age and characteristics of residents, and the higher risk environment of institutional settings, often exacerbated by insufficient resourcing and staffing [116], [117], [118]. Failures and subsequent outbreaks in aged care facilities were particularly noted in our Spanish and Belgian case studies, consistent with reports by Medicine Sans Frontier [119]. In Spain, 43% of all deaths in the two first waves occurred in aged care homes [120], which had poor infection control and preparedness [119]. In Belgium, where almost two thirds of deaths in the first wave were in aged care [121], the sector was found to be underprepared, with shortfalls of PPE and disinfectants, and only a 60% adherence with isolating COVID-19 positive patients [121]. Belgium has a mix of public, private, and not-for-profit aged care facilities, but no difference in COVID-19 infection rates was found between these different types of facilities [121,122]. Other studies in the US have also failed to find any differences in COVID-19 outcomes between public, private, and not-for-profit aged care homes [123]. However, Armstrong and colleagues [124] raise several concerns about privatised aged care in regards to COVID-19: that it shifted decision-making power to private companies, who are afforded secrecy on commercial competition grounds; that they have been found in general to provide lower quality care than public aged care in the US and UK; and that staff in private aged care have lower pay and higher precarity, which increases the risk that they will be a vector for COVID-19 infection into nursing homes. In Australia, which also had a high number of deaths in aged care, especially in Victoria, weaknesses were identified particularly in privatised aged care [125]. COVID-19 outbreaks in that state were found to be more likely in private aged care than public facilities [126]. Residential care in the UK, increasingly dominated by private equity with complex business models in which care homes are, in effect, a means of monetarising vulnerable people with profits accruing in tax havens, were especially hard hit [127]. Prisons COVID-19 outbreaks in prisons have been a significant health concern in many countries [128], the close quarters confinement representing a high transmission risk [128,129]. COVID-19 in prisons was raised in two of our case studies – Thailand and Australia, but there are reports of COVID-19 outbreaks in prisons in China [130], African countries [131], and the US [128,129]. This becomes a commercial determinant of health when the extent of privatisation of prisons is considered. The US relies extensively on private prisons [132,133], even though for-profit prisons have many flaws, including poorer prisoner safety [133], and there were many shortcomings throughout both the private and public the US prison sector COVID-19 response [129,134]. Similarly, the UK, which has a substantial proportion of private prisons [135], faced criticism over their COVID-19 handling [134]. Australia has one of the highest proportions of prisoners incarcerated in private facilities in the world [136]. Payne and Hanley [136] provided an example of an outbreak in an Australian private prison, and raised concerns about how little influence governments can have over the COVID-19 response of private prisons. Policy and political-oriented practices Commercial actors can have considerable political power, and their interests can run counter to public health. In Australia, the government's ongoing management of COVID-19 was subject to extensive commercial lobbying from the business sector, including an open letter calling for lockdowns to end [137]. There was a legal challenge from mining company owner and pro-business politician Clive Palmer to Western Australia's COVID-19 triggered state border closure when the closure threatened Palmer's business operations [138]. In the New Zealand case study, the business sector lobbied, relatively unsuccessfully, to curtail COVID-19 measures recommended by epidemiologists, citing the more permissive Australian regulations. In South Africa, the mining industry lobbied for vaccine priority for their workers based on societal economic benefits, and the tobacco and alcohol industries lobbied against restrictions on their products during lockdowns, spending millions on public marketing campaigns [139]. These examples show the potential for privileging commercial interests over public health. Discussion It is accepted that social determinants of health exert considerable influence on people's experience of the COVID-19 pandemic and their health outcomes [26,28,29]. We show how central commercial determinants of health have also shaped the course of the pandemic, the government and health system response, and health outcomes. It is important to distinguish commercial determinants from other social determinants because they require a different set of responses [2]. While governments can invest in positive determinants of health through the provision of resources such as housing and education, addressing the negative impacts of commercial determinants of health requires different approaches. Addressing commercial determinants may require strategies such as regulation of business practices, winding back privatisation, altering financial incentives and subsidies to address commercial practices, civil society activism, and litigation to combat harmful commercial determinants of health [2]. Addressing the commercial determinants necessitates balancing economic considerations and the interests of often powerful commercial actors against the health, equity and wellbeing of the population [140]. The COVID-19 pandemic highlighted the areas of conflict between commercial and public health interests, and there was concern documented in many case studies that governments prioritised commercial interests over population health in some instances. This is despite the finding that countries who fared worse in terms of COVID-19 deaths also fared worse in loss of gross domestic product [141], which suggested that commercial and public health goals ought to have been considered more in alignment rather than in opposition. We found evidence of the influence of commercial determinants of health in the origin and spread of COVID-19, through; shaping working conditions that influenced workers’ risk of contracting and spreading the SARS-CoV-2 virus; affecting the procurement of PPE, rapid antigen tests, and other goods and services that help protect against spread of the virus; and commercial actors lobbying against public health regulations that have the potential to reduce transmission, when those public health regulations threatened profit. The commercial determinants of health also influenced the availability and rollout of vaccines that had the capacity to reduce the severity of illness, and the health system response to the virus, which was complicated in countries with an extensive private health sector, who often failed to engage in the public response to COVID-19, and in some cases instead used it as an opportunity to refuse care to people with COVID-19, and to engage in price gouging for care for COVID-19. Thus, the commercial determinants of health would also have directly influenced the health outcomes of the pandemic in addition to exacerbating the spread of the virus. We found that there is only mixed evidence on the effects of privatisation of aged care and prisons on the spread of COVID-19. What is known raises important questions about how the pursuit of profit by the companies that run these facilities can be balanced with the health and wellbeing of the residents. If adequate regulation is not put in place, private facilities will remain difficult to mobilise for public health goals, and it will remain a governance challenge to ensure the health and wellbeing of residents, workers, and the broader community are not compromised in the pursuit of profit [124,136]. These findings illustrate the urgent need that Diderichsen, Dahlgren and Whitehead [15] identified of including privatisation as a central concern in the commercial determinants of health, so we can understand better how the introduction of these market forces into previously public spheres affect health and equity. COVID-19 cases and mortality have been distributed inequitably in the population, along socioeconomic, racial, and other social exclusion lines [26,142]. Our findings show that commercial determinants of health may be one of the contributing factors to this inequity by exacerbating inequities that already affect many groups. For example, we found commercial interests shaping working conditions that were detrimental to the health and safety of low skilled and foreign-born workers, that private prisons may have adversely affected the health of prisoners, and that shortcomings in vaccine availability particularly affected low-income countries. Inequity has to-date received little attention in the commercial determinants of health literature, although there are indications that the effects of commercial determinants on inequities have been recognised without explicitly framing them as commercial determinants [3]. However, it should be noted that manufacturers of harmful commodities, corruption, and procurement failures are explicitly addressed in the evidence review underpinning the work of the Pan-European Commission on Health and Sustainable Development, which is being taken forward by the World Health Organization Regional Office for Europe [143]. Experiences of the pandemic have been heavily gendered, including the way that essential health and care workers are predominantly female. Cohen and van der Meulen Rodgers [144] also argue that in capitalist countries, capitalism has exacerbated these gender inequities during the pandemic. Our findings show the value in being guided by a framework, in this case Freudenberg et al.’s [18], to ensure different component practices, and, importantly, the societal structures that support the prioritisation and impact of commercial determinants, are examined. The political economy within which the COVID-19 pandemic has played out is vital to understanding the differing economic and commercial forces influencing COVID-19 outcomes in different countries [145]. Bump and colleagues argue that “Internationally, the political economy of COVID-19 reflects global patterns of extraction that were established in colonial times” and our case studies reinforced this [145]. Greater international action is needed to address these longstanding inequities, to ensure future pandemics do not face the same extent of between country inequities in vaccine access and health outcomes. Within countries, our findings support the argument for more pushback against privatisation in the health system, aged care, prisons, and other sectors, to better safeguard the health and wellbeing of the population. The recurrence of similar negative impacts across the 16 countries suggests that, rather than reflecting a few misbehaving companies, the examples we have presented indicate how the practices of the private sector are influenced by their for-profit motive, and fail to prioritise equity, access, healthy workplace conditions, and worker wellbeing. This means that careful public oversight is needed (such as in the South African case for ventilators, and the Taiwan case for masks) to ensure the outcomes arising from involving the private sector in a public pandemic response are positive and equitable. Renationalising privatised services would allow equity and public good goals to be prioritised and safeguard the rights and wellbeing of residents and recipients of these services. At the very least, improving state oversight and governance of private services to make them more likely to pursue public good goals is critical. Our analysis also indicates that we need greater advocacy to governments to hold them accountable for governing for health, rather than governing for the profit of commercial actors [146]. Our findings add to calls for stronger regulation of private sector business practices to ensure healthy and just employment practices, including greater requirements on companies to safeguard the health of their workers from infectious diseases. Our findings also add to calls for more equitable vaccine distribution, including addressing the profits private companies have made off vaccinations at the expense of more comprehensive and equitable availability. Limitations This research looked at the country level to examine the impact of commercial determinants of health. Freudenberg and colleagues [18] also flag the influence of commercial interests on global organisations that are critical to public health such as the World Health Organization and the World Trade Organization, reducing global co-ordination to safeguard population health against commercial determinants. De Lacy-Vawdon, Vandenberg, and Livingstone [12] and Barlow and colleagues [37] also highlight the role of multinational free trade agreements. These global structures also warrant scrutiny for their influence of COVID-19 health outcomes. Another global consideration that sat outside our case studies is the role that private media [147] and social media companies, such as Facebook and Twitter [148], play in the dissemination of health information, misinformation and disinformation (the latter designed to mislead). COVID-19 disinformation has spread inaccurate beliefs about COVID-19 and can decrease the seriousness with which people treat the pandemic, increasing vaccine hesitancy and refusal, and strengthening opposition to governments enforcing public health rules, all of which hampered the containment of the pandemic [147]. Yamey and Gorski have documented how a US libertarian group funded by, among others, the oil and tobacco industries, supported one of the most widely disseminated documents undermining public health messaging [149]. The value of using case studies of countries with country experts was to provide a rich, rigorous overview of critical factors influencing the countries’ experiences of the COVID-19 pandemic informed by local expertise, supported by grey and peer reviewed literature, including media, and material in languages other than English. We found the depth and quality of data far surpassed what would have been possible using a standard literature review method. We used the best available evidence for each point, giving the rapidly changing nature of the COVID-19 pandemic, and the inclusion of political and other perspectives not common in the peer reviewed literature. We included case studies of countries with higher and lower rates of COVID-19 cases and mortality. However, assigning attribution to particular drivers of these outcomes is a very difficult undertaking given the complexity of factors and contexts that differed between countries that interacted to produce COVID-19 outcomes, including geography, political leadership, health systems, previous experience of epidemics, and government responses [26]. Nevertheless, the positive health outcomes for countries such as Taiwan, South Korea, Thailand, Vietnam and New Zealand, who all emphasised a strong public sector response to the pandemic (exemplified in Vietnam's motto of ‘saving lives is prioritised above consideration of the economic loss’ [150]), contrasted with the high COVID-19 toll in countries plagued with controversies around commercial interests, such as private sector procurement arrangements, and heavy reliance on a private health sector, such as the UK, US, and South Africa. Conclusions We found extensive scope for the commercial determinants of health to lead to adverse health outcomes in the COVID-19 pandemic through a multitude of avenues: they are likely to have affected the origin and spread of the virus, the health system response to the virus, and the availability and rollout of COVID-19 vaccines. It is crucial to extend scholarship on the commercial determinants of health and health equity beyond its main focus on unhealthy commodities and non-communicable diseases to understand their influence on both the spread and control of infectious diseases such as COVID-19. Understanding the influence of commercial determinants of health will help build evidence to advocate for the role of governments in governing for health, wellbeing, and equity, and provide knowledge to underwrite regulation to address the negative impacts of commercial determinants of health that may otherwise undermine these goals. CRediT authorship contribution statement Toby Freeman: Conceptualization, Methodology, Formal analysis, Writing – original draft, Funding acquisition. Fran Baum: Conceptualization, Methodology, Formal analysis, Writing – review & editing, Supervision, Funding acquisition. Connie Musolino: Conceptualization, Methodology, Formal analysis, Writing – review & editing, Project administration, Funding acquisition. Joanne Flavel: Methodology, Formal analysis, Investigation, Writing – review & editing. Martin McKee: Investigation, Writing – review & editing. Chunhuei Chi: Investigation, Writing – review & editing. Camila Giugliani: Investigation, Writing – review & editing. Matheus Zuliane Falcão: Investigation, Writing – review & editing. Wim De Ceukelaire: Investigation, Writing – review & editing. Philippa Howden-Chapman: Investigation, Writing – review & editing. Nguyen Thanh Huong: Investigation, Writing – review & editing. Hani Serag: Investigation, Writing – review & editing. Sun Kim: Investigation, Writing – review & editing. Carlos Alvarez Dardet: Investigation, Writing – review & editing. Hailay Abrha Gesesew: Investigation, Writing – review & editing, Funding acquisition. Leslie London: Investigation, Writing – review & editing. Jennie Popay: Investigation, Writing – review & editing, Funding acquisition. Lauren Paremoer: Investigation, Writing – review & editing. Viroj Tangcharoensathien: Investigation, Writing – review & editing. T Sundararaman: Investigation, Writing – review & editing. Sulakshana Nandi: Investigation, Writing – review & editing. Eugenio Villar: Investigation, Writing – review & editing. Declaration of Competing Interest Declarations of interest: none Appendix Supplementary materials Image, application 1 Acknowledgements Funding: This work was supported by a contract from the World Bank, and by an Australian National Health and Medical Research Council Investigator Fellowship (Baum, grant number 20099223). The funding sources had no role in the design, collection, analysis, interpretation, or writing of the manuscript. Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.healthpol.2023.104860. ==== Refs References 1 Glover R Petticrew M. Defining the commercial determinants of health after COVID-19 European Journal of Public Health 31 2021 ckab164 808 2 Lee K Freudenberg N. Public health roles in addressing commercial determinants of health Annual Review of Public Health 43 2022 375 395 3 Maani N Collin J Friel S Gilmore AB McCambridge J Robertson L Petticrew MP. Bringing the commercial determinants of health out of the shadows: a review of how the commercial determinants are represented in conceptual frameworks European Journal of Public Health 30 2020 660 664 31953933 4 Gilmore AB Fabbri A Baum F Bertscher A Bondy K Chang H-J Demaio S Erzse A Freudenberg N Friel S. Defining and conceptualising the commercial determinants of health The Lancet 401 2023 1194 1213 5 Baum F Sanders DM Fisher M Anaf J Freudenberg N Friel S Labonté R London L Monteiro C Scott-Samuel A. Assessing the health impact of transnational corporations: its importance and a framework Globalization and Health 12 2016 1 7 26739193 6 Baum F Margaret Anaf J Transnational corporations and health: a research agenda International Journal of Health Services 45 2015 353 362 25674798 7 Kickbusch I Allen L Franz C. The commercial determinants of health The Lancet Global Health 4 2016 e895 e8e6 27855860 8 Lacy-Nichols J Marten R Power and the commercial determinants of health: ideas for a research agenda BMJ Global Health 6 2021 e003850 9 McKee M Stuckler D. Revisiting the corporate and commercial determinants of health American journal of public health 108 2018 1167 1170 30024808 10 Lee K Crosbie E. Understanding Structure and Agency as Commercial Determinants of Health: Comment on" How Neoliberalism Is Shaping the Supply of Unhealthy Commodities and What This Means for NCD Prevention International Journal of Health Policy and Management 9 2020 315 32613804 11 Gómez EJ. Enhancing our understanding of the commercial determinants of health: Theories, methods, and insights from political science Social Science & Medicine 301 2022 114931 12 de Lacy-Vawdon C Vandenberg B Livingstone CH. Recognising the elephant in the room: the commercial determinants of health BMJ Global Health 7 2022 e007156 13 Jones R. In whose interest?: The privatisation of child protection and social work. Policy Press, 2018. 14 Lethbridge J. Privatisation of migration and refugee services and other forms of state disengagement. 2017. 15 Diderichsen F Dahlgren G Whitehead M. Beyond ‘commercial determinants’: shining a light on privatization and political drivers of health inequalities European Journal of Public Health 31 2021 672 673 34137787 16 Baru RV. Privatisation of health services: A South Asian perspective Economic and Political Weekly 2003 4433 4437 17 Sonntag D Schneider S Mdege N Ali S Schmidt B. Beyond Food Promotion: A Systematic Review on the Influence of the Food Industry on Obesity-Related Dietary Behaviour among Children Nutrients 7 2015 8565 8576 26501319 18 Freudenberg N Lee K Buse K Collin J Crosbie E Friel S Klein DE Lima JM Marten R Mialon M Defining priorities for action and research on the commercial determinants of health: a conceptual review American Journal of Public Health 111 2021 2202 2211 34878875 19 Lee K Freudenberg N Zenone M Smith J Mialon M Marten R Lima JM Friel S Klein DE Crosbie E. Measuring the commercial determinants of health and disease: a proposed framework International Journal of Health Services 52 2022 115 128 34723675 20 Sanders D Sengupta A Scott V. Ebola epidemic exposes the pathology of the global economic and political system International Journal of Health Services 45 2015 643 656 26391140 21 Kentikelenis A King L McKee M Stuckler D. The International Monetary Fund and the Ebola outbreak Lancet Glob Health 3 2015 e69 e70 25540014 22 Zhang L. The origins of COVID-19: China and global capitalism. Stanford University Press, 2021. 23 Thoradeniya T Jayasinghe S. COVID-19 and future pandemics: a global systems approach and relevance to SDGs Globalization and Health 17 2021 1 10 33397377 24 United Nations. Goal 15 Life on Land. 2020. 25 Rietig K Cashore B Clough E Long G Nathan I Peringer C Cansino H Censoro J Muggleton E. The'net'in net-zero greenhouse gas emissions: Achieving just transitions in the forestry sector through climate policy integration and learning 2022 The British Academy London, UK 26 Baum F Freeman T Musolino C Abramovitz M De Ceukelaire W Flavel J Friel S Giugliani C Howden-Chapman P Huong NT. Explaining COVID-19 performance: what factors might predict national responses? BMJ 2021 372 27 Sagan A Webb E McKee M Greer SL Karanikolos M Williams GA Cylus J Richardson E Waitzberg R Lessof S Figueras J Falkenbach M Hernandez-Quevedo C Klasa K Mauer N Panteli D Permanand G Quentin W Rechel B Rozenblum SD Thomas S Willoughby EL Winkelmann J Wismar M. Health systems resilience during COVID-19: Lessons for building back better Copenhagen (Denmark): European Observatory on Health Systems and Policies 2021 28 Paremoer L Nandi S Serag H Baum F. Covid-19 pandemic and the social determinants of health BMJ 2021 372 29 Flavel J Baum F. The influence of socio-economic conditions on the epidemiology of COVID-19 in Australia The Medical Journal of Australia 216 2022 344 345 35318679 30 Devakumar D Selvarajah S Abubakar I Kim SS McKee M Sabharwal NS Saini A Shannon G White AIR Achiume ET. Racism, xenophobia, discrimination, and the determination of health Lancet 400 2022 2097 2108 36502848 31 Dall'Alba R Rocha DG. Brazil's response to COVID-19: commercial determinants of health and regional inequities matter The Lancet Global Health 9 2021 e726 e7e7 33857501 32 Hellmann F Homedes N. An unethical trial and the politicization of the COVID-19 pandemic in Brazil: The case of Prevent Senior Developing World Bioethics 2022 Early View 33 Ventura D Aith F Reis R. Crimes against humanity in Brazil's covid-19 response—a lesson to us all BMJ 2021 375 34 Maani N Van Schalkwyk MC Petticrew M Galea S. The Commercial Determinants of Three Contemporary National Crises: How Corporate Practices Intersect With the COVID-19 Pandemic, Economic Downturn, and Racial Inequity The Milbank Quarterly 99 2021 503 33783862 35 Lalani HS Avorn J Kesselheim AS. US Taxpayers Heavily Funded the Discovery of COVID-19 Vaccines Clin Pharmacol Ther 111 2022 542 544 34243221 36 Gleeson D Tenni B Townsend B. Four actions Australia should take to advance equitable global access to COVID-19 vaccines Australian and New Zealand Journal of Public Health 2022 37 Barlow P van Schalkwyk MC McKee M Labonté R Stuckler D. COVID-19 and the collapse of global trade: building an effective public health response Lancet Planet Health 5 2021 e102 e1e7 33581061 38 van Schalkwyk MCI Maani N McKee M. Public health emergency or opportunity to profit? The two faces of the COVID-19 pandemic The Lancet Diabetes & Endocrinology 9 2021 61 63 33417836 39 Yin RK. Case study research and applications: Design and methods 2018 SAGE Los Angeles, CA 40 Ketokivi M Choi T. Renaissance of case research as a scientific method Journal of Operations Management 32 2014 232 240 41 Institute for Health Metrics and Evaluation. COVID-19 projections: Cumulative deaths 2022. 42 Mialon M. An overview of the commercial determinants of health Globalization and Health 16 2020 1 7 31898532 43 Maiangwa B Suleiman MD Anyaduba CA. The nation as corporation: British colonialism and the pitfalls of postcolonial nationhood in Nigeria Peace and Conflict Studies 25 2018 3 44 Maiangwa B. The colonial enterprise hard-baked violence in Nigeria: how it can be fixed. The Conversation, 2018. 45 CA-COVID. Private sector coalition against COVID-19. 2022. 46 Zondo R. The judicial commission of inquiry into allegations of state capture, corruption and fraud in the public sector including organs of state. Department of Justice and Constitutional Development Notice 396 of 2018 2018. 47 Cruywagen V, Heywood M. Murder of Gauteng health official Babita Deokaran: Investigators probe link to her PPE whistle-blowing. Daily Maverick. South Africa, 2021. 48 Chi C. Taiwan and the politics of vaccine warfare East Asia Forum 2021 49 Schwalb A Seas C. The COVID-19 pandemic in Peru: what went wrong? The American Journal of Tropical Medicine and Hygiene 104 2021 1176 33591940 50 de la Puente J. La gran depresión y el fracaso peruano: balance de la primera ola del Coronavirus. 2021. 51 Le T-AT Vodden K Wu J Atiwesh G. Policy responses to the COVID-19 pandemic in Vietnam International Journal of Environmental Research and Public Health 18 2021 559 33440841 52 Huynh D Tosun MS Yilmaz S. All-of-government response to the COVID-19 pandemic: The case of Vietnam Public Administration and Development 40 2020 236 239 53 Hengel KMO Burdorf A Pronk A Schlünssen V Stokholm ZA Kolstad HA van Veldhoven K Basinas I van Tongeren M Peters S. Exposure to a SARS-CoV-2 infection at work: development of an international job exposure matrix (COVID-19-JEM) Scandinavian journal of work, environment & health 48 2022 61 54 Tweedie D, Chan S. Precarious work and globalisation in Australia: Growth, risks, and future (s). Identifying and Managing Risk at Work: Routledge, 2021:82-96. 55 Park SY Kim Y-M Yi S Lee S Na B-J Kim CB J-i Kim Kim HS Kim YB Park Y Coronavirus disease outbreak in call center South Korea. Emerging Infectious Diseases 26 2020 1666 32324530 56 Choi H Kim SY Kim JW Park Y Kim MH. Mainstreaming of Health Equity in Infectious Disease Control Policy During the COVID-19 Pandemic Era Journal of Preventive Medicine and Public Health 54 2021 1 7 33618493 57 Webb A McQuaid R Rand S. Employment in the informal economy: Implications of the COVID-19 pandemic International Journal of Sociology and Social Policy 40 2020 1005 1019 58 Reid A Ronda-Perez E Schenker MB. Migrant workers, essential work, and COVID-19 American Journal of Industrial Medicine 64 2021 73 77 33355943 59 Lotta G Kuhlmann E. When informal work and poor work conditions backfire and fuel the COVID-19 pandemic: Why we should listen to the lessons from Latin America The International Journal of Health Planning and Management 36 2021 976 979 33368589 60 Fassani F Mazza J. A vulnerable workforce: Migrant workers in the COVID-19 pandemic 2020 Publications Office of the European Union Luxembourg 61 Saparamadu AADNS Sharpe A Kim S Barbosa BLFA Pereira A. Low-wage migrant workers during coronavirus disease 2019: a social determinants analysis Journal of Public Health Policy 42 2021 452 464 34417557 62 Rajatanavin N Tuangratananon T Suphanchaimat R Tangcharoensathien V. Responding to the COVID-19 second wave in Thailand by diversifying and adapting lessons from the first wave BMJ Global Health 6 2021 e006178 63 Suresh R James J RSj B. Migrant workers at crossroads–The COVID-19 pandemic and the migrant experience in India Social Work in Public Health 35 2020 633 643 32970543 64 Kohler T Hill R. The distribution and dynamics of South Africa's TERS policy: Results from NIDS-CRAM Waves 1 to 5. Cape Town 2021 University of Cape Town South Africa 65 Asundi A O'Leary C Bhadelia N Global COVID-19 vaccine inequity: The scope, the impact, and the challenges Cell Host & Microbe 29 2021 1036 1039 34265241 66 Bayati M Noroozi R Ghanbari-Jahromi M Jalali FS. Inequality in the distribution of Covid-19 vaccine: a systematic review International Journal for Equity in Health 21 2022 1 9 34980116 67 Asadu C. Nigeria faces deadly Covid-19 Delta wave as infections jump 150% The Africa Report 2021 68 Bajaj SS Maki L Stanford FC. Vaccine apartheid: global cooperation and equity The Lancet 399 2022 1452 1453 69 Storeng KT Stein F de Bengy Puyvallée A. COVAX and the many meanings of sharing BMJ Global Health 6 2021 e007763 70 Oxfam International. Pfizer, BioNTech and Moderna making $1,000 profit every second while world's poorest countries remain largely unvaccinated. 2021. 71 Our World in Data. Coronavirus (COVID-19) Vaccinations. 2022. 72 Wouters OJ Shadlen KC Salcher-Konrad M Pollard AJ Larson HJ Teerawattananon Y Jit M. Challenges in ensuring global access to COVID-19 vaccines: production, affordability, allocation, and deployment The Lancet 397 2021 1023 1034 73 Oxfam International. Vaccine monopolies make cost of vaccinating the world against COVID at least 5 times more expensive than it could be. 2021. 74 Furlong A Aarup SA Horti S. Who killed the COVID vaccine waiver? Politico 2022 75 Green A. WTO finally agrees on a TRIPS deal But not everyone is happy. Devex 2022 76 Burnside N. In choosing 'strollout' as its Word of the Year, the National Dictionary Centre alludes to a uniquely Australian problem. ABC News, 2021. 77 Davies M Furneaux R. Pfizer backs down over ‘unreasonable’ terms in South Africa vaccine deal Mail & Guardian 2021 78 Lee S Woodward R. From financing social insurance to insuring financial markets: The socialisation of risk and the privatisation of profit in an age of irresponsibility The Withering of the Welfare State: Regression 2012 121 136 79 Reuters SA's vaccine compensation fund could cost R250m in first year - Mkhize IOL 2021 80 SABC. J&J refusing to sign off on supply of 20 million COVID-19 vaccine doses: Mkhize. SABC News, 2021. 81 Roveri E, Falcao MZ, Navarrete AC. Innovation as a public strategy: The experience of Brazilian public laboratories in the development and production of vaccines against Covid-19. Sao Paulo, Brazil: Instituto Brasileiro de Defesa do Consumidor. 82 Perez-Brumer A Silva-Santisteban A. The Peruvian COVID-19 vaccine scandal and re-thinking the path to public trust Global Public Health 17 2022 3119 3125 34813717 83 People's Health Institute. COVID-19 Vaccinations and People's Rights. South Korea: People's Health Institute, 2021. 84 Turner M Kwon S-H O'donnell M State effectiveness and crises in East and Southeast Asia: the case of COVID-19 Sustainability 14 2022 7216 85 Kim S. Success of K-pandemic prevention lies in 'publicness of the system', not the market Rapportian 2020 86 Wang CJ Ng CY Brook RH. Response to COVID-19 in Taiwan: Big Data Analytics, New Technology, and Proactive Testing JAMA 323 2020 1341 1342 32125371 87 Garg S Bebarta KK Tripathi N Krishnendhu C. Catastrophic health expenditure due to hospitalisation for COVID-19 treatment in India: findings from a primary survey BMC Research Notes 15 2022 1 7 34983646 88 Nandi S Das N Chapter 21: Impact of Covid-19 on health and the health system: Glimpses from Chhattisgarh Jain Y Nabiya S COVID-19- A View from the Margins 2022 Manohar New Delhi 89 Tomlinson C. A rapid situational analysis of barriers and enablers to equitable access to COVID-19 health technologies in South Africa. Cape Town: People's Health Movement− South Africa 2020 90 National Institute for Communicable Diseases COVID-19 Testing Summary South Africa Week 34 2021 2021 NICD Johannesburg 91 McKee M. The UK's PPE procurement scandal reminds us why we need ways to hold ministers to account BMJ 372 2021 n639 33685915 92 Conn D. Revealed: Tory peer Michelle Mone secretly received £29m from ‘VIP lane’ PPE firm 2022 The Guardian UK 93 Williams J. Mobile units turn up late - and sometimes not at all as testing in Greater Manchester descends into chaos after being taken off the army and given to private sector. Manchester Evening News, 2020. 94 Scally G Jacobson B Abbasi K. The UK's public health response to covid-19 BMJ 369 2020 m1932 32414712 95 Briggs ADM Fraser C. Is NHS Test and Trace exacerbating COVID-19 inequalities? Lancet 396 2021 1972- 33285140 96 Pooler M. Dyson says its ventilator not needed in UK. Financial Times, 2020. 97 van der Merwe C. South Africa's ventilator project shows value of basic research Research Professional News 2021 98 Kirkpatrick D Bradley J.U.K Paid $20 Million for New Coronavirus Tests They Didn't Work. NY Times 2020 99 Rawlinson K. Coronavirus PPE: all 400,000 gowns flown from Turkey for NHS fail UK standards. The Guardian, 2020. 100 Dwyer G. Hiding in plain sight: Vulnerability, public administration, and the case of Covid-19 hotel quarantine Australian Journal of Public Administration 80 2021 1002 1016 101 Grout L Katar A Ait Ouakrim D Summers JA Kvalsvig A Baker MG Blakely T Wilson N Failures of quarantine systems for preventing COVID-19 outbreaks in Australia and New Zealand Medical Journal of Australia 215 2021 320 324 34472122 102 VRTNWS Suspected fraud at Flemish contact tracing call centre VRTNWS 2021 103 Van Hoang M Tran AT Vu TT Duong TK. Covid-19 Preparedness and Response Capability: A Case Study of the Hanoi Primary Healthcare System Health Services Insights 14 2021 11786329211019224 104 Global Initiative for Economic Social and Cultural Rights, Justice & Empowerment Initiative. The failure of commercialised healthcare in Nigeria during the COVID-19 pandemic: Discrimination and inequality in the enjoyment of the right to health. DOI: 10.53110/ZYQT7031. 2022. 105 David Williams O Yung KC Grépin KA The failure of private health services: COVID-19 induced crises in low-and middle-income country (LMIC) health systems Global Public Health 16 2021 1320 1333 33471633 106 Tangcharoensathien V Witthayapipopsakul W Panichkriangkrai W Patcharanarumol W Mills A. Health systems development in Thailand: a solid platform for successful implementation of universal health coverage The Lancet 391 2018 1205 1223 107 Sachdev S Viriyathorn S Chotchoungchatchai S Patcharanarumol W Tangcharoensathien V. Thailand's COVID-19: How public financial management facilitated effective and accountable health sector responses The International Journal of Health Planning and Management 2022 108 Thiagarajan K. COVID-19 exposes the high cost of India's reliance on private healthcare BMJ 2020 370 109 Arora VS McKee M. Covid-19 has decimated independent US primary care practices—how should policymakers and payers respond? BMJ 2020 110 Howden-Chapman P, Gatzweiller F, Luginaah I, Cooper R. Cities Under COVID-19: A Systems Approach. Springer Press, in press. 111 Chi C. Country Responses to the Covid19 Pandemic- Taiwan's Response to the Coronavirus Pandemic. Health Economic Policy and Law blog series, 2020. 112 Assa J Calderon C. Privatization and pandemic: a cross-country analysis of COVID-19 rates and health-care financing structures 2020 The New School for Social Research New York 113 Blumenthal D Fowler EJ Abrams M Collins SR. Covid-19 - Implications for the Health Care System New England Journal of Medicine 383 2020 1483 1488 32706956 114 Banthin J Simpson M Buettgens M Blumberg LJ Wang R. Changes in health insurance coverage due to the COVID-19 recession: preliminary estimates using microsimulation 2020 Urban Institute Washington, DC 1 9 115 Cole B. The Impact of the COVID-19 Pandemic on Access to Health Care National Academy of Social Insurance 2020 116 Causa R Nievas DA Tamayo CB. COVID-19 and functional dependence: cohort study of an outbreak in a nursing home for elderly Rev Esp Salud Publica 95 2021 e202103045 117 Van Houtven C Miller K Gorges R Campbell H Dawson W McHugh J McGarry B Gilmartin R Boucher N Kaufman B. State Policy Responses to COVID-19 in Nursing Homes Journal of Long-Term Care 2021 118 Mahase E. Covid-19: Neglect was one of biggest killers in care homes during pandemic, report finds British Medical Journal 375 2021 n3132 34937696 119 Thornton J. COVID-19: Care homes in Belgium and Spain had “alarming living conditions,” says MSF report BMJ (Online) 370 2020 10.1136/bmj.m3271 120 Zunzunegui MV. COVID-19 in care homes: equity will be needed to avoid new catastrophes Gaceta Sanitaria 36 2021 3 5 34400008 121 Medicins Sans Frontieres Left behind in the times of COVID-19: Médecins Sans Frontières/Doctors Without Borders (MSF) sharing experiences from its intervention in care homes in Belgium 2020 MSF Brussels, Belgium 122 Peckeu-Abboud L van Kleef E Smekens T Latour K Dequeker S Panis LI Laga M. Factors influencing SARS-CoV-2 infection rate in Belgian nursing home residents during the first wave of COVID-19 pandemic Epidemiology & Infection 150 2022 123 Braun RT Yun H Casalino LP Myslinski Z Kuwonza FM Jung H-Y Unruh MA. Comparative performance of private equity–owned US nursing homes during the COVID-19 pandemic JAMA Network Open 3 2020 e2026702-e 124 Armstrong P Armstrong H Bourgeault I. Privatization and COVID-19: a deadly combination for nursing homes. Vulnerable: The Law, Policy and Ethics of COVID-19 2020 University of Ottawa Press Ontario, ON, Canada 451 452 125 Lucas C. Why Victorian public aged care homes were spared the worst of coronavirus. The Age. Melbourne, Victoria, 2022. 126 Ibrahim JE Li Y McKee G Eren H Brown C Aitken G Pham T. Characteristics of nursing homes associated with COVID-19 outbreaks and mortality among residents in Victoria Australia. Australasian Journal on Ageing 40 2021 283 292 34288338 127 Rajan S Comas-Herrera A McKee M. Did the UK government really throw a protective ring around care homes in the COVID-19 pandemic? Journal of Long-Term Care 2020 2020 185 195 128 Kinner SA Young JT Snow K Southalan L Lopez-Acuña D Ferreira-Borges C É O'Moore Prisons and custodial settings are part of a comprehensive response to COVID-19 The Lancet Public Health 5 2020 e188 32197116 129 Hummer D. United States Bureau of Prisons’ Response to the COVID-19 Pandemic Victims & Offenders 15 2020 1262 1276 130 Yang H Thompson JR. Fighting COVID-19 outbreaks in prisons BMJ 369 2020 m1362 32241756 131 Muntingh LM. Africa Prisons and COVID-19 Journal of Human Rights Practice 12 2020 284 292 132 Craig R Pond Cummings AD Abolishing private prisons: A constitutional and moral imperative U Balt L Rev 49 2019 261 133 Ahmed H. How private prisons are profiting under the Trump administration. Center for American Progress https://wwwamericanprogressorg/issues/democracy/reports/2019/08/30/473966/privateprisons-profiting-trump-administration 2019. 134 Burki T. Prisons are “in no way equipped” to deal with COVID-19 The Lancet 395 2020 1411 1412 135 Crewe B, Liebling A. Quality, professionalism and the distribution of power in public and private sector prisons. The Private Sector and Criminal Justice: Springer, 2018:161-94. 136 Payne JL Hanley N. COVID-19 and corrections in Australia: a summary review of the available data and literature Victims & Offenders 15 2020 1367 1384 137 Marsh S. Australia's biggest businesses implore state leaders to end lockdowns once nation reaches 80 per cent vaccination rate 9News 2021 138 Wesson M, Murray I. Explainer: why did the High Court rule against Clive Palmer and what does the judgment mean? The Conversation, 2021. 139 Ngqangashe Y. What South Africa's COVID alcohol restrictions point to for future policy. The Conversation, 2021. 140 Baum F. Governing for health 2018 Oxford University Press USA 141 Smithson M. Data from 45 countries show containing COVID vs saving the economy is a false dichotomy. The Conversation, 2020. 142 Organization WH. Health inequity and the effects of COVID‑19: assessing, responding to and mitigating the socioeconomic impact on health to build a better future. World Health Organization. Regional Office for Europe, 2020. 143 McKee M. Drawing light from the pandemic: A new strategy for health and sustainable development European Observatory on Health Systems and Policies 2021 144 Cohen J van der Meulen Rodgers Y. The feminist political economy of Covid-19: Capitalism, women, and work Global Public Health 16 2021 1381 1395 33905301 145 Bump JB Baum F Sakornsin M Yates R Hofman K. Political economy of COVID-19: extractive, regressive, competitive BMJ 2021 372 146 Friel S Collin J Daube M Depoux A Freudenberg N Gilmore AB Johns P Laar A Marten R McKee M. Commercial determinants of health: future directions The Lancet 401 2023 1229 1240 147 Motta M Stecula D Farhart C. How Right-Leaning Media Coverage of COVID-19 Facilitated the Spread of Misinformation in the Early Stages of the Pandemic in the U.S Canadian Journal of Political Science 53 2020 335 342 148 Yang K-C Pierri F Hui P-M Axelrod D Torres-Lugo C Bryden J Menczer F. The COVID-19 Infodemic: Twitter versus Facebook Big Data & Society 8 2021 20539517211013861 149 Yamey G Gorski D. Covid-19 and the new merchants of doubt BMJ 2021 150 Nguyen TV Tran QD Phan LT Vu LN Truong DTT Truong HC Le TN Vien LDK Nguyen TV Luong QC Pham QD. In the interest of public safety: rapid response to the COVID-19 epidemic in Vietnam BMJ Global Health 6 2021 e004100
PMC010xxxxxx/PMC10287179.txt	==== Front Int J Ind Organ Int J Ind Organ International Journal of Industrial Organization 0167-7187 1873-7986 Elsevier B.V. S0167-7187(23)00057-7 10.1016/j.ijindorg.2023.102976 102976 Article The home bias in procurement. Cross-border procurement of medical supplies during the Covid-19 pandemic. Hanspach Philip ⁎1a European University Institute ⁎ Corresponding author. 1 I thank Giacomo Calzolari, David K. Levine, Stephen Davies, Laurens Vandercruysse, Joosua Virtanen and two anonymous referees for comments, as well as participants at CLEEN 2021, CRESSE 2021 and working groups at EUI. The usual disclaimer applies. Funding by the German Academic Exchange Service is gratefully acknowledged. 23 6 2023 23 6 2023 10297616 5 2023 19 5 2023 © 2023 Elsevier B.V. All rights reserved. 2023 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Public procurement markets are often national a general agreement national preferencing. I exploit shocks occurring during the Covid-19 pandemic to two important factors, crisis urgency, measured through local infection rates, and increased buyer discretion, to study home bias in public procurement. Two causal difference-in-difference analyses on novel data for medical supplies in Europe show that home bias is not inevitable. An increase in local infection rates by one standard deviation locally increases the share of cross-border procurement by 19.3 percentage points over a baseline of 1.5 percent. Also, deregulation that allowed for buyer discretion caused cross-border procurement to increase by more than 35 percentage points. A simple theoretical model systematizes these findings. Keywords public procurement home bias regulation difference-in-differences Covid-19 ==== Body pmc1 Introduction Public sector procurement accounts for a large share of the economy, at 15-20% of gross domestic product in the European Union (EU). Economists have identified sources of misallocation that result in inefficiencies, for example, due to favoritism of public buyers towards firms from their own country (home bias). Home bias can emerge when buyers care not only about the purchase itself, but also about secondary goals, such as boosting local jobs or pursuing political goals. In the EU, persistent home bias also undermines the policy goal of completing the “Single Market”. This is in spite of regulation aimed at promoting competition and a level playing field. Instead, most contracts are awarded nationally, fragmenting public procurement markets along national borders. In this paper, I study home bias using the impact of the early Covid-19 pandemic. Two factors have large effects on home bias in procurement in an emergency: crisis urgency and buyer discretion. These effects are not explained by potential supply-side limitations that may have pushed buyers to purchase abroad. Using the natural experiment that the pandemic poses for certain regions and certain product categories, I draw lessons for procurement beyond the context of the pandemic and argue that the effective lever to reduce home bias is in buyer incentives. However, given that the costs of the pandemic are estimated anywhere between 12.5 trillion USD globally through 2024 (by the IMF)2 to as much as 16 trillion USD for the US alone (Cutler and Summers, 2020), also avoiding misallocation in the context of emergencies is an important application of these results. In 2020, the Covid-19 pandemic shocked the global economy on many levels, impacting international supply chains and firm activities. Changing market conditions overturned the conventional wisdom of the economics of procurement. As the crisis intensified, rules which prioritize transparency and competition were abandoned. In some places, public authorities found themselves competing against private sector buyers and each other. In the EU, the European Commission eventually dropped publication and transparency requirements and gave buyers full discretion in their purchasing decisions for certain goods. The focus of the analysis is on this deregulation, together with the effect of crisis urgency. Potential misallocation in the procurement for medical supplies has come under scrutiny and has garnered high levels of public attention. Anti-corruption activists caution against the risk that vested private interests capture public resources and distort decisions as procurement is highly vulnerable to corruption.3 Media reports uncover problems ranging from low-quality products in Austria4 and Switzerland5 , over failures to organize distribution6 and payment of deliveries7 to even major irregularities in contract awards due to political influence-taking in Germany8 and Finland9 . Examples from other countries abound, making it clear that misallocation and inefficiencies did not fully disappear in this crisis situation. This article contributes to the study of home bias and cross-border procurement: I create a novel data set of procurement contracts for medical supplies. I analyze tender documents published between 2018 and 2020 in Tenders Electronic Daily (TED), an online register of European procurement contracts. The descriptive statistics show that a temporary and large surge in cross-border procurement during the first wave of infections coincides with a period of few competitive tenders and many direct awards. This surge of direct cross-border awards already suggests the presence of a phenomenon other than mutual unawareness of international buyers and sellers that explains home bias. Among the potential reasons for home bias, the shock provided by the pandemic illustrates two in particular, crisis urgency and buyer discretion. We use two distinct analyses: For crisis urgency, I study the effect of local infection rates on cross-border procurement on a sample of Covid-19-related medical products. This analysis finds that an increase in infection rates by one standard deviation increases the likelihood of a cross-border award for medical supplies by more than 19 percentage points (over a pre-pandemic baseline of 1.5 percent). A country-by-country leave-one-out analysis finds a still sizeable lower bound on the effect size of 9.3 percentage points. Then, I analyze the lifting of restrictions on buyers purchasing these Covid-19-related medical supplies. Under the new rules, buyers purchased directly from sellers rather than posting calls for tender. To estimate the effect of this regulation change, I draw upon additional data of procurement contracts of medical supplies which were not affected by these new rules but still similar in product group and therefore, presumably, similar in their supply characteristics. I compare the share of cross-border procurement for these supplies with medical supplies from related product categories that were unaffected by deregulation using additional data. Following this increase in buyer discretion, the probability of a cross-border award increased by 35.7 percentage points for affected products and services (with a lower bound of 16 percentage points in a country-by-country leave-one-out analysis). Informed by a simple model of monitoring, I suggest that the observed changes in buyer behavior are caused by the changed incentives and informational advantages of the buyers. Importantly, the empirical results persist in an analysis on a subsample of contracts where there is domestic spare capacity, conservatively approximated as the extensive margin of supply (additional domestic firms standing ready to compete for a contract but not winning it). Therefore, a lack of domestic production capacity alone does not explain the surge in cross-border awards during the pandemic. At the same time, this implies that the low share of cross-border awards outside of the emergency is not driven by a lack of competitive foreign sellers, for example due to potential transportation cost differentials or frictions related to regulatory, language or cultural barriers. I conclude that home bias can subside temporarily in an emergency. The empirical findings then vindicate the crisis response of lifting restrictions. However, a return to pre-pandemic levels of cross-border procurement in late 2020 shows that both effects were just transitory. Overcoming home bias in procurement therefore requires more permanent policies. If regulation to limit buyer discretion has the straightforward effect of lowering this monitoring cost, one would expect regulation to decrease misallocation. However, the above-mentioned scrutiny and public attention on procurement of medical supplies during the pandemic might have worked in the opposite direction by disciplining buyer behavior. The rest of the paper is organized as follows: In Section 2, I review the literature on misallocation in procurement. In Section 3, I present the data set. I outline the empirical strategy and regression results in Section 4. Section 5 concludes. 2 Literature This paper relates most closely to the empirical literature on the effect of procurement design, in particular buyer discretion, on various outcomes of the procurement process. Coviello et al. (2017) study procurement in Italy with 2000-2005 data and find that increased buyer discretion on average improves the functioning of the procurement system. In contrast, Baltrunaite et al. (2021) find that an increase in bureaucrat discretion in Italy increased the rents of politically well-connected and less efficient firms in data from 2009-2013. This paper is the first to study the economic channels of crisis urgency and increase of buyer discretion on cross-border awards. This analysis is not limited to contracts around specific thresholds and contributes a broad, European overview. An older strand of the literature studies how misallocation in procurement can occur due to favoritism in auctions (McAfee, McMillan, 1989, Laffont, Tirole, 1991), outright corruption (Burguet, 2017), or “buy national” policies (Cernat and Kutlina-Dimitrova, 2015). Allocation of procurement contracts in the EU is heavily skewed towards domestic firms (Vagstad, 1995), indicating strong discrimination against foreign firms. Based on an empirical analysis of procurement in 29 OECD countries, Hessami (2014) documents misallocation in procurement due to political corruption and rent-seeking behavior in OECD countries. I call this skew towards domestic firms “home bias”, which need not arise from corruption, but which can also be a deliberate policy choice.10 Procurement processes can encourage home bias to promote small-and-medium-sized firms, jobs, growth, or innovation. Dynamic considerations can justify home bias economically (Barbosa and Boyer, 2021). Laffont and Tirole (1991) suggest that the share of cross-border awards can serve as a simple screen for favoritism. They specifically analyze collusion of domestic buyers and firms against the government and conclude that under some conditions a “law of large numbers” should hold for awards to the disfavored group. I study this outcome empirically in the context of European medical supplies. In spite of its richness and accessibility, few articles have used procurement data published on TED. Prier et al. (2018) describe a consolidated data set published by TED for the years 2009-2015. La Cour and Olykke (2018) find that data completeness differs depending on the country submitting tender information. La Cour and Milhøj (2013) attempt a mainly statistical exploration of the data on Danish contracts. Similar to this paper, Kutlina-Dimitrova and Lakatos (2016) use data for cross-border contract awards in Europe and argue that strong product market regulation may act as a hidden anticompetitive barrier. Carboni et al. (2018) also discuss empirical methods to study discrimination of foreign firms in procurement.11 I discuss the issues raised by previous scholars with regards to the present data set for sample selection and identification. Although missing values are common in the data set, they seem uncorrelated with the award of contracts to foreign or domestic firms. This is the first paper to my knowledge that uses TED data to investigate procurement in the context of the Covid-19 pandemic and creates an original data set from individual contract award notices, rather than using the consolidated data sets which are only published with a lag of several years. So far, the effect of Covid-19 on procurement has been studied by legal scholars (Lalliot, Yukins, 2020, Sanchez-Graells, 2020) while Hoekman et al. (2021) focus on the trade dimension and the implications for public-private partnerships in the global medical industry (Casady, Baxter, 2022, Vecchi, Cusumano, Boyer, et al., 2020). These early responses to the pandemic lay out the policy challenges. By studying empirically the outcomes of procurement for medical supplies during the Covid-19 pandemic, this paper helps to quantify these concerns and to draw lessons for the future. 3 Data This section provides a brief summary of each each data set. Additional details can be found in Appendix Appendix A. 3.1 Data sets used Procurement data: Contract award notices (CAN) for public procurement 2018-2020 in Europe come from “TED: tenders electronic daily, Supplement to the Official Journal of the EU” (TED).12 According to its website, “TED publishes 746 thousand procurement award notices a year, including 235 thousand calls for tenders which are worth approximately € 545 billion.” Each CAN describes the outcome of a tender, including information on the buyer (name, type and location of the authority), the object (total value of the procurement, product category), and, where applicable, the division of the contract into individual lots. For each lot, the CAN reports the number of companies that bid for the contract (bidders), as well as the number of bidders that are foreign, non-EU, and small-and-medium-sized firms. The bidder winning the lot, the contractor, is listed with its location as well as an initial estimate and final reported value of the award. The unit of observation is a separate contract award to a contractor. I study the cross-section of awards as there are too few repeat observations for firms and buyers to create a panel. A contract award is defined by a contract date and a contractor and may represent an entire contract or just a contract lot. Covid-19 infection rates: Regional infection rates come from the European Centre for Disease Prevention and Control (ECDC).13 These are reported as average cases per 100,000 inhabitants over a 14-day period. Regions are listed by NUTS2-code. Population data: Population data at national and regional (by NUTS2-code) levels from Eurostat are used for a robustness check to compute average national infection rates (weighted by population) excluding individual regions. Exchange rates: Non-Euro currencies are converted to Euro using data from the European Central Bank (supplemented in a few cases by online sources described in Appendix Appendix A). 3.2 Sample selection Entries with incomplete data or token values are removed. Especially the largest countries often do not report the total values of procurement contracts or only report token values (such as 1 EUR) which are treated as missing. Of the 295 documents for medical supplies where no total value is reported (out of a total of 8,054 CAN that remain after dropping failed tenders from the original 9,322 CAN), 75 percent have buyers from Germany, France, and the UK. This is consistent with differences in reporting described by previous researchers (see Section 2). The sample is further restricted to contract awards by national governments and public authorities (excluding EU bodies) from 2018 - 2020 with contract values reported. Thus, I obtain a regression data set of 67,638 observations of individual awards for medical supplies.14 These contracts were designated by TED as related to the Covid-19 pandemic and are the subject of a comparison of cross-border procurement across regions by their infection rates. For the analysis of the deregulation, I identify another 252,575 observations which are used as the control group in Section 4.3. These products are identified from belonging to the same larger product groups and are mostly medical in nature as well. They were, in contrast with the previous group, not identified as Covid-relevant. For example, the former data set includes “antiseptics and disinfectants” which are Covid-relevant while, based on the CPV hierarchy level, “antipsoriatics” are not Covid-related and enter the control group. Additional examples are listed in Table A.4 .Table 1 Descriptive statistics of tenders for Covid-19-related product groups, n = 67,638 Table 1 Mean Std. dev. Minimum Median Maximum Lot value (excluding VAT) in EUR 137,115.50 3,640,437.11 0.02 1,999.87 420,499,749 Indicator: domestic award 0.99 0.08 0 1 1 14-day average infection rate per 100 inhabitants 0.10 0.08 0 0 2.35 Infection rate at seller location 0.03 0.12 0 0 2.35 Total number of bidders 5.00 10.52 1 2 350 Share of foreign bidders 0.00 0.06 0 0 1 Table A1 Selection of product groups by CPV for treatment and control group Table A1Treatment Control 45215142 Intensive-care unit construction work 45215141 Operating theatre construction work, 45215143 Diagnostic screening room construction work, 45215144 Screening rooms construction work, 45215145 Fluoroscopy room construction work, 45215146 Pathology room construction work, 45215147 Forensic room construction work, 45215148 Catheter room construction work 35113400 Protective and safety clothing 35113100 Site-safety equipment, 35113200 Nuclear, biological, chemical and radiological protection equipment, 35113300 Safety installations 18143000 Protective gear 18143100 Work gloves, 18143200 Safety visors 18424300 Disposable gloves 18424400 Mittens, 18424500 Gauntlets 33141420 Surgical gloves 33141410 Wire cutter and bistoury 33157000 Gas-therapy and respiratory devices 33151000 Radiotherapy devices and supplies, 33152000 Incubators, 33153000 Lithotripter, 33154000 Mechanotherapy devices, 33155000 Physical therapy devices, 33156000 Psychology testing devices 33192120 Hospital beds 33192110 Orthopaedic beds, 33192130 Motorised beds, 33192140 Psychiatric couches, 33192150 Therapy beds, 33192160 Stretchers, 33192200 Medical tables, 33192300 Medical furniture except beds and tables, 33192400 Dental workstations, 33192500 Test tubes, 33192600 Lifting equipment for health care sector 33195110 Respiratory monitors 33195200 Central monitoring station 33670000 Medicinal products for the respiratory system 33610000 Medicinal products for the alimentary tract and metabolism, 33620000 Medicinal products for the blood, blood-forming organs and the cardiovacular system, 33640000 Medicinal products for the genitourinary system and hormones, 33650000 General anti-infectives for systemic use, vaccines, antineoplastic and immunodulating agents, 33660000 Medicinal products for the nervous system and sensory organs, 33680000 Pharmaceutical articles, 33690000 Various medicinal products 33363600 Antiseptics and disinfectants 33363100 Antifungals for dermatological use, 33363200 Emollients and protectives, 33363300 Antipsoriatics, 33363400 Antibiotics and chemotherapeutics for dermatological use, 33363500 Corticosteroids for dermatological use and dermatological preparations, 33363700 Anti-acne preparations, 33700000 Personal care products, 33900000 Post-mortem and mortuary equipment and supplies 39330000 Disinfection equipment 39310000 Catering equipment, 39340000 Gas network equipment, 39350000 Sewerage works equipment, 39360000 Sealing equipment, 39370000 Water installation 3.3 Descriptive statistics Table 1 presents summary statistics for Covid-19-related tenders. Further descriptive statistics, including on the control group data, are in Appendix Appendix B. The most important findings of this section are that the value of cross-border awards increases from 1.5% on average pre-pandemic to more than 50% between April and June 2020 and that a majority of tenders in April and May 2020 were direct awards to foreign sellers. Spending patterns: The total spending on Covid-19-related medical supplies by country is summarized in Table B.6 . It is apparent that the amount procured by a country is not just a function of the size of its economy or population.15 The five countries that procured the greatest amount during the pandemic are the United Kingdom (UK), Ireland, Germany, Norway and Romania. In both tables, the UK represents a large amount of spending. This is due to some exceptionally large contracts for goods from China and Hong Kong. Most contracts are worth between 1,000 and 10,000 EUR (Figure B.3 ). The ten largest individual lots are reported in Table B.7 . As a result of the presence of outliers, and to prevent one country from impacting the estimation by too much, we include a robustness check by leaving one country out at a time.Table B2 Total value of individual contract awards published 2018-2020 by buyer country (EUR equivalent, excluding VAT) Table B2Country Total value Country Total value Country Total value United Kingdom 3,825,440,159.92 Hungary 98,826,759.14 Greece 5,806,930.67 Poland 1,172,433,382.38 Slovakia 72,286,834.36 Cyprus 5,735,658.40 Ireland 771,395,491.32 Belgium 65,088,034.40 Portugal 3,481,540.76 France 767,024,944.62 Slovenia 62,360,505.66 Iceland 475,862.22 Italy 517,696,183.13 Germany 60,982,075.65 Lithuania 336,143.81 Romania 423,224,861.66 Finland 51,526,898.00 Estonia 10,190.00 Austria 402,774,108.00 Norway 43,775,788.39 Denmark 336,247,169.13 Croatia 33,510,701.05 Spain 215,683,821.81 Bulgaria 30,046,518.24 Czech Republic 205,983,843.36 Netherlands 18,257,500.94 Sweden 109,544,231.72 Latvia 6,166,053.67 Sum of the individual contract awards by buyer country, published as “result[s] of tenders related to COVID-19” on a page with “COVID-19-related tenders” on https://ted.europa.eu/TED/. Fig. B1 Distribution of lot value for contract awards, Covid-19 related tenders in EUR (log scale) Fig. B1 Table B3 Largest cross-border purchases Table B3Buyer country Seller country Contract date Contract value (EUR equivalent) United Kingdom China 20/05/2020 346,725,371 Ireland China 03/04/2020 225,128,510 United Kingdom Hong Kong 28/05/2020 126,582,278 United Kingdom China 06/06/2020 110,173,256 United Kingdom United States 04/06/2020 91,642,825 United Kingdom Austria 30/05/2020 87,951,739 United Kingdom Hong Kong 04/05/2020 82,553,660 United Kingdom China 03/06/2020 81,705,908 United Kingdom China 21/05/2020 79,801,871 United Kingdom United States 04/06/2020 74,041,759 Finally, there are purchases completely outside the EU procurement system that do not appear in the data set. For example, the German “open house” purchase of face masks in March 2020 accepted face masks from all sellers who committed to a minimum delivery amount. According to media reports, circa a billion face masks at a unit price of 4.50 EUR were ordered.16 While the contract notice is posted on TED, no CAN exist. In contrast with traditional contract awards, all firms that fulfilled certain conditions could deliver goods at the posted price.17 Net change in cross-border awards: The outcome of interest is the share of contracts that were awarded cross-border. This is an indicator variable that takes the value 1 if buyer and seller are located in the same country and 0 if not. At the start of the pandemic, the value of cross-border awards increases drastically.18Figure 1a shows the average monthly share of domestic procurement. The overall level of domestic procurement before the pandemic is consistent with previous studies on procurement in Europe by Vagstad (1995).19Fig. 1 Fig. 1 Figure 1 a shows that in a sudden reversal, the volume of domestic procurement dips in April 2020. Less than 50% of contract volume were awarded to domestic companies between April and June 2020, reaching a low of 22.8% in May 2020. Purchases in several European countries, including France, the UK, and Italy, contribute to this dip. Countries that received a large number of cross-border orders include large non-European countries, including the US and China, and within Europe Switzerland and Austria. Overall, the value of domestic contract awards are 98.5% of the total in the pre-pandemic period from January 2018 to January 2020 but only 60.3% in the pandemic period from February to December 2020.Competition and direct awards: Competition for contract awards decreased at the start of the pandemic. Figure 1b describes the share of the total value that buyers awarded competitively in each month. I define competitive contract awards as having more than one bidder and not being categorized as “Contract awards without prior publication” or “Negotiated without a prior call for competition”. Buyers awarded over 90% of contract volume non-competitively in April and May 2020. Direct awards to foreign sellers represent 56.5% and 77.1% of the total observed contract volume in these two months. While the presence of several bidders does not guarantee a competitive auction, it is noteworthy that buyers placed many contracts directly with foreign sellers. This suggests that search frictions, such as a lack of awareness of buyers about foreign sellers or of firms about foreign procurement tenders, are unlikely to explain home bias before the pandemic. Explanatory variables: One of the main explanatory variables is the average 14-day infection rate per 100 inhabitants. The overall shape of the infection rate is plotted in Figure B.14 .20 This figure shows the average infection rate per month, weighting regions by their monthly contract value. This curve tracks closely the total European infection rate reported by the ECDC with a first wave in early 2020 and a much higher second wave in late 2020, as plotted in Figure B.15 . The ECDC figure contains weekly data separate by country. Although there are some differences, such as a slightly steeper drop in infection rates towards the end of the sample period, this comparison suggests that the weight of different European regions in the data set follows the general trend of the pandemic.Fig. B12 Monthly averages of the 14-day infection rate per 100k inhabitants in 2020 (weighted by purchase value) Fig. B12 Fig. B13 Weekly European infection rate according to ECDC, (c) 2021 European Centre for Disease Prevention and Control Fig. B13 The variable “Regulation change” is a dummy that takes the value 1 for contracts signed after April 1, 2020, the publication date of the “European Commission guidelines on procurement” and 0 before. It always takes the value 0 in the control group. The purpose of the change in regulation was to increase buyer discretion. I test the hypothesis that the publication of these guidelines, as buyers had more discretion in awarding contracts, changed the share of cross-border procurement. Further discussion on missing data and potential sample selection is in Appendix B.1. Procurement documents are sometimes incomplete, leading to missing values. These values are most likely not missing at random, yet a regression analysis shows that they likely just result in a downward bias of regression estimates, leaving us with a lower bound on possible effect sizes. Also, contracts below an administrative notification threshold don’t have to be reported. Contracts that should normally fall above a notification threshold can be split or shaded intentionally to fall below the threshold and go unreported. While I cannot exclude the possibility that this also affects tenders for medical supplies, they likely do not impact the estimates. Only systematic differences in contract size between times of high cross-border awards and other periods would be of concern. The main period of high cross-border awards, April to June 2020, does not appear to be anomalous from visual inspection. There is also no significant seasonal variation in the main outcome or independent variables. 3.4 Summary of the data Purchasing patterns across countries and product-groups are heterogeneous and lot values are highly dispersed. Beyond the documented differences in reporting, there are other unobserved differences between countries. For example, a country with a more technocratic public sector may have a higher base-line of cross-border awards and recalibrate less due to the pandemic.21 This motivates a leave-one-out robustness check by country. Descriptive statistics show that the net effect of the pandemic and deregulation was a temporary surge in cross-border procurement. Direct awards to foreign sellers represent a high share of the total observed contract volume, especially in April and May 2020. While the presence of many bidders does not guarantee a competitive auction, it is noteworthy that buyers placed most contracts directly with foreign sellers in these months. This suggests that there is no lack of mutual awareness of buyers and sellers across borders driving home bias in normal times. 4 Estimating the impact of the emergency and buyer discretion I estimate economically large effects of crisis urgency (measured through local infection rates) as well as increased buyer discretion (measured through a change in regulation) on the likelihood of cross-border procurement. First, I analyze medical supplies that are designated as Covid-19 related. Using local infection rates as a treatment for a staggered difference-in-difference analysis, there is an increase in the likelihood of a cross-border award by 19.3 percentage points following a one-standard deviation increase in infection rates. The contracts in the first analysis were all subject to deregulation that increased buyer discretion. For a second analysis, I draw upon additional data for similar products from the same product categories but for which regulations were not lifted. The likelihood of a cross-border award increased by 35.7 percentage points for products and services for which the regulation was lifted relative to the control group. However, each analysis finds a large increase in cross-border procurement compared to the pre-pandemic baseline when only 1.5 percent of the total value was awarded internationally. Even the lower bound effect sizes in a “leave-one-out” analysis remain economically very large, at 9.3 and 16 percentage points, respectively.22 4.1 Empirical strategy Ideal experiment: What would be the ideal experiment to causally identify the effect of increased buyer discretion and crisis urgency? For the former effect, consider a counterfactual Europe with identical infection rates but without the lifting of regulation. That setting would allow a direct causal interpretation of the regulation effect. It is apparent that the ideal experiment does not exist: non-European procurement markets, such as in Asia or the United States enacted their own regulatory responses to the pandemic. Both the pandemic and policy responses evolved at different speeds across the world. Alternatively, with complete information on the bidding history of all firms, one could analyze firms’ likelihood of winning bids (conditional on their bid and other bids) around the regulation change. This data is not available, as individual bids are not released, nor is the identity of non-winning bidders known. Likewise, a naive regression on infection rates would admit a causal interpretation only if infection rates were randomly assigned. This is clearly not the case: infection rates are endogenous, as they depend on previous infection rates, infection rates in neighboring regions, and factors that are correlated with factors such as the efficiency of the public sector or existing trade links. Changes in infection rates might not be random, as they might be influenced, e.g., government quality which might impact the likelihood of cross-border procurement and influence infection rates. Crisis urgency: Nonetheless, our setting provides a unique chance to identify the effects of crisis urgency and buyer discretion. The first analysis exploits the fact that the pandemic intensified in different locations at different times. Using only data set on Covid-19-related medical supplies, local infection rates act as a treatment in a staggered difference-in-differences setup, where higher levels of treatment imply greater volumes of cross-border procurement. To obtain an unbiased estimate of this treatment effect, we have to assume i) parallel trends and ii) the linearity of the infection effect. While we cannot test whether infection rates would have developed in parallel in different locations, we do compare regions that were treated earlier and later further below. As we have data on multiple locations and time periods, our data lends itself to a two-way fixed effects analysis with staggered treatment (Callaway, Sant’Anna, 2020, Athey, Imbens, 2021). A two-way fixed effects analysis generalises the canonical two-period difference-in-differences approach to multiple periods. The characteristic of the staggered treatment is that units are treated at different times. The unit that is treated is not the contract award (our unit of observation) itself, but the geographic region for which the contract is specified. The control group for treated units (regions) consists both of never-treated units and potentially also units that have been treated in the past. This setting also features varying treatment intensity as the treatment variable, infection rates, vary across locations and periods. Wooldridge (2021) shows that also with a staggered treatment, the two-way fixed effects estimator is appropriate to account for treatment intensity, covariates and interactions. I estimate the following baseline equation which permits an interpretation of the effect of infection rates on cross-border awards.(1) yit=αi+λt+βIit+β→Xit+ϵit αi denotes a contract’s performance location, λt is the week. Recall that the indicator variable yit takes the value 1 if the contractor is located in the same country as the buyer and 0 otherwise. Iit is the infection rate at the buyer’s location for observation i.23 Xi is a vector of control variables, the infection rate at the seller location, the share of foreign bidders, the total number of bidders, and dummies for country and product-group. Buyer discretion: By drawing upon additional data, the events surrounding the pandemic allow us also to study the effect of a deregulation act. We collect additional data on contracts for products which are in adjacent product groups relative to the previously analyzed ones. These contracts fall outside the classification as “Covid-19 related”. Between these contracts and the previously analyzed ones, a deregulation action on April 1st, 2020 provides a treatment that only affects the latter group but not the former, which acts as a control group of products with similar supply characteristics for which the procurement rules were unchanged. Identification of the effect of deregulation requires parallel trends between the control and treatment group. We show that trends in procurement for these products are indeed comparable to the treatment group. There are procedural restraints (open and competitive tenders, transparent award criteria, review by external authorities) which were lifted in an announcement by the European Commission on April 1st, 2020, two months after the start of the pandemic.24 After that date, buyers could award contracts directly and as fast as possible. The “Guidance from the European Commission on using the public procurement framework in the emergency situation related to the COVID-19 crisis” (2020/C 108 I/01) states:“for a situation such as the current COVID-19 crisis which presents an extreme and unforeseeable urgency, the EU directives do not contain procedural constraints. [...] [P]ublic buyers may negotiate directly with potential contractor(s) and there are no publication requirements, no time limits, no minimum number of candidates to be consulted, or other procedural requirements. No procedural steps are regulated at EU level. In practice, this means that authorities can act as quickly as is technically/physically feasible - and the procedure may constitute a de facto direct award only subject to physical/technical constraints related to the actual availability and speed of delivery.” Generally, this announcement applied only to tenders for certain goods and services that were seen as directly related to the pandemic, not to procurement overall. This communication from the European Commission was not a change in hard law. It merely clarified how to use the procurement framework. However, notes in procurement contracts refer explicitly to the use of expedited rules under the conditions of the pandemic.25 I estimate the following equation:(2) yit=τi+λt+βDit+β→Xit+ϵit Compared to equation 1, τi now denotes whether a contract awards belongs to the treatment or control group, the set of control variables does not include product group dummies (to avoid co-linearity with treatment group status), and Dit takes the value 1 for observations in the treatment group after the treatment date (April 1st, 2020). Again, a leave-one-out analysis by country allows to investigate the importance of outliers. 4.2 Emergency effect The first hypothesis is that greater levels of emergency lead to a greater share of cross-border awards. Measuring the degree of emergency implied by the pandemic through local infection rates, I investigate the impact of emergency on cross-border awards. If the absence of cross-border awards is associated with misallocation, cross-border procurement should increase when crisis urgency is greater. Consistent with the descriptive statistics from Section 3.3, any effect of infection rates on cross-border procurement is, if anything, transitory. The overall plummet in cross-border procurement coincides with the first wave of Covid-19 infections, but not the higher, second wave. This supports the interpretation of infection rates as a measure of crisis urgency which presumably played a bigger role when the world was initially confronted with the pandemic than in later stages of the pandemic. The present analysis differs from a standard setup of the staggered difference-in-difference estimator first because of the binary outcome variable. Both linear probability models and non-linear alternatives such as probit are available and have been used in these cases (Finkelstein, 2002) although the interpretation of the treatment effect in the non-linear model is not straightforward (Puhani, 2012). I do not repeat the exercise with a non-linear model as the focus is on the demonstration of an effect of considerable economic size, rather than a precise estimation of magnitude. Difference-in-differences diagnostics: First, I investigate the standard “parallel trends” assumption and potential spillover in treatment or outcomes. A standard approach to demonstrate parallel trends is to plot control and treatment group around the threshold and visually confirm that their trends are approximately parallel. Here, this is difficult due to the large number of regions with distinct infection profiles. Therefore, I aggregate regions into two groups depending on the date when infections were first recorded. Figures B.16 and B.17 in the Appendix compare the share of domestic purchases and average infection rates for these two aggregated groups. The former figure shows differences with two visible spikes in cross-border awards for the “early” group and a spike in-between for the “late” group. Looking at the flat section of the graphs in Figure B.17 until the beginning of April, it is also clear that both groups do not experience a fundamentally different development in the infection numbers for the first few weeks. This strengthens the case for areas that were infected a few weeks later as a control group for those that were infected earlier.Fig. B14 Domestic purchases by infection date Fig. B14 Fig. B15 Average infection rate per 100 inhabitants in regions infected early/late Fig. B15 Estimation results: The outcome of interest is the share of the value awarded abroad, so each observation is weighted by the value of the award. This avoids distortion from potentially arbitrary divisions of contracts into more lots. I use White’s heteroskedasticity-robust standard errors to account for the well-known fact that linear regressions in a binary outcome framework have heteroskedastic residuals. The results are summarized in Table 2 .Table 2 Difference-in-difference analysis with staggered treatment through infection rates Table 2 All obs. No UK Dep. var.: Contract awarded to domestic company ATET 14-day average infection rate -1.376* -0.463* (0.318) (0.127) Controls Infection rate at seller location 2.242* 0.529* (0.400) (0.183) Total number of bidders 0.002* 0.000 (0.001) (0.000) Share of foreign bidders -0.363* -0.428* (0.100) (0.061) Standard error of infection rate 0.14 0.20 1-std.dev. increase -0.193 -0.093 Dummies yes yes N 67,638 67,387 Robust standard errors in parentheses, * p<0.01. Infection rates are 14-day moving average per 100 inhabitants at the NUTS-region reported as performance location (buyer infection rate) or as location of the contractor (seller infection rate). Share of foreign bidders computed as number of foreign bidders from EU and non-EU countries divided by total number of bidders. The estimated average treatment effect on the treated (ATET) is for a unit increase of the 14-day average infection rate per 100 inhabitants. It is best interpreted relative to observed standard deviations during the pandemic. I compute the standard deviation of the infection rate for observations with non-zero infection rates.26 Column one includes all observations for medical supplies where the lot value is not missing. In the whole sample, the effect of a one-standard deviation increase in infection rates is an increase of 19.3 percentage points in the likelihood of a cross-border award. Due to heterogeneity between countries and the presence of outliers, I conduct a “leave-one-out” analysis in which I exclude one country at a time. For all of these tests, the effect remains statistically significant and when leaving out Czech Republic, it even increases by about 50% relative to the full sample. Unsurprisingly in light of the previously documented outliers from the UK, leaving out the UK diminishes the coefficient the most. Even this case, however, I find that a one-standard deviation increase in infections increases the likelihood of a cross-border awards by 9.3 percentage points (see column 2 of Table 2). Given that cross-border contracts only represent 1.5 percent of value pre-pandemic (see Figure ), even an increase by just over 9 percentage points is economically large. Spillover might occur if the treatment status of some units influences the treatment of other units. In the context of a pandemic, this is clearly a concern as the treatment literally spreads and may be transmitted by people moving between regions. However, whether this spread of infections also influences procurement decisions is less clear, as buyers should react to local infection rates independent of whether local infections arise because of movement between regions or infections within a region. More importantly, are buyers influenced by information from other regions when making purchasing decisions? In other words, do infections in other regions provide additional information beyond what local (lagged) infection rates predict? To test this, for each region I test for Granger-causality of infection rates against own lags and lags of other regions in the same country (see Appendix B.2). Allowing for up to six lags of the weekly updated infection data, I fail to reject the Null-hypothesis that the infection rates in one region are not Granger-caused by infection rates in other regions (see Table B.9 ).27 Table B5 Test for Granger-causality of local infection rates Table B5Dep. var.: Local infection rate Infection rate other regions -0.017 (0.017) Number of periods 349 Number of units 46 Standard errors allowing for cross-sectional heteroskedasticity in parenthesis. Up to nine lags of the explanatory variable tested, six lags chosen based on Bayesian Information Criterion. I also test for spillover in outcomes (Table B.10 in Appendix B.2): Does treatment status by one region impact outcomes in another region? Using data on population and infection rates, I compute for every region and every week the national infection rate excluding that region. In other words, it is a national measure of infection rates that excludes the contract location. Note that this measure is strongly correlated with local infection rates: this tells us that most regions at most times did not diverge too far from national trends in the infection rate. Still, even in the small subset of observations (842 observations representing ca. 45 million Euros worth of contracts) where local infections were below the overall median infection rate while national rates were above, there is no significant impact of the national infection rate on outcomes.Table B6 Test for Granger-causality of domestic awards Table B6Dep. var.: Domestic award Infection rate other regions 0.092 (0.085) Constant 0.990* (0.006) R-squared 0.001 N 842 Robust standard errors in parenthesis, * p<0.01. 4.3 Suspension of regulation As described in Section 4.1, the European Commission published guidance on using the procurement framework in which the normally very restrictive rules were widely suspended. Did the suspension of these rules lead to additional cross-border procurement? These suspensions increased buyer discretion, which previous scholars identified as a potential source of better procurement in some contexts and as a potential source of misallocation in others (Section 2). This hypothesis follows also from the theoretical model (Section Appendix D). The European Commission has confirmed that these rules were indeed applied restrictively to Covid-related tenders. I leverage the TED classification of Covid-related tenders to identify those products that fell under these rules. The control group includes all product groups that are at the same CPV hierarchy level as the Covid-related tenders. This yields a selection of closely related products. For example, from the category “beds for medical use” (33192100), the product group “hospital beds” (33192120) is Covid-related and all contract award notices 2018-2020 are part of the treatment group. Contract award notices for the remaining products in the category “beds for medical use”, which are “orthopaedic beds” (33192110), “motorised beds” (33192130), “psychiatric couches” (33192140), “therapy beds” (33192150), and “stretchers” (33192160), enter the control group. A complete list of relevant product groups is provided in Table A.4. The main assumption is that outcomes for products in the control group would have a similar potential outcome to the products in the treatment group. This could mean that for a contractor, there is a high degree of supply-side substitutability between, say, “intensive-care unit construction work” (treatment group) and “diagnostic screening room construction work” (control group). It could also mean that, even if different firms produce “protective gear” (treatment group) on the one hand and “work gloves” and “safety visors” (control group) on the other hand, changes in market conditions such as increasing input costs affect these firms similarly. Difference-in-differences diagnostics: The control group is based on product similarity in the product classification system.28 Figure 2 shows the average monthly rates of cross-border procurement for the products in the treatment and control group. This is unsurprising given the soft nature of the suspension of buyer discretion: instead of representing a change in “hard law”, it is a clarification on the interpretation of existing rules. Nonetheless, a clear impact of the regulation on the impacted goods, but not similar goods can be seen. At the same time, the graph shows broadly parallel trends pre-April 2020, save for a handful of spikes in individual months.Fig. 2 Average share of cross-border awards in the treatment group with greater buyer discretion and the control group Fig. 2 For most product groups, even in the control group the average monthly spending during the pandemic exceeds spending before the pandemic by a large amount. Only in two product groups (albeit including the largest by spending, “various medicinal products” alongside “antineoplastic agents”), spending is slightly lower during the pandemic. For most product groups both in the control and treatment group, average spending during the pandemic exceeds average spending before the pandemic, sometimes by an order of magnitude. This suggests that differences in cross-border procurement between goods in the treatment and control group are likely not driven by an uneven spike in demand that would only affect the treatment group. Estimation results: Estimation results are reported in Table 2. The baseline result on the full sample indicates that the increase in buyer discretion increased the share of cross-border awards for Covid-19 related tenders by circa 36 percentage points. I treat the smallest absolute regression coefficient in the leave-one-out analysis as a lower bound of this effect. This value is only 16 percentage points when the UK is removed, reflecting large cross-border contracts from the UK. The overall results are unsurprising, given the size of the drop in cross-border procurement, as well as the visually clear difference between the treatment group and the control group in Figure 2. 4.4 Robustness Potential challenges to the empirical strategy include the impact of the pandemic on the manufacturers of medical supplies, but also the measurement of buyer discretion and emergency. The pandemic has lead to disruptions for manufacturers as well as for distribution and international supply chains which could impact inference on awarding practice. For example, a decrease in contract awards to foreign companies could be related to border closures or export restrictions on medical supplies that were applied in 2020. All companies may face higher costs or lower production capacity due to sick workers, new safety measures, or general uncertainty arising from the circumstances of the pandemic and therefore decide to pursue fewer contracts or only bid on nearby (domestic) contracts. Depending on a firm’s industry, production technology and location, rising infection rates can imply• decreased production capacity, including factory shutdowns, due to worker sickness; • decreased labor productivity, or increased production cost, due to increased safety standards (e.g. encumbering safety clothing, regular disinfection protocols, maintaining physical distance between workers); • lack of physical access or uncertain access to raw materials due to interrupted supply chains; • lack of physical access or uncertain access to buyers due to border shutdowns and export restrictions. To rule out that the observed contracting practices are driven by these supply-side factors, company size would be available as a control. However, firm identity and size are not reported, only the total number of bidders is observed, and the numbers of bidders from other EU and non-EU countries. The control variables in the main regression, the number of firms bidding and the share of foreign bidders, already account for some of these supply-side effects. I also consider effects that would impact foreign awards even after controlling for participation. Capacity constraints on the supply-side: A buyer might turn to foreign sellers because domestic industries cannot supply the short-term need for all goods and services. Then, cross-border awards should increase during the pandemic even absent any home bias. However, even applying a conservative definition of spare capacity does not change the findings of the main analysis. Excess domestic capacity is present when there is a larger number of domestic bidders than domestic winners. For each contract, we subtract from the largest number of domestic bidders of any given lot the number of distinct domestic winners (counting over firm names after cleaning inconsistently spelled ones). This method likely underestimates domestic spare capacities because it only counts entire domestic companies that would stand ready to serve a contract, the extensive margin of domestic supply, while leaving unobserved intra-firm spare capacity, the intensive margin. This measure is also conservative because bidders are anonymous, so if different domestic firms bid for different lots on a contract, it is not possible to distinguish between them: the lot with the largest number of domestic firms is still just a lower bound of distinct domestic firms ready to supply a certain product at the time of contract award. A potential challenge for the analysis of infection rates as a measure of crisis urgency could be that the outcome variable impacts infection rates. That is, less misallocation and better procurement might, for some reason, directly result in lower infection rates. However, exploiting variation in infection rates within country-month-product group brackets means that any impact of procurement on infection rates would have to occur within the same month to create this kind of endogeneity. The effects of emergency and the regulation are robust to an alternative approach to measuring emergency that specifically circumvents such potential feedback effects: the infection rate in other regions of the same country (computed via population statistics and national infection rates, see Table C.11 ). However, these measures result in a noisier estimation than the main estimation, so I only consider them as robustness checks.Table C1 National infection rate (excluding local infections) as alternative explanatory variable Table C1 All countries Excluding UK Dep. var.: Contract awarded to domestic company ATET Infection rate other regions same country -1.793* -0.527* (0.405) (0.141) Controls Infection rate at seller location 2.050* 0.427 (0.398) (0.168) Total number of bidders 0.002* 0.000 (0.001) (0.000) Share of foreign bidders -0.361* -0.428* (0.099) (0.061) Dummies yes yes Standard error of infection rate 0.14 0.20 1-std.dev. increase -0.251 -0.105 N 67,638 67,387 Robust standard errors in parentheses, p<0.05, * p<0.01. The results are reported in Tables C.12 and C.13 . The difference-in-differences analysis by regional infection rates has now 23,443 observations (34.7% of the full sample). The effect of the infection rate remains significant and the effect size slightly increases in absolute terms to -19.7 percentage points for a one-standard deviation increase in infections. For the analysis on the regulation effect using a control group of additional products, I use the full control group. Here, the effect size is lower, at -8.8 percentage points post regulation. The coefficients are still significant but less so than in the main analysis. The leave-one-out robustness check does not change the results much, except that is pushes up the standard error for the regulation effect so that it is no longer significant at the 5% level. Overall, the analysis shows that a lack of domestic spare capacity is not a sufficient explanation for the surge in cross-border procurement. Large effects, especially from rising infection rates remain for contracts on which domestic suppliers bid.Table C2 Main regression Section 4.2 on a subsample with domestic spare capacity Table C2 All countries Excluding UK Dep. var.: Contract awarded to domestic company ATET 14-day average infection rate per 100 -0.896* -0.908* (0.252) (0.258) Controls Infection rate at seller location 1.212* 1.222* (0.405) (0.408) Total number of bidders 0.001 0.001 (0.000) (0.000) Share of foreign bidders 0.057 0.051 (0.107) (0.110) Dummies yes yes Standard error of infection rate 0.22 0.23 1-std. dev. increase -0.197 -0.209 N 23,443 23,294 Robust standard errors in parentheses, p<0.05, * p<0.01. Table C3 Main regression Section 4.3 on a subsample with domestic spare capacity Table C3 All countries Excluding UK Dep. var.: Contract awarded to domestic company ATET Regulation change -0.088 -0.089* (0.043) (0.052) Controls Infection rate at seller location 0.214 0.033 (0.137) (0.020) Total number of bidders -0.000 0.001* (0.000) (0.000) Share of foreign bidders -0.147* -0.545*** (0.027) (0.079) Dummies yes yes N 276,018 275,080 Robust standard errors in parentheses, * p<0.1, p<0.05, * p<0.01. Another supply-side concern could arise from changes in the pool of sellers. For example, the increase in domestically awarded contracts could be due to buyers using their increased discretion to purchase via resellers or trading companies. These companies may not qualify for procurement contracts for medical supplies under normal circumstances but might have privileged access to foreign manufacturers. Anecdotally, the German government called upon companies with trading links to China to purchase additional face masks around Easter 2020. While the data is not sufficiently rich to describe composition differences in their industries, I check that companies which have as name components “trade”, “logistic” or “distribution” are not more common among contractors during the pandemic than before and make up only a small portion of total sales, suggesting that this is not a large concern. A further robustness check is to ignore the lot value and instead count the number of lots awarded domestically or cross-broder. The results do not replicate when ignoring differences in value between individual lots. In this case there is no effect of the infection rate and a negative effect but very small effect of the regulation change when looking at all contracts, including those with missing lot values (see Table C.14 ). This is unsurprising, as this check is very much driven by two countries whose contract award notices are typically split into many, sometimes very small lots: Romania, representing 63% of all contract awards, and Poland, representing 20% of the absolute number of lots.29 Table C4 Unweighted regression for contract award to domestic companies Table C4 Infection rate Regulation change Dep. var.: Contract awarded to domestic company ATET Infection rate -0.012 (0.011) Regulation change -.004* (0.000) Controls Infection rate at seller location 0.029* 0.006 (0.011) (0.003) Total number of bidders 0.000 0.000* (0.000) (0.000) Share of foreign bidders -0.505* -0.597* (0.023) (0.011) Dummies yes yes Standard error of infection rate 0.22 1-std.dev. increase -0.003 N 120,922 540,413 Robust standard errors in parentheses, p<0.05, * p<0.01. Additionally, in a subset of countries with a series of failed procurement tenders during the pandemic, there is no evidence of any impact of failed procurement on contract awards (Table C.15 ). I also test whether correlation between procurement contracts within countries, for example due to similarity in regulatory and legal environments or similarities in training, doctrine, and perspective among public sector buyers, plays a role. I re-estimate the difference-in-differences analyses with country-clustered standard errors, as well as standard errors that are clustered by contract. Both types of clustered standard errors are moderately larger than White-robust standard errors (see Tables C.16 and C.17 ), but the estimates remain significant at conventional significance levels. Therefore, these robustness checks do not change the conclusions regarding the impact of crisis urgency and buyer discretion.Table C5 Including the cumulative sum of failed tenders Table C5 All countries Dep. var.: Contract awarded to domestic company ATET Infection rate -1.207* (0.300) Controls Cumulative sum of failed tenders 0.000 (0.000) Infection rate at seller location 2.346* (0.432) Total number of bidders 0.003* (0.001) Share of foreign bidders -0.614* (0.073) Standard error of infection rate 0.21 1-std. dev. increase -0.254 N 21,111 Robust standard errors in parentheses, * p<0.01. Table C6 Difference-in-difference analysis for infection rates with clustered standard errors Table C6 All countries Excluding UK Dep. var.: Contract awarded to domestic company ATET 14-day average infection rate per 100 -1.376* -0.463* (0.718) (0.166) Controls Infection rate at seller location 2.242 0.529 (1.398) (0.312) Total number of bidders 0.002 0.000 (0.001) (0.001) Share of foreign bidders -0.363 -0.428** (0.136) (0.189) Dummies yes yes N 67,638 67,387 Country-clustered standard errors in parentheses, * p<0.1, p<0.05, * p<0.01. Infection rates are 14-day moving average per 100 inhabitants at the NUTS-region reported as performance location (buyer infection rate) or as location of the contractor (seller infection rate). Share of foreign bidders computed as number of foreign bidders from EU and non-EU countries divided by total number of bidders. Table C7 Difference-in-difference analysis for the effect of deregulation with clustered standard errors Table C7 All countries Excluding UK Dep. var.: Contract awarded to domestic company ATET Regulation change -0.357* -0.160 (0.062) (0.074) Controls Infection rate at seller location 0.394 0.058 (0.319) (0.060) Total number of bidders 0.001* 0.001* (0.000) (0.001) Share of foreign bidders -0.030 -0.504*** (0.135) (0.146) Dummies yes yes N 320,213 319,173 Country-clustered standard errors in parentheses, * p<0.1, p<0.05, * p<0.01. Infection rates are 14-day moving average per 100 inhabitants at the NUTS-region reported as performance location (buyer infection rate) or as location of the contractor (seller infection rate). Share of foreign bidders computed as number of foreign bidders from EU and non-EU countries divided by total number of bidders. As mentioned above, differences across countries, for example differences in how procurement of medical supplies interacts with different health systems or in the quality of different public sectors, could impact results. For example, some countries might face lower base-lines of misallocation to begin with and consequently experience lower effects of the pandemic. While it is not possible to control for every conceivable difference between countries, in line with our goal of demonstrating the importance of both crisis urgency and increased buyer discretion, we account for these differences by including country-fixed effects, capturing different baselines, and with the “leave-one-out” analysis which tells us the lowest absolute effect size found among subsets in which one country is left out. These still continue to find economically large and statistically significant effects. Our estimation of these effects requires that trends in procurement awards would have been parallel across different regions independent of when they were treated with infections in the first analysis. The best argument for this is perhaps the long-lasting and consistently high share of domestic procurement for the product categories of interest which do not suggest that anything would have changed absent the pandemic. For the second analysis, a similar argument holds that we see persistently high shares of domestic awards throughout the pre-pandemic period for all goods, and substantial cross-border procurement only after the deregulation well into the pandemic. 4.5 Discussion The overall surge in cross-border awards is driven by two channels: both crisis urgency and increased buyer discretion lead to an increase in cross-border awards. This does not exclude the possibility that other channels may have an impact, too. At least the result on buyer discretion is surprising as buyer discretion may in principle also contribute to misallocation. The effects are statistically significant, even when accounting for across-country heterogeneity, and economically large: the increase in the likelihood of a cross-border award following a one-standard deviation increase in infection rates is over 10 times larger than the baseline rate, and over 20 times in the case of the deregulation. Interpreting this in light of the descriptive statistics and charts, the large drop in value awarded to domestic companies coincides with the first, lower wave of infections, and does not repeat itself when infection rates rise much higher during a second wave in late 2020. This suggests that any mitigating effect of urgency on home bias was at best temporary. While the absence of cross-border procurement is not direct evidence of misallocation, there are good reasons to presume that low shares of cross-border procurement indicate misallocation. Firstly, large variations in cross-border procurement as a result of a change in buyer discretion is inconsistent with a hypothesis of equal treatment. Secondly, a pure cost-based explanation could not explain the drastic surge in cross-border procurement from almost nothing pre-pandemic to more than 50 percent in some months of 2020. As argued above, the results are not driven by a lack of competitive foreign sellers. The magnitude of the observed effects is economically large relative to potential exclusionary effects of, say, fixed costs that might arise from translation of tender documents. I rationalize these findings using a game of monitoring that is presented in Appendix Appendix D to explain especially the puzzling effect of buyer discretion. In this game, a buyer chooses either selects the unconditionally best supplier or a favored supplier and the government can choose to make a costly review of that purchase. Emergency appears as the differential in the payoff that the government receives when a purchase is made either from the best supplier or the favored supplier. It is plausible that this differential is higher in an emergency due to the health threat posed by the pandemic where “procuring well” (choosing a supplier that aligns with tender requirements) is of the essence. Buyer discretion enters the model through the cost of monitoring. When regulation on buyer discretion decreases these costs and makes the collusive outcome less likely, the lifting of such regulation should result in more misallocation. There is evidence to the contrary in the empirical analysis. The conclusion within the proposed model is that in the context of the pandemic, the lifting of regulation does not increase the cost of monitoring collusion. Indeed, as the media reports cited in the introduction show, misallocation and collusive practices were under great scrutiny. While normally, procurement outcomes rarely make headlines, in the pandemic there was great public interest in the procurement of medical supplies. Therefore, in spite of the known instances of misallocation, overall buyers were put under more pressure to avoid misallocation, potentially reducing the cost of uncovering and fighting undesirable outcomes. This can explain the observed surge in cross-border procurement. 5 Conclusion Novel data set of procurement tenders in Europe allows us to study the natural experiment provided by the onset of the Covid-19 pandemic. A temporary surge in cross-border awards in 2020 was driven by suspended rules on buyer discretion and an emergency effect of the pandemic. The results of the empirical analysis suggest that buyer discretion and crisis urgency are important channels of procurement. Both channels exert an economically large impact on cross-border procurement, with effect sizes an order of magnitude greater than the baseline rate. Some open questions merit follow-up investigation. While this paper investigates the channels that affect home bias, the data is silent on contract performance indicators, such as cost overruns, delays, or terminations. Additional data could directly demonstrate the effect of misallocation both for economic and health outcomes. The paper also does not account explicitly for the role of trade policy instruments. An interesting subject for follow-up research would also be a network analysis of buyer-seller decisions following the pandemic to study the long-term effects on buyer-seller relationships in Europe. Beyond the context of the pandemic, we see that buyers can rapidly adjust their behavior in emergencies. In spite of decades of low shares of cross-border procurement, contradicting the EU policy goal of completing the “Single Market”, there was a surge of cross-border procurement at the onset of the pandemic. Both the local extent of emergency, measured through infection rates, and increasing buyers’ discretion, had contributed to this surge. This finding can be explained through a model in which the costs of monitoring buyers and the extent of their informational advantage impact misallocation. While the monitoring cost decreased due to heightened media attention, the informational advantage increased as procuring well for medical supplies is of heightened interest during a medical emergency. Also, in normal times as well as during emergencies, buyer incentives and regulatory constraints are important to foster EU Common Market policy. The prevalence of direct awards and non-competitive tenders in spring 2020 suggests that buyers and sellers are aware of each other. Search frictions, such as mutual unawareness of buyers and sellers, are then likely not an important causes of the low volume of cross-border contracts absent the pandemic. To the contrary, buyers seem capable of selecting foreign sellers if they wish to do so. This helps to rule out policies that seem well-intended but are likely ineffective to remedy home bias. For example, making it easier for sellers to enter their bids into foreign procurement tenders by reducing language barriers is likely not effective against home bias. This does not mean that existing regulation in this direction, such as Europe-wide publication requirements for large tenders, are not helpful for market integration. They might have built up existing mutual awareness of buyers and sellers that enabled cross-country procurement in the pandemic. For policy-evaluation, the analysis suggest that lifting the regulations was successful in creating a more integrated response to the pandemic that allowed the most affected regions to draw on the Single Market, in spite of the transitory nature of the effect. Economic and social policies during the pandemic have been critically examined and found lacking in some regards. In the context of procurement policy, this is likely unjustified. The main caveat is that the relationship between procurement outcomes and medical outcomes is unobserved. Yet, the procurement system at large proved capable of adjusting to the pandemic with an unprecedented increase in cross-border purchases. Appendix A Creation of the data set The data set is created from individual XML files related to Contract Award Notices published on the website “TED Tenders electronic daily - Supplement to the Official Journal of the European Union”.30 The selection of relevant procurement contracts is via the pre-selected filter that is offered on a “COVID-19 dedicated page for tenders related to medical equipment needs”.31 The default filter is for contracts published between February 1, 2020 and December 31, 2020. The descriptive statistics distinguish between pre- and post-Covid contracts using this datum, which is in line with the WHO designation (see footnote 23). I expand the default filters to include notices published as early as January 2018 to obtain a comparison period immediately preceding the pandemic. The contract date is given separately for the entire contract and its individual lots. In particular, the contract date for any individual lot, which is the unit of observation, may precede the publication date. The earliest lot in a sampled contract goes back as far as the year 2000. The filter also includes pre-selected codes from the Common Procurement Vocabulary (CPV).32 The sample includes 461 distinct CPV codes at different levels of granularity. The first five digits of a CPV code describe the category, while three additional digits provide greater detail about the products. Different individual products are aggregated by category, using the first 5 digits of the CPV code, resulting in 311 different product categories in the data set. I download a total of 9,233 contract award notices and voluntary ex-ante transparency notices.33 These documents are complex and rich in features, each describing a procurement tender, giving information on the buyer (name, type and location of the authority), the object (total value of the procurement and product categories, of which only the first, main product category is observed if there are several), and possibly division of the contract into individual lots. For each lot, the document reports the number of companies that bid for the contract, as well as the number of foreign, non-EU, and small and medium-sized firms as bidders. To create a data set at the level of individual contract awards, I process the files in two stages: first, I collect data that is common to all lots within one contract award notice, such as the date of publication, the location of the buyer, and the total value and currency of the procurement tender. In the second step, the section including the individual contract award is split at lot numbers to extract features that are specific to individual awards, such as the location and name of the contractor, the initially estimated and final value of the award, the contract date, as well as the number of bidders, further divided into bids from other EU countries, bids from outside the EU, and bids from small and medium-sized firms. These award features are then combined with the contract-specific information. The unit of observation is a separate contract award to a contractor. Each lot that is awarded to a single company is one observation. Where several contract awards are listed under one lot number, each contract award is also counted as one observation. Entries where no contractor name or contract date exist represent failed procurement tenders. In these cases, no contract was awarded. These observations are removed from the main analysis. I do, however, construct a weekly panel by country of the total amount of failed procurement at the start of each week, which I combine with the cross-section of contract awards for a robustness check (see Appendix Appendix C). In the remaining data set, there are 146 awards where no contract date has been listed. In these cases, contract date is approximated by publication date. The main outcome variable is an indicator that takes the value 1 if the country of the contractor is identical to the country of the buyer. Procurement by EU agencies is dropped because the outcome variable related to domestic firms does not apply to procurement by the EU itself. Thus, I obtain a total of 125,301 observations of individual awards, some of which do not enter the main regression analysis due to lack of data on contract values. This data is combined with monthly average exchange rates for all non-Euro currencies. Exchange rates are provided by the Statistical Data Warehouse of the European Central Bank, data set “EXR : Exchange Rates”, supplemented by individual values for the Macedonian Denar and Icelandic Kronur for six different months from “currencies.zone”, an exchange-rate information provider. All non-Euro values are then converted to Euro using the prevailing exchange rate. In a few cases, token values such as 0.01, 1, 99,999,999 or 999,999,999 appear in the data set. These are set to missing. I manually verified for the large values that these are indeed token or placeholder values, which is sometimes explicitly noted as a comment in the document or apparent from inconsistencies between total value and the value of awards. Some contracts don’t specify a total value of the procurement at the top of the contract, but instead report a lower and upper range. Missing values for the total contract value are imputed with midpoint between the higher and lower bound of this range. To be conservative and not introduce noise (e.g. from very imprecise ranges), I only do this in cases when the upper bound is less than 10 times larger than the lower bound, so the error is not greater than an order of magnitude. In any case, the total value of of the contract only enters descriptive statistics and a robustness check on the total amount of failed procurement. The main regressions use the value at the award level for which the value is always an exact amount. Next, I add data on infection rates from the European Centre for Disease Prevention and Control (ECDC) at a national and sub-national level.34 Infection rates are matched with the performance regions of the contracts. This is a measure the pandemic intensity at the location of interest to the buyer. Some data cleaning is required to deal with inconsistent entries (e.g. “CZ0” appears as well as “CZ”) and to match infection rates and contract location at the same level. Infection rates are available at different levels of aggregation depending on the country. Where only infection rates for the region at the NUTS3 level is available when contract information is at the NUTS2-level, the observation is matched with the first NUTS3-region in the data. In most cases, the location is reported at a more detailed level (NUTS3), which can be easily matched with a NUTS2 code through the first two numbers following the country code. In rare cases, the performance location is listed as a NUTS1 code, in which case it is matched to the infection rate of the first applicable NUTS2 code. When only the country is listed as location of performance, I use national data on the infection rate. I also use national data to complete entries before the 14-day sub-national average is started being reported. Dates before the first reported case in a country are assumed to have 0 infections. The procedure is repeated to match infection rates with seller location NUTS-codes and Covid-death rates for robustness checks reported in Appendix Appendix C. Finally, recent population data for the different NUTS codes from Eurostat is added to compute the average national infection rate excluding individual regions one at a time for a robustness check. The data for the control group of goods closely related to the Covid-19 related medical supplies are retrieved and treated similarly. The control group comprises all product groups of the CPV that are at the same hierarchy level as the Covid-related tenders. The CPV of Covid-19 related tenders are pre-defined by TED. Table A.4 shows a sample of the 969 CPV that were chosen to complement the treatment product groups: Appendix B Data description This section reports additional summary statistics. All sums are converted to Euros, where necessary, and exclude VAT. Table B.5 reports the summary statistics for the control group of the analysis of the regulation effect, that is, all those contracts that were in product groups adjacent to the Covid-19-related medical supplies for which procurement rules were lifted.Table B1 Descriptive statistics of tenders for product groups in the control group, n = 252,575 Table B1 Mean Std. dev. Minimum Median Maximum Lot value (excluding VAT) in EUR 162,292.00 7,076,075.20 0.07 1,380.92 1,990,000,000 Indicator: domestic award 0.99 0.07 0 1 1 14-day average infection rate per 100 inhabitants 0.02 0.10 0 0 1.87 Infection rate at seller location 0.02 0.10 0 0 1.87 Total number of bidders 4.18 7.26 1 3 350 Share of foreign bidders 0.00 0.06 0 0 1 Table B.6 reports the sum of lot values for all contracts in the data set (published between 2018 and 2020, but including some contracts with earlier contract dates) and thus represents the weights of different countries in the regression analyses. lot sizes. Table B.7 lists the 10 largest cross-border contract awards after February 2020. Except for one, they are all purchases from the UK. All of these purchases were under the CPV code for “Garments for biological or chemical protection” which encompasses the most common types of PPE, including face masks. These contracts range in value from almost 350 million Euros to just under 80 million Euros and most have been awarded to non-EU countries. This suggests that while cross-border procurement reached unprecedented levels in the pandemic, Covid-19 did not “integrate the Single Market”, but rather pushed buyers that previously purchased domestically to an international procurement strategy.35 To compare monthly spending by country and in the 20 largest product categories, Fig. B2, Fig. B3, Fig. B4 to Fig. B5, Fig. B6, Fig. B7 are reproduced for monthly values by dividing the pre-pandemic sum by 25 and the sum during the pandemic by 11, reflecting the period covered in the data set. Of interest is the observation that for some countries average spending was higher in the two years preceding the pandemic than in the first 11 months of the pandemic, including Germany, Croatia, and Slovenia. This comparison also shows that monthly average spending did remain approximately constant in some product groups, including construction (related to medical work), which might reflect long-term planning for some kinds of products and services. In most of the largest product groups (Figure B.10 ), average monthly spending during the pandemic exceeded average monthly spending before the pandemic by a wide margin.Fig. B2 Spending 2018-2020 on Covid-19- related product groups, top 14 countries Fig. B2 Fig. B3 Spending 2018-2020 on Covid-19- related product groups, other countries Fig. B3 Fig. B4 Spending 2018-2020 by product category, top 9 categories Fig. B4 Fig. B5 Spending 2018-2020 by product category, other categories Fig. B5 Fig. B6 Average monthly spending 2018 - 2020 by country on Covid-19-related products, top 14 spenders Fig. B6 Fig. B7 Average monthly spending 2018 - 2020 by country on Covid-19-related products, other spenders Fig. B7 Fig. B8 Average monthly spending 2018 - 2020 by Covid-19-related product category, top 9 categories Fig. B8 One possible concern is that the two sets of products are not comparable due to a surge in demand for the products in the treatment group, which might not carry over to the products in the control group. If this was the case, total purchases for products in the control group would have remained constant while purchases in the treatment group spiked. Average monthly purchases by product group are reported in Figures B.10 and B.11 . For comparison, the average monthly purchases of the 20 largest product groups in the control group are shown in Figures B.12 and B.13 .Fig. B9 Average monthly spending 2018 - 2020 by Covid-19-related product category, other categories Fig. B9 Fig. B10 Average monthly spending 2018 - 2020 by product category, top 9 categories (control group products Section 4.3) Fig. B10 Fig. B11 Average monthly spending 2018 - 2020 by product category, other categories (control group products Section 4.3) Fig. B11 The following weeks are weeks where some region reports its first infection: those starting on January 20, 27, February 3, 17, 24, and March 9, 16. Based on this, observations are categorized into those that experience their first infections either early or late. For easier visualization, I classify regions that first experienced infections in the weeks starting January 20 or 27 as “early” and the rest as “late”. Figure B.16 compares the evolution of the share of domestic purchases and the average infection rates. In both figures, two red vertical lines mark the weeks beginning January 20 and 27, as a marker of the “early infection” group. B1 Sample selection Procurement documents are sometimes incomplete, leading to missing values. These values are most likely not missing at random, yet a regression analysis shows that they likely just result in a downward bias of regression estimates, leaving us with a lower bound on possible effect sizes. Table B.8 presents a least-squares regression of a dummy variable that takes the value 1 if the value of a contract award is missing and 0 if not. The regressor is the main dependent variable, an indicator variable for contract award to a domestic company.Table B4 Likelihood of missing value conditional on domestic awards Table B4Dep. var.: Award value is missing Domestic award 0.013 0.084* (0.019) (0.019) Controls & dummies no yes N 121,120 120,962 Robust standard errors in parentheses, * p<0.01. Absent control variables, there is no strong correlation between domestically awarded contracts and missing information on contract value. Controlling for local infection rates for the buyer and seller, number of bidders, share of foreign bidders, as well as dummies for buyer country, product group and month-year, domestic contracts are on average circa eight percentage points more likely to not have information on award values. Contracts below an administrative notification threshold don’t have to be reported. Contracts that should normally fall above a notification threshold can be split or shaded intentionally to fall below the threshold and go unreported. While this may also affect tenders for medical supplies, they are unlikely to impact the estimates in the absence of systematic differences in contract size between times of high cross-border awards and other periods. Figure B.18 shows the share of contracts of different sizes in 2020. The categories follow the different thresholds that the EU uses to identify contracts presumed to be of cross-border interest.36 I aggregate contracts by size and plot their relative frequency by month in 2020.37 The main period of high cross-border awards, April to June 2020 (marked by a dashed box), is not obviously anomalous from visual inspection beyond usual fluctuations. There is also no significant seasonal variation in the main outcome or independent variables. Also, one would expect that contracts where misallocation plays a role to be more likely to go unreported. Suppose such contract awards are more likely to be concealed, for example to avoid scrutiny by oversight authorities. Then this would imply that cross-border awards are oversampled. However, since the baseline of cross-border awards is already very small (less than 1% as a simple share of all contracts), such an effect would likely be small as well.Fig. B16 Contract award sizes in 2020 in EUR Fig. B16 Expanding on the discussion of missing values in Section 3, the presence of selection bias and unreported values would suggest that the distribution of contract award sizes was highly abnormal in this time period. Although the share of contracts in these different size categories exhibits some fluctuation year-round, the average within each category seems to be not too far off for the three months within the box and the remaining nine months. An absence of large contracts in the period of high cross-border awards would be more concerning because this might suggest that large contracts would have been deliberately not reported or shaded to avoid exceeding reporting thresholds. Visual inspection of Figure B.18 does not suggest lower reporting of large contracts in the period of large cross-border awards. B2 Granger causality and spillover This Section contains the tables referenced in Section 4.2 to test for the information value of infection rates in other regions and possible spillover effects. Allowing for up to six lags of the weekly updated infection data, I fail to reject the Null-hypothesis that the infection rates in one region are not Granger-caused by infection rates in other regions. Appendix C Robustness This section includes robustness checks to the regression analysis in Section 4. Alternative emergency variables The claim of the infection rates as a measure for crisis emergency, or urgency, should persist for alternative measures. In particular, one might worry about possible feedback effects from procurement on local infection rates. To alleviate such concerns, an alternative measure of crisis urgency is used to address feedback effects from procurement on infection rates. I compute the infection rate in all other regions of a country except for the one of the performance location of the contract. This helps excluding effects of local procurement outcomes on the infection rate. This measure is computed from the absolute number of infections in each country and the region indicated in the contract award using population and the infection rate, then computing the difference between the two, and dividing by country population minus regional population to obtain the average infection rate in the country excluding the region of the observation. This measure helps us capture the common trend of infections within a country while excluding potential local feedback effects. Additionally, spillover effects are studied separately in Section 4.2 with the result that other regions of a country do not provide additional information about future infections over local past infections. Using this alternative variable in the regression analysis (Table C.11) yields find large and statistically significant effects for the regulation change. They are of similar quantitative size both for the full sample and the leave-one-out analysis where again leaving out the UK provides a lower bound of the effect size: In the full sample (sample excluding UK), a one-standard-deviation increase in national infections excluding the local region, results in a 25.1 (10.5) percentage points higher share of cross-border procurement. Domestic capacity constraints The results are robust even when accounting for buyers that had to turn to foreign sellers because of domestic capacity constraints. Consider a subsample of contracts for which the greatest number of domestic bidders for a contract (across all lots) is greater than the number of distinct domestic winners for lots on that contract. In this subsample, we identify the extensive margin of domestic spare capacity for these contracts. As discussed in Section 4.4, this method is likely conservative for domestic spare capacity. The results are reported in Tables C.12 and C.13. The effect of a one standard deviation in the average infection rate at the buyer’s location is approximately the same, at circa 20 percentage points. The effect of regulation is mitigated to circa 9 percentage points, compared to 35 percentage points in the baseline regression. Still, the effect size and statistical significance of the main variables remains high. Domestic capacity constraints were certainly important during the pandemic (affecting up to almost two-thirds of contracts) but do not suffice to explain the movements in cross-border procurement in the pandemic. Unweighted regression I re-run the first difference-in-difference regression, but do not weight contracts by value. The results are reported in Table C.14. I also use contracts for which no value is available, leading to a greater number of observations. However, a contract split into multiple small lots (observations) now carries much greater weight than a single, large contract compared to the main regression. As there is no significant variation of the outcome variable at this level (documented in Section 3.3), I do not expect to find any correlation with the explanatory variables. Indeed, due to the great differences in award values, as well as the more significant variation of the outcome variables, the results are null. Although ignoring contract size allows us to use more observations, no meaningful analysis with regards to the outcome of interest is possible. Effects disappear (or become economically insignificant in the second column) when attempting to estimate this regression without taking into account the differences in lot size through their monetary value. Analysis of failed procurement contracts Information about previously failed procurement contracts is potentially useful. For example, repeatedly failing to purchase desired goods or services may induce a buyer that discriminates against foreign sellers to reduce the extent of discrimination to avoid failed tenders in the future. I analyze the total contract value of failed tenders (i.e., where no contractor and contract date are given) and create a weekly time series of the cumulative failed value of procurement for every country, starting in February 2020, thus restricting this analysis only to the time period that is indeed related to the Covid-19 pandemic. I add the Euro-equivalent value of total failed procurement at the time of each award and a linear time-trend as explanatory variables. The results are reported in Table C.15. There is no statistically significant effect of failed previous procurement, nor does inclusion of this variable change the estimated effect drastically. Inference based on clustered standard errors Errors may be correlated across countries due to similarity in regulatory and legal environments and similarities in training, doctrine, and perspective among public sector buyers within each country. I re-estimate the difference-in-difference regressions in Tables 2 and 3 using clustered standard errors with clustering at the country-level. Standard errors are indeed larger with clustering, although only moderately. Only for the full-sample regression regarding infection rates are the results no longer statistically significant at the 5%-level. In the remaining regressions, the precision of the estimates remains high relative to the point estimates.Table 3 Difference-in-difference analysis with targeted deregulation Table 3 All obs. No UK Dep. var.: Contract awarded to domestic company ATET Regulation change -0.357* -0.160* (0.055) (0.061) Controls Infection rate at seller location 0.394*** 0.058* (0.121) (0.033) Total number of bidders 0.001 0.001* (0.000) (0.000) Share of foreign bidders -0.030 -0.504* (0.036) (0.075) dummies yes yes N 320,213 319,173 Robust standard errors in parentheses, * p<0.1, p<0.05, * p<0.01. Infection rates are 14-day moving average per 100 inhabitants at the location of the contractor (seller infection rate). Share of foreign bidders computed as number of foreign bidders from EU and non-EU countries divided by total number of bidders. Appendix D Model This section briefly illustrates the impact of urgency and buyer discretion in a simple model of monitoring. Even though this setting is simplified, it is informative to predict the presence of misallocation in procurement. The key comparative statics include a government’s cost of reviewing and punishing collusion and the importance of a procurement agent’s information. The former may represent the time and effort that needs to be expended to review a tender decision, litigate an outcome, and enforce a compensation for foregone benefits. Regulation that limits the discretion of buyers can be understood to directly reduce these costs. For example, when a regulator limits buyer discretion by mandating transparent tenders, prescribing scoring rules, and making the bidding process more transparent, he effectively reduces the cost of verifying violations of such rules. Think of the value of the agent’s information as increasing during an emergency. In particular, consider the multitude of policy objectives regarding procurement, some of which are completely unrelated to each tender (such as promoting jobs or small businesses). I assume that an increase in the urgency of a crisis makes objectives related to the tender relatively more important than secondary policy goals which may not require such information. For example, a good tender decision might require detailed information about the quality the seller, while simply spending money to secure local jobs does not. So, an increase in the urgency of a crisis should increase the importance of the bureaucrat’s information. From this model, I predict that an increase in crisis urgency should result in less misallocation. It is inspired by the mechanism-design models based upon McAfee and McMillan (1989) and Laffont and Tirole (1991). These models concern optimal auction contracts in a principal-agent-firm framework where the agent can collude with the firm (as opposed to collusion among firms). Whereas Laffont and Tirole (1991) is concerned with collusion-proof auctions and how they differ from the first-best, other models generate collusion in equilibrium (Burguet, Che, 2004, Burguet, 2017), or study dynamics and steady-state levels of corruption (Menezes and Monteiro, 2006). Branco (1994) argues that foreign profits should not enter the utility function of the principal when the principal is a national government. In this case, favoritism may be an efficient outcome also for the principal, although it implies an international coordination problem between governments. This problem has been compared to a prisoner’s dilemma where governments “want domestic protectionism and foreign liberalization, but they may prefer mutual liberalization to mutual protectionism” (Rickard and Kono, 2014). Arozamena and Weinschelbaum (2011) offer a different view and suggest that when entry into the auction is important, the principal should not favor a subset of bidders even if it cares about their payoffs, but not about the payoffs of other bidders. However, these models are not straightforward to take to the data as they often consider different cases, such as verifiable and unverifiable information, often with different implications. This model considers an explanation that Laffont and Tirole (1991) suggest to explain the small fraction of cross-border procurement, a fact which they suggest as a screen for misallocation, but that to my knowledge has not received much attention in this literature: rather than designing the auction himself, the principal might have to contend with reviewing procurement decisions and punishing collusion ex-post. To be sure, Laffont and Tirole (1991) think of this as a reputation game: concerns about a reputation to “keep their mouths shut” might disincentivize firms that were unfairly treated to blow the whistle on a buyer who discriminates against them to avoid losing whatever rents they were left with. By contrast, I study a simple monitoring game between a bureaucrat who decides whether to bias his decision and a government that decides to review (and punish) without modeling firms explicitly. It suggests that the ease of monitoring the buyer and the size of his informational advantage are important channels in this context. D1 Setup Consider a simultaneous game of complete information between a bureaucrat and a government. For a procurement tender, the bureaucrat has to decide whether to bias his decision towards a domestic supplier (cheat, C) or not (NC). I call this “cheat” for brevity, but it represents home bias as a form of misallocation in the widest sense such that not the procurement outcome is prioritized, but some other objective of the bureaucrat, say, because of regulatory capture. If the bureaucrat plays C, this results in a low-quality product being delivered to the government from which the government obtains a payoff of S_, and the bureaucrat receives a private benefit, or bribe, B>0. If the bureaucrat doesn’t cheat, a high-quality product of value S¯>S_ is delivered to the government and the bureaucrat receives a payoff normalized to 0. The government decides to either review the tender (R) or not (NR). If the government plays R, the government imposes a cost of kG onto itself and kB onto the bureaucrat. In case of review and if the bureaucrat plays C, the bureaucrat compensates the government for the foregone quality S¯−S_=ΔS, effectively increasing government payoff to S¯ (before subtracting the review cost). This is summarized in Table D.18 which depicts a 2x2 matrix in which the government is the row player and receives the first payoff listed in each cell and the bureaucrat is the column player and receives the second payoff listed in each cell.Table D1 Review-and-collusion game between government and bureaucrat Table D1 Bureaucrat C NC Gov’t R S¯−kG, B−kB−ΔS S¯−kG, −kB NR S_, B S¯, 0 Assume that the following two conditions hold to ensure that the game is interesting: ΔS>kG ensures that the government has an interest in playing R if the bureaucrat plays C. If kG is too high, then the government will strictly prefer to play NR whatever the bureaucrat’s strategy. The other condition is ΔS>B, the difference in quality must be greater than the benefit, so the bureaucrat prefers NC when the government plays R. If this was not the case and C was the bureaucrat’s dominant strategy, the government’s best response would be R. These two conditions result in a well-known kind of monitoring game which has no Nash Equilibrium in pure strategies, but only in mixed strategies where the government plays R with some positive probability p and the bureaucrat plays C with some positive probability q. These probabilities p,q are chosen such that the other player is indifferent between his strategies, so:(D.1) p(B−kB−ΔS)+(1−p)B=p(−kB)+(1−p)·0 (D.2) q(S¯−kG)+(1−q)(S¯−kG)=qS_+(1−q)S¯ After some simplification, this is straightforward to solve for(D.3) p=BΔS (D.4) q=kGΔS As for comparative statics, it is easy to see that the equilibrium probability p of the government playing R (reviewing a tender) increases with the size of the benefit B and decreases with the difference in the two quality levels ΔS. The equilibrium probability q of the bureaucrat playing C (favoring domestic suppliers) increases with the cost of review and enforcement kG and decreases with the difference in the two quality levels ΔS. D2 Discussion of the results The comparative statics for equation D.4 are measurable within the empirical framework as was suggested in Section 4.5. The lifting of buyer discretion, which we analyze in Section 4, amounts to an increase in kG. This model predicts misallocation to become more frequent when kG increases. However, the effect found in the empirical analysis in the opposite direction. This suggests that in spite of the suspension of rules on buyer discretion, government review cost need not have increased. Perhaps the scrutiny of procurement of medical supplies was higher during the pandemic than before. The media reports cited in the introduction are an example of how procurement received unusual public attention. ΔS stands for the difference between the high quality and the low quality product. Rather than just narrowly describing product quality in the sense of vertically differentiated products, this quality parameter S has been interpreted and described, e.g., by Laffont and Tirole (1991), as the fit of a seller with tender requirements, or more broadly as the value of the information of an agency. This motivates the focus on procurement of medical supplies. In the context of an emergency, such as the Covid-19 pandemic, the value of information on the actual quality of the seller of medical supplies is increased. In Section 3, I measure urgency through the local Covid-19 infection rate, but also consider alternative urgency measures. The empirical analysis confirms the model prediction that an increase in the value of information should decrease misallocation (and also lead to fewer reviews which is not observable, however). What justifies the assumption that the quality differential is larger than the bribe? If this condition wasn’t true, then collusion would be efficient in the model in the sense that it maximizes total surplus. The idea behind the second condition is that the bribing firm cannot offer a bribe that is greater than ΔS which is motivated by the notion of bilateral interim efficiency developed in Laffont and Tirole (1991). This idea was developed for the case where an agency can collude symmetrically with two firms and information about firm quality is verifiable. Then under a bilateral interim efficient auction, there are no incentive-compatible side transfers between the bureaucrat and any firm, and no announcement strategy by the agency or that firm, given the auction which is designed by the principal in the Laffont-Tirole model when taking truth-telling by the other firm as given. Why is the bureaucrat capable of compensating for shortcomings in procurement tenders? While this question might arise of the agencies in the mechanism-design models, which are bureaucratic agencies that are merely paid to screen sellers on behalf of the principal in the spirit of Laffont and Tirole (1991), buyers in the empirical analysis typically purchase goods out of their own budgets rather than just running auctions for goods and services for which the government pays (for example, hospitals). The implications and comparative statics of the model will not change drastically when assuming imperfect recovery of foregone surplus (i.e., if the bureaucrat pays the government ΔS if caught cheating, but the government only receives a fraction λ∈(0,1) of that sum, q would increase to kgλΔS). Data availability Data will be made available on request. 2 Reuters, 20.01.2022: https://www.reuters.com/business/imf-sees-cost-covid-pandemic-rising-beyond-125-trillion-estimate-2022-01-20/ 3 Transparency International, 29.03.2021: https://www.transparency.org/en/blog/g20-italy-covid-19-recovery-corruption-priorities 4 Der Standard, 09.12.2020: https://www.derstandard.de/consent/tcf/story/2000122115865/mangelhafte-ffp2-masken-werden-zurueckgeholt-und-beschaffung-rechtlich-geprueft [in German] 5 blue News, 08.03.2021: https://www.bluewin.ch/de/news/international/armee-tauscht-schutzmasken-der-firma-emix-um-616790.html [in German] 6 Osnabrücker Zeitung, 05.03.2021: https://www.noz.de/lokales/osnabrueck/artikel/2246556/osnabrueck-loest-corona-materiallager-auf [in German] 7 Die Zeit 13.06.2020: https://www.zeit.de/politik/deutschland/2020-06/mundschutz-atemschutzmasken-lieferproblem-zahlungsverzug-coronavirus [in German] 8 Deutsche Welle, 08.03.2021: https://www.dw.com/en/german-lawmaker-resigns-over-face-mask-scandal/a-56798497 9 Politiikka, 09.04.2020: https://www.hs.fi/politiikka/art-2000006469197.html [in Finnish] 10 The European Commission claims that “the public sector can use procurement to boost jobs, growth and investment, and to create an economy that is more innovative, resource and energy efficient, and socially-inclusive” (https://ec.europa.eu/growth/single-market/public-procurement_en). Loader (2007) writes that “UK government and small firms believe that public sector procurement is a good way of helping to support small business”. See also Loader (2016) on favoring small and medium-sized firms in procurement. 11 This is also the subject of a study commissioned by the European Commission using TED data (Chur, 2011). 12 https://ted.europa.eu 13 https://qap.ecdc.europa.eu/public/extensions/COVID-19/COVID-19.html 14 The number of observations exceeds the number of CAN because one CAN may include several individual lots. 15 An economic explanation is that countries with a well-funded healthcare system may already possess large stocks of commodities such as disinfectant or PPE and need to procure less in an emergency than countries with an under-funded healthcare system. However, I do not observe such stocks and in any case this would only represent a part of the spending. The overall decisions to procure medical supplies also depend on the incentives of buyers (e.g. differences in the funding of the healthcare system between countries, Beveridge vs. Bismarck systems) and differences in capacity planning for goods such as intensive care units (ICU) or medical labs (the data set includes construction services for both ICU and medical labs). 16 See Tagesschau, 16.05.2021: https://www.tagesschau.de/investigativ/ndr-wdr/open-house-101.html [in German] 17 See https://ted.europa.eu/udl?uri=TED:NOTICE:147548-2020:TEXT:EN:HTML 18 I do not observe local subsidiaries of foreign firms or foreign value-added to contracts awarded to domestic firms, e.g., through traded inputs or sub-contracting. Accounting for these alternative channels would result in a larger value share to foreign companies. See e.g. Cernat and Kutlina-Dimitrova (2015). 19 Before the pandemic, buyers awarded over 90% of total contract value domestically. Over 99% of all contracts are awarded domestically, which implies that cross-border awards are of higher value on average. 20 Scaled down by a factor of 1,000 from the original data for readable regression coefficients. 21 I thank a referee for this suggestion. 22 I emphasize that the two analyses have different treatment and control groups and effects are not like-for-like or additive. 23 The 14-day average rate of infection per 100,000 inhabitants published by the ECDC is scaled up by a factor of 1000 to improve readability of the coefficient estimates. 24 The WHO declared the novel Coronavirus outbreak a public health emergency of international concern (PHEIC), WHO’s highest level of alarm, on January 30, 2020, see https://www.who.int/emergencies/diseases/novel-coronavirus-2019/interactive-timeline. 25 One example is contract award notice 600222-2020, a British purchase of PPE from Austria, concluded on May 30, 2020. It says “In March the NHS experienced severe shortages of PPE [...]. In these circumstances, a procurement following the usual timescales under the PCR 2015, including accelerated options, was impossible. PPE manufacturers and supply chains were under immediate and unprecedented global pressure to provide products. A delay in engaging with the market by running a usual procurement process ran the risk of failing to acquire the necessary stock of PPE equipment and presenting a significant risk to life. The Department is content the tests permitting use of the negotiated procedure without prior publication (Regulation 32(2)(c)) are met: 1) The purchasing of PPE was identified as strictly necessary to meet anticipated demand. 2) It is responding to Covid-19 immediately because of public health risks presenting a genuine emergency. 3) The events that led to the need for extreme urgency were unforeseeable: the Commission itself confirmed: ’The current coronavirus crisis presents an extreme and unforeseeable urgency - precisely for such a situation our European rules enable public buyers to buy within a matter of days, even hours, if necessary.’ 4) There was no time to run an accelerated procurement under the open, restricted or competitive procedures with negotiation that would secure products within the required timescales. 5) The situation is not attributable to the contracting authority: It has not done anything to cause or contribute to the need for extreme urgency.” 26 Otherwise the standard deviation of the infection rate would be artificially diminished by pre-pandemic observations. 27 I flexibly allow for up to nine lags and then choose the model based on Bayesian Information Criterion. Allowing for longer or shorter lag structures, e.g. six weeks or twelve weeks of past information, yields the same result. 28 A similar method has been used by Eurostat, the statistical agency of the EU, to analyze trends in international trade related to Covid. Based on the trade statistics in their Comext data base, they compare Covid-19 specific product groups with other, similar product groups which are selected by chapters of the “Harmonised System” product classification. See Eurostat, Archive:EU trade in COVID-19 related products (31.03.2021), last accessed 01.05.2023. 29 EU regulation suggests that contracts may be aggregated to achieve cost savings, while a division into smaller lots might be motivated by a desire to allow smaller companies to participate, for example in Directive 2014/24/EU, preamble, paragraph 59. However, these competing forces determining contract size and lot division are outside the scope of the paper. 30 https://ted.europa.eu 31 https://simap.ted.europa.eu/web/simap/covid-related-tenders 32 “The CPV establishes a single classification system for public procurement aimed at standardising the references used by contracting authorities and entities to describe the subject of procurement contracts.” https://simap.ted.europa.eu/web/simap/cpv. 33 The latter, which only make up 0.1% of documents and 0.3% of awarded value, do not enter the regression analysis, because they contain no information about the number of bidders or foreign bidders - they all refer to directly negotiated contract awards. They are considered in Figure 1b to assess the share of non-competitive tenders. 34 https://qap.ecdc.europa.eu/public/extensions/COVID-19/COVID-19.html#subnational-transmission-tab 35 Among the 50 largest lots, which includes contracts above 2.8 million Euros, there are also three French purchases of protective gear from Germany and Switzerland, one Czech purchase of medicinal products from Ireland, one Romanian purchase of medical breathing devices from Germany, three Italian purchase of medical consumables and garments from Singapore and China, while the remainder are UK purchases mostly again of garments for biological or chemical protection from mostly non-EU countries. 36 See the website of the European Commission under “Internal Market, Industry, Entrepreneurship and SMEs”, https://ec.europa.eu/growth/single-market/public-procurement/rules-implementation/thresholds_en. 37 The main value thresholds above which contracts have to be reported are 135,000 EUR for goods and services and 5,350,000 EUR for (subsidized) works contracts, which in the present context can apply to construction work. A few other thresholds may apply in special circumstances which is not identifiable from contract observables. See https://ec.europa.eu/growth/single-market/public-procurement/rules-implementation/thresholds_en ==== Refs References Arozamena L. Weinschelbaum F. On favoritism in auctions with entry Economics Letters 110 3 2011 265 267 10.1016/j.econlet.2010.11.036 https://www.sciencedirect.com/science/article/pii/S016517651000409X Athey S. Imbens G.W. Design-based analysis in Difference-In-Differences settings with staggered adoption Journal of Econometrics 2021 10.1016/j.jeconom.2020.10.012 https://www.sciencedirect.com/science/article/pii/S0304407621000488 Baltrunaite A. Giorgiantonio C. Mocetti S. Orlando T. Discretion and Supplier Selection in Public Procurement The Journal of Law, Economics, and Organization 37 1 2021 134 166 10.1093/jleo/ewaa009 Barbosa K. Boyer P.C. Discrimination in Dynamic Procurement Design with Learning-by-doing International Journal of Industrial Organization 77 2021 102754 10.1016/j.ijindorg.2021.102754 https://linkinghub.elsevier.com/retrieve/pii/S0167718721000473 Branco F. Favoring domestic firms in procurement contracts Journal of International Economics 37 1 1994 65 80 10.1016/0022-1996(94)90025-6 https://www.sciencedirect.com/science/article/pii/0022199694900256 Burguet R. Procurement Design with Corruption American Economic Journal: Microeconomics 9 2 2017 315 341 10.1257/mic.20150105 https://www.aeaweb.org/articles?id=10.1257/mic.20150105 Burguet R. Che Y.-K. Competitive Procurement with Corruption The RAND Journal of Economics 35 1 2004 50 68 10.2307/1593729 Publisher: [RAND Corporation, Wiley]. https://www.jstor.org/stable/1593729 Callaway B. Sant’Anna P.H.C. Difference-in-Differences with multiple time periods Journal of Econometrics 2020 10.1016/j.jeconom.2020.12.001 https://www.sciencedirect.com/science/article/pii/S0304407620303948 Carboni C. Iossa E. Mattera G. Barriers towards foreign firms in international public procurement markets: a review Economia e Politica Industriale 45 1 2018 85 107 10.1007/s40812-017-0089-2 Casady C.B. Baxter D. Procuring healthcare public-private partnerships (PPPs) through unsolicited proposals during the COVID-19 pandemic Journal of Public Procurement 22 1 2022 6 16 10.1108/JOPP-07-2020-0061 Cernat L. Kutlina-Dimitrova Z. International Public Procurement: From Scant Facts to Hard Data Robert Schuman Centre for Advanced Studies Research Paper 2015 10.2139/ssrn.2682582 https://papers.ssrn.com/abstract=2682582 Chur R. Cross-border procurement above EU thresholds Technical Report 2011 DG Internal Market and Services http://op.europa.eu/en/publication-detail/-/publication/0e081ac5-8929-458d-b078-a20676009324. Copenhagen Coviello D. Guglielmo A. Spagnolo G. The Effect of Discretion on Procurement Performance Management Science 64 2 2017 715 738 10.1287/mnsc.2016.2628 Publisher: INFORMS. https://pubsonline.informs.org/doi/abs/10.1287/mnsc.2016.2628 Cutler D.M. Summers L.H. The COVID-19 Pandemic and the $16 Trillion Virus JAMA 324 15 2020 1495 10.1001/jama.2020.19759 33044484 https://jamanetwork.com/journals/jama/fullarticle/2771764 Finkelstein A. The effect of tax subsidies to employer-provided supplementary health insurance: evidence from Canada Journal of Public Economics 84 3 2002 305 339 10.1016/S0047-2727(00)00155-9 https://linkinghub.elsevier.com/retrieve/pii/S0047272700001559 Hessami Z. Political corruption, public procurement, and budget composition: Theory and evidence from OECD countries European Journal of Political Economy 34 2014 372 389 10.1016/j.ejpoleco.2014.02.005 https://www.sciencedirect.com/science/article/pii/S017626801400024X Hoekman B. Shingal A. Eknath V. Ereshchenko V. COVID-19, public procurement regimes and trade policy The World Economy 2021 10.1111/twec.13118 _eprint: https://onlinelibrary.wiley.com/doi/abs/10.1111/twec.13118 Kutlina-Dimitrova Z. Lakatos C. Determinants of direct cross-border public procurement in EU Member States Review of World Economics 152 3 2016 501 528 10.1007/s10290-016-0251-3 La Cour L. Milhøj A. An explorative study of the contracts of the Tender Electronic Daily (TED) Symposion i anvendt statistik: 28.-29. januar 2013 2013 178 194 Copenhagen https://core.ac.uk/reader/17279010 La Cour L. Olykke G.S. Public Procurement in the EU: Another Explorative Study Symposium i anvendt statistik, Vol. 25 2018 Copenhagen Laffont J.-J. Tirole J. Auction design and favoritism International Journal of Industrial Organization 9 1 1991 9 42 https://www.sciencedirect.com/science/article/pii/0167718791900034 Lalliot L.F. Yukins C.R. COVID-19: Lessons learned in public procurement. Time for a new normal? Concurrences 3 2020 46 58 https://www.concurrences.com/en/review/issues/no-3-2020/articles/covid-19-lessons-learned-in-public-procurement-time-for-a-new-normal-95667-en Loader K. The Challenge of Competitive Procurement: Value for Money Versus Small Business Support Public Money & Management 27 5 2007 307 314 10.1111/j.1467-9302.2007.00601.x Publisher: Routledge _eprint: Loader K. Is local authority procurement supporting SMEs? An analysis of practice in English local authorities Local Government Studies 42 3 2016 464 484 10.1080/03003930.2016.1157068 Publisher: Routledge _eprint: McAfee R.P. McMillan J. Government procurement and international trade Journal of International Economics 26 3-4 1989 291 308 10.1016/0022-1996(89)90005-6 https://www.sciencedirect.com/science/article/pii/0022199689900056 Menezes F.M. Monteiro P.K. Corruption and auctions Journal of Mathematical Economics 42 1 2006 97 108 10.1016/j.jmateco.2005.04.002 https://www.sciencedirect.com/science/article/pii/S0304406805000698 Prier E. Prysmakova P. McCue C.P. Analysing the European Union’s Tenders Electronic Daily: possibilities and pitfalls International Journal of Procurement Management 11 6 2018 722 747 10.1504/IJPM.2018.095655 Publisher: Inderscience Publishers. Puhani P.A. The treatment effect, the cross difference, and the interaction term in nonlinear “difference-in-differences” models Economics Letters 115 1 2012 85 87 10.1016/j.econlet.2011.11.025 https://www.sciencedirect.com/science/article/pii/S0165176511004769 Rickard S.J. Kono D.Y. Think globally, buy locally: International agreements and government procurement The Review of International Organizations 9 3 2014 333 352 10.1007/s11558-013-9177-x http://link.springer.com/10.1007/s11558-013-9177-x Sanchez-Graells A. Procurement in the Time of COVID-19 Northern Ireland Legal Quarterly 71 1 2020 81 87 https://heinonline.org/HOL/Page?handle=hein.journals/nilq71&id=85&div=&collection= Vagstad S. Promoting fair competition in public procurement Journal of Public Economics 58 2 1995 283 307 10.1016/0047-2727(94)01472-Z https://www.sciencedirect.com/science/article/pii/004727279401472Z Vecchi V. Cusumano N. Boyer E.J. Medical Supply Acquisition in Italy and the United States in the Era of COVID-19: The Case for Strategic Procurement and Public-Private Partnerships: The American Review of Public Administration 50 6-7 2020 642 649 10.1177/0275074020942061 Publisher: SAGE PublicationsSage CA: Los Angeles, CA Wooldridge J.M. Two-Way Fixed Effects, the Two-Way Mundlak Regression, and Difference-in-Differences Estimators SSRN Electronic Journal 2021 10.2139/ssrn.3906345 https://www.ssrn.com/abstract=3906345
PMC010xxxxxx/PMC10287180.txt	==== Front Heliyon Heliyon Heliyon 2405-8440 The Authors. Published by Elsevier Ltd. S2405-8440(23)04785-0 10.1016/j.heliyon.2023.e17577 e17577 Article Stock market optimization amidst the COVID-19 pandemic: Technical analysis, K-means algorithm, and mean-variance model (TAKMV) approach Navarro Maricar M. abc Young Michael Nayat a Prasetyo Yogi Tri de∗ Taylar Jonathan V. f a School of Industrial Engineering and Engineering Management, Mapúa University, 658 Muralla St., Intramuros, Manila 1002, Philippines b School of Graduate Studies, Mapúa University, 658 Muralla St., Intramuros, Manila 1002, Philippines c Department of Industrial Engineering, Technological Institute of the Philippines Quezon City, 938 Aurora Blvd, Cubao, Quezon City 1109, Metro Manila, Philippines d International Bachelor Program in Engineering, Yuan Ze University, 135 Yuan-Tung Road, Chung-Li 32003, Taiwan e Department of Industrial Engineering and Management, Yuan Ze University, 135 Yuan-Tung Road, Chung-Li, 32003, Taiwan f Department of Computer Engineering, Technological Institute of the Philippines Quezon City, 938 Aurora Blvd, Cubao, Quezon City 1109, Metro Manila, Philippines ∗ Corresponding author. International Bachelor Program in Engineering, Yuan Ze University, 135 Yuan-Tung Road, Chung-Li 32003, Taiwan. 22 6 2023 22 6 2023 e1757710 2 2023 17 6 2023 21 6 2023 © 2023 The Authors. Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The Philippine stock market, just like most of its neighbors in the region, was seriously impacted by the global pandemic COVID-19. Investors remain hopeful while continuing to seek great ones in the damaged market. This paper developed a methodology for portfolio selection and optimization with the use of technical analysis, machine learning techniques, and portfolio optimization model. The combined methods of technical analysis, K-means clustering algorithm, and mean-variance portfolio optimization model will result in the development of the proposed TAKMV method. The study aims to integrate these three important analyses to identify portfolio investments. This paper uses the average annual risk and annual rate of return data for the years 2018 and 2020 to form the clusters and assessed the stocks that correspond to the investor's technical strategy such as Moving Average Convergence/Divergence (MACD) and Hybrid MACD with Arnaud Legoux Moving Average (ALMA). This paper solved the risk minimization problem on selected shares of the companies, based on the mean-variance portfolio optimization model. There are 230 and 239 companies for 2018 and 2020, respectively, listed in Philippine Stock Market, and all simulations were performed in MATLAB environment platform. Results showed that MACD strategy dominates the MACD-ALMA strategy in terms of the number of assets with a positive annual rate of return. The MACD works efficiently in the pre-COVID-19 condition while MACD-ALMA works efficiently during-COVID-19 condition, regardless of the number of assets with a positive annual rate of return. The results also show that the maximum expected portfolio return (RP) can be achieved using the MACD and MACD-ALMA in the pre-and during-COVID-19 conditions, respectively. The MACD-ALMA shows an advantage during high-risk market conditions and can also provide maximum RP. The performance of the TAKMV method was validated by applying its results and comparing it to the next year's historical price. The 2018 results were compared to 2019 data and the 2020 results were compared to 2021 data. For consistency, the comparison was applied to the same company per portfolio. Simulation results show that the MACD strategy is more effective compared to MACD-ALMA. Keywords Technical analysis Machine learning Optimization Stock market COVID-19 ==== Body pmc1 Introduction The worldwide stock market was globally affected by the COVID-19 pandemic. In most countries, stock markets were negatively influenced by the spread of the COVID-19 disease [[1], [2], [3], [4], [5], [6], [7], [8], [9]]. The Philippine Stock Market, also known as the Philippine Stock Exchange (PSE), has been significantly impacted by the COVID-19 pandemic. As with many stock markets around the world, the PSE has experienced volatility and uncertainty as investors react to the pandemic. In the early months of the pandemic, the PSE saw significant declines in stock prices as investors responded to concerns about the impact of the pandemic on the Philippine economy. The PSE Index, which tracks the performance of the largest companies listed on the exchange, fell to its lowest level in years. Despite the challenges posed by the pandemic, some investors have continued to invest in the Philippine Stock Market. Some have seen the pandemic as an opportunity to invest in companies that have the potential to thrive in the new normal, such as those in the technology and healthcare sectors. Investors in the Philippines have also had to adapt to changes brought about by the pandemic, such as the shift to remote work and digital channels. This has presented new challenges for investors who may be used to more traditional methods of investing. Investors remain hopeful about the Philippine Stock Market, while retail investors continue to seek great ones in the damaged market. These are generally young, tech-savvy millennials with a lot of money to trade online. Many big stock brokerage companies had to temporarily block new account openings or invest in additional bandwidth to accommodate the rush of online trades as a result of their rapid flood into the market. Retail investors' increased engagement is a strong sign that the local stock market still has plenty of chances for those willing to look beyond short-term gains. With increased knowledge of the local market and increased investment maturity, these investors have boosted their interest in listed shares, which allows the economy to recover more quickly from the pandemic's effect. Despite certain setbacks, like the continued rise in COVID-19 cases, which prompted the government to revert to more restrictive regulations in Metro Manila and neighboring provinces, the Philippine Stock Exchange Index (PSEi) is expected to have a stronger year in 2023 than it did in 2020–2022. In addition, the vaccine roll-out boosted the investor's confidence. In today's world of rapid technological progress, institutional and retail investors in the Philippines have a variety of choices, including online trading. These types of investors can benefit from the online trading platform, though they are not all the same, and there are several distinctions between institutional investors and non-institutional, or retail, investors. An organization or person that trades assets in huge amounts to be eligible for privileged dealing and less expensive fees is known as an institutional investor, they do not invest their funds; instead, they invest the funds of others on their behalf. A retail investor, on the other hand, is an individual or non-professional investor who buys and trades stocks through brokerage firms. They frequently invest in brokerage or retirement accounts for their benefit. There is a significant increase in retail investors who used online stock trading platforms in the Philippines during COVID-19. From a newbie investor who requires assistance in developing an investment strategy to a seasoned investor who can use an online trading platform to execute a strategy. In the Philippines, online stock trading platforms are widely used, making investment extremely convenient and accessible to the majority of Filipinos. It became the new normal's trend that supports the rapid spread of online investing and trading whether it was on mobile or web-based platforms. These platforms were targeted by institutional and retail investors and are widely accessible and given mostly by stock brokerage firms in the Philippines. In addition, these platforms were useful in assessing the stock performance of the top gainers and top losers for each day of individual stock performance. Institutional and retail investors can easily check their stock portfolio on these online trading platforms. The record of transactions is available online, and investors can buy and sell at any time based on their preference without having to depend on the broker which has complete control over the stock portfolio. A stock portfolio is a collection of equities in which investors and traders specifically invest intending to make a profit. To become a resilient investor, one should diversify his investments that span several sectors, for the reason that if one area suffers a setback, the investments in other industries aren't necessarily impacted. Diversification is an important part of building a portfolio, yet full diversification is impossible because investing in each company needs a significant amount of money. Apart from that, not all companies are worth investing in, which could be due to their illiquidity. As an institutional or retail investor, it is obvious to develop a successful strategy that will select stocks that will give higher returns and lower risk. Philippine Stock Market has faced challenges during this time, there are also opportunities for investors who are willing to navigate the new normal and identify companies with the potential to succeed in the post-pandemic world. Broad market pullbacks such as the COVID-19 pandemic situation obviously can suffocate the entire portfolio without any good strategy employed. In this study, we proposed a combined method of Technical Analysis, K-means Algorithm, and Mean-Variance Model named as TAKMV method. The diversification of this study will help to protect an investor's portfolio from the systematic risk that could expose the portfolio to losses. The TAKMV method was introduced to help investors, traders, managers, and decision-makers in analyzing the stock market and was used to identify possible portfolio investments. 2 Literature review Portfolio selection is the process of determining a mix of securities from a huge variety of choices. It is a method of allocating wealth to a group of assets to attain long-term objectives [10,11]. Its purpose is to maximize investment returns for investors. According to Markowitz [12], investors must choose between profit maximization and risk minimization. Risk minimization for a predetermined level of return or return maximization for a calculated risk level is two options available to investors. Markowitz calculated investment return as the expected value of securities' profits. The divergence from the expected value, according to Markowitz, is a risk. The Mean-Variance (MV) optimization method solves the portfolio problem by combining two basic indicators such as expected returns (represented by the mean return) and risk (measured by the volatility of the return). Understanding the behavior of financial asset prices and predicting its future behavior has always been a difficult task for practitioners. As a result, precisely anticipating price swings is critical for making gratifying and profitable investing decisions [13]. Portfolio selection has been extensively researched in a variety of fields, including quantitative and conventional finance, machine learning, and artificial intelligence [14]. Prior research has focused on financial factors on different Investment behavior [[15], [16], [17], [18], [19]]. Fundamental and technical analysis has traditionally been the two most extensively used ways of analyzing stock market data [20]. The concept of intrinsic value is used in fundamental analysis, which means that the present price is based on both qualitative and quantitative data. Fundamental analysis seeks to determine a company's fair market value by exploring all components of the firm, as well as the market, industry, domestic and large-scale environment. Whoever conducts the analysis should analyze the firm's overall performance and financial statements, as well as all recent corporate news. Investors should determine if the market correctly incorporated all relevant data into the stock price. The investor must analyze all aspects of financial statements, including profits, assets, revenues, and expenses, do a year-by-year comparison, compare to industry standards, observe specific tendencies in their behavior, and value the shares appropriately based on all of this. While recent studies proposed different algorithm models to analyze the performance of the selected portfolio [[21], [22], [23], [24], [25]] using Fundamental variables, it has many limitations and flaws in the search for future price forecasting, making it even more difficult to construct models that accurately describe stock variations. Because it relies on a deeper understanding of all aspects of a company and individual stocks, accumulating all of these qualities for a thorough company valuation can be time-consuming and expensive, making it impracticable in many cases, especially from a retail investor's perspective. Technical analysis, on the other hand, seeks to recognize a pattern in data, such as price movements and past returns that may be used to predict future price movement for securities and the market as a whole [26]. It is a well-known method that uses stock prices or technical indicators as inputs. Based on technical analysis, the stock price has already represented all of the major underlying variables, it focuses only on the stock price and trading volume, both of which are recorded and displayed in various tables and graphs. Investors can learn about specific trend forms and regularities in price movement, trading volume, and interdependence by evaluating these graphs. The high and low prices for each trading period, as well as the opening and closing prices, are plotted for each trading period. Moving averages are frequently used as technical analysis indicators in related literature such as Simple Moving Average [[27], [28], [29], [30]]; Weighted Moving Average [31]; Exponential Moving Average [28,32]. Some studies used Moving Average Convergence/Divergence (MACD) as a technical indicator with machine learning applications [27,29]. Fundamental analysis examines stocks over a longer period than a technical analysis which takes place over a short period, such as days, weeks, or months [33]. In the literature, variables such as macroeconomic, financial, and technical indicators have been examined as the most significant ones influencing stock price movements [34]. In most prediction studies, technical indicators play an essential role in buy and sell signals for stocks and are widely applied as input variables. Prior studies used technical analysis to examine stock behaviors [35,36]. While past research has focused on portfolio selection based on fundamental analysis, technical analysis should also be considered when using machine learning to select a portfolio. Clustering is an unsupervised data mining approach for grouping things based on their similarities. It's used to examine a variety of datasets. K-means is one of the most commonly used unsupervised clustering algorithms. However, it is difficult to determine the value of the k parameter, which represents the number of clusters, one of the most commonly used methods for determining the number of clusters is the cluster validity index. Numerous internal and external validity indexes are used to find suitable cluster numbers based on the characteristics of datasets [37]. In the stock market, Cluster analysis helps to distinguish stocks with different characteristics. Prior Studies include K-means on selecting stocks index, particularly in Asia. In the study of [38], they consider the data preprocessing using trimmed k-means clustering for robust mean-variance portfolio selection. The optimum portfolio is formed by selecting the stock representation for each cluster using the Sharpe ratio in the Indonesian Stock Exchange. Another paper studied the Fuzzy Time Series Markov Chain model [39] and offered a new strategy that combines a combination of absolute differences and k-means clustering. Based on the data it made the interval more flexible and compact. The Taiwan Capitalization Weighted Stock Index (TAIEX) was employed as a benchmark in their study. In addition [40], investigated the use of the K-means clustering method in stock forecasting, with the K-means algorithm being improved by incorporating the artificial fish swarm algorithm (AFSA) which is named KAFSA. Closing price, price-earning ratio, earnings per share, and return on net assets were used to verify the prediction results. The findings revealed that there were considerable variances between A and B equities when divided by KAFSA, with B stocks showing significantly larger discrepancies than A stocks. Moreover [41], proposed a model that takes into account heterogeneity and looks for homogeneous groups of enterprises with high importance. The multiple kernel learning technique and K-means clustering which is used to forecast stock price changes and incorporate information from the target company and its homogenous cluster. The experiment was conducted utilizing three years of data from the Republic of Korea. The results reveal that in the vast mainstream of circumstances, the suggested strategy outperforms current methods in terms of predictability. The findings also suggest that the need for cluster analysis is dependent on the sector's heterogeneity and that as the heterogeneity develops, it is necessary to undertake cluster analysis with a bigger number of clusters. Furthermore, the work of [42] proposed an analytical approach for determining thresholds and migration within clusters using simulations with k-means clustering and homogeneous Markov chains. Quarterly financial data from a sample of 35 public organizations from July 1, 2006 to March 28, 2020 (companies listed on the stock exchanges of the United States, Mexico, Brazil, and Chile) was used. Furthermore [43], established a six-step portfolio selection methodology in the Mongolian stock exchange, taking into account the economic, legal, political, and corporate governance implications. They used the methodology to acquire stock price information from companies in the TOP-20 index. Using the k-means method, grouping of share return, and risk evaluations, hierarchical clusters were created based on the correlation matrix of share return. They identified the most efficient portfolio and solved the return maximization objective on selected companies based on Markowitz's mean-variance model. Lastly, the work of [44] suggested a new indicator such as the Simple Moving Average of Price Change Ratios (SMA-PCR-N). It's an adjusted version of the standard Simple Moving Average (SMA). They show how to establish a diversified stock portfolio using k-means clustering and SMA-PCR-N. For the fiscal years 2015–2017, data on around 300 stocks were acquired from the Stock Exchange of Thailand and utilized in trials to evaluate the performance of their proposed approach. In most cases, portfolios built using SMA-PCR-N outperformed portfolios built with SMA clusters, according to the findings. Assessing the risk and returns of multiple firms can be difficult; grouping more than two hundred thirty equities in the Philippine stock market is nearly possible with the help of the clustering technique. Cluster analysis can help by aggregating returns and risk so the investor or trader can concentrate on each group rather than trying to make decisions based on individual stocks. This study focused on the PSE, which was founded in 1992. Using the Philippine Stock Exchange (PSE), portfolio selection, like any other decision-making problem, is influenced by a variety of factors, both directly and indirectly. In this regard, researchers, managers, investors, and practitioners have found it difficult to investigate, recognize, rank, and use criteria to analyze, select, and optimize portfolios. Therefore, this paper developed a methodology for portfolio selection and optimization with the use of technical analysis, machine learning techniques, and a portfolio optimization model. TAKMV methodology, a combined method of Technical Analysis, K-means clustering algorithm, and Mean-Variance portfolio optimization model was proposed in this paper. The study aims to integrate these three important analyses to come up with the best portfolio. This paper uses the average annual risk and an annual rate of return data for the years 2018 and 2020 to form the clusters and assessed the stocks that correspond to investors' technical strategies such as Moving Average Convergence/Divergence (MACD) and Hybrid MACD with Arnaud Legoux Moving Average (ALMA). In the empirical experiment, To select the efficient portfolio, this paper solved the risk minimization problem on selected shares of the companies, based on the mean-variance portfolio optimization model. There are 230 and 239 companies for 2018 and 2020, respectively, listed in Philippine Stock Market, and all simulations were performed in MATLAB environment platform. To the best of the authors' knowledge, this is the first study to use this TAKMV method in the Philippine Stock Market pre- and during COVID-19 conditions. This paper can help to improve people's understanding of machine learning and technical analysis, as well as dispel their misconceptions about the difficulty of analyzing the stock market in the Philippines. 3 Research methodology This paper consisted of five stages. Stage 1 constituted the data collection from Marketwatch [45]. Stage 2 was based on the investment strategy using a hybrid technical indicator: Moving Average Convergence/Divergence (MACD) and Arnaud Legoux Moving Average (ALMA). Stage 3 was stock clustering in the Philippine Stock Market based on average annual risk and an annual rate of return. Stage 4 evaluates stock average annual risks and returns according to clustered MACD or MACD-ALMA strategy. And lastly, Stage 5 identified the most efficient portfolio using the mean-variance portfolio optimization model. Combining three important analyses (Technical Analysis, K-Means Clustering, and Mean-Variance Model) will result in the proposed TAKMV methodology. Fig. 1 shows the flowchart of the proposed methodology.Fig. 1 Flowchart of the proposed TAKMV methodology. Fig. 1 3.1 Stage 1: data collection The data used in this study corresponded to the 2018 and 2020 historical prices of the Philippine Stock Market pre- and during-COVID-19 conditions. It composes of the Bank and Financial sector, Commercial and Industrial, Conglomerates, Consumer, Index, Insurance, mining and oil, Properties, Services, and Telecoms sectors with a total of 230 and 239 companies in the Philippines for 2018 (pre-COVID-19) and 2020 (during-COVID-19), respectively. To determine the performance of the model, it was then validated by applying the results and comparing it to 2019 (for 2018 results) and 2021 (for 2020 results) data. No validation was conducted between 2019 and 2020 since this was the transition of pre- and during-COVID-19 conditions. 3.2 Stage: 2: Technical analysis Technical Analysis indicators are linear functions that calculate recurrent values using historical trading data such as open, high, low, or close prices, volume, open interest, advances, declines, and so on. Short-term investment benefited from the use of Technical Analysis and Price Actions. In this paper, we suggested a technical indicator for an investment strategy to identify profitable stocks, such as the Moving Average Convergence/Divergence Method (MACD) and Arnaud Legoux Moving Average (ALMA). The MACD (12, 26, 9), MACD (4,22,3), MACD-ALMA (12, 26, 9), and MACD-ALMA (4,22,3) were the technical investment strategies used in this paper. 3.2.1 Moving average convergence/divergence (MACD) method MACD stands for moving average convergence/divergence and is one of the most widely utilized momentum indicators in technical analysis. Gerald Appel created this at the end of 1970 [46]. It is usually used both for long-term and short-term investors. The MACD line is calculated as the difference between the 12th and 26th-day exponential moving averages. It explains that the MACD line and the MACD moving average line show oscillations around the zero-level line over time, as well as divergent, convergent, and crossover movements [47]. MACD is commonly used with n 1, n 2, and n 3 combinations but other values can be substituted depending on goals. This is usually represented in the form MACD (n 1, n 2, n 3). For the case of (12, 26, 9), it was represented as MACD (12, 26, 9). The MACD indicator was used in its classic form, which calculates using exponential moving averages and prioritizes the most recent data in its weighting process. It is calculated by subtracting the shorter exponential moving average (EMA) of window length n 1 from the longer EMA of window length n 2. The MACD trading indicator consists of the following three elements.1. The MACD line (equation (2)): the difference between the short- and long-term exponential moving averages (EMA). 2. The Signal line (equation (3)): an exponential moving average of the MACD line. 3. Histogram (equation (4)): a graphical representation of the distance between the MACD line and the Signal line. The EMA is defined in equation (1) as(1) EMAt(N)=[2N(It−EMAt−1(N))]+EMAt−1(N) (2) MACDt(n1,n2)=EMAt(n1)−EMAt(n2) (3) Signalt(n3)=[2n3+1×(MACDt(n1,n2)−Signalt−1(n3))]+Signalt−1(n3) (4) Histogramt=MACDt−Signalt Where.EMAt(N) = Exponential moving average at time t N = window length of EMA (e.g., n 1 and n 2) n3 = denotes the period for the EMA calculation of the MACDt series It = closing day price (index value) at time t Signalt(n3) = EMA of MACD line at time t; (Signal Line) Signalt-1 = initial previous signal line that starts at 2nd period EMA of MACD line at time t 3.2.2 Moving average convergence/divergence (MACD) trading rules The MACD and the signal line move up and down the zero axis or midline to show the following trends such as overbought or oversold. When the EMA points are close together, it is called convergence, and when they are far apart, it is called divergence. The MACD line reacts more strongly when the moving average is shorter. Signal line crossovers, centerline crossovers, and divergence are some MACD indicators. The signal line is the MACD line's EMA. As a result, it follows the average line and aids in detecting turns in the MACD. When the MACD crosses over the signal line, it shows bullish and is called a bullish crossover. A bearish crossover occurs when the price falls below the signal line. Below are the MACD trading rules used in this paper and is shown in equations (5), (6):(5) BuySignal:Histogramt=MACDt−Signalt>0 (6) SellSignal:Histogramt=MACDt−Signalt<0 Also, the annual return is calculated as shown in equation (7),(7) RA=∑i=1MPSell−∑j=1NPBuy And the annual rate of return as shown in equation (8),(8) Ri=RARBS×100% Where; RA = Annual Return Ri = Annual Rate of Return RBS = Closing index value where the first transaction occurs (Buy or Sell) M = number of sell signal N = number of a buy signal PBuy and PSell are the closing index values on the days to execute buying and selling transactions, respectively A positive (negative) value of R A indicates a profit (loss), which is applicable in both the case of a long and a short trade. This paper uses the single-line crossover as a MACD indicator. Both MACD (12, 26, 9) and MACD (4,22,3) [48] were used in the analysis. Fig. 2 shows the same sample stock with the corresponding buying and selling index values for MACD (12, 26, 9) and MACD (4,22,3). “B” and “S” denote buying and selling point, respectively.Fig. 2 Trading rules for MACD (12, 26, 9) and MACD (4,22,3). (Upper: Candlestick; Middle: MACD and Signal Line of (12, 26, 9) Window; Lower: MACD and Signal Line of (4,22,3) Window). Fig. 2 3.2.3 Arnaud Legoux moving average (ALMA) The Arnaud Legoux Moving Average, abbreviated ALMA, is a relatively new addition to the family of moving average technical indicators. The ALMA was created in 2009 by Arnaud Legoux and Dimitrios Kouzis Loukas and has quickly received attention in the trading community. It is a kind of weighted moving average, and the shape of the coefficient is a Gaussian filter. A normal Gaussian filter is a symmetrical bell type with the highest center, but ALMA uses an asymmetric Gaussian filter with the peak shifted to the nearest position to improve price tracking [49]. Because the ALMA is based on the moving average indicator, it is universally acceptable across markets and time frames. Equations 9–12 shows the calculation of ALMA line.(9) ALMA=∑(PiCi)∑(Ci),1Norm∑i=1sizeP(i)e(i−offset)2σ2 (10) Ci=expexp{−(i−offset)2σ2} (11) σ=NA (12) Offset=Truncation{B(N−1)} Where.Pi (i = 1,2,3, …,N) = is each closing price Ci (i = 1,2,3, …,N) = is an arbitrary coefficient related to each closing price A, B, N are arbitrary, and (A = 6, B = 0.85 is the default) ALMA's formula is shown in (9), and it employs a Gaussian distribution shift with an offset so that it is not evenly centered on the window but is biased towards the more recent days. The offset can be adjusted, allowing us to trade off smoothness and responsiveness. The second parameter is the sigma parameter, which alters the shape of the filter, making it wider (larger sigma) or more focused (smaller sigma) (smaller sigma). The default value of 6 was inspired by the six sigma process, which provides excellent performance. 3.2.4 MACD-ALMA trading rules For this paper, ALMA is integrated into the MACD trading rules. If the signal and MACD were bullishly crossed in a buying condition, the investor would buy stocks. It signals a buying position if MACD is in an upward trend and the closing or opening price is higher than ALMA. The Sell position, on the other hand, will state that if the opening and closing price were both below ALMA, the first selling point should be sold immediately. This two-indicator hybrid was employed in a clustered group to select stocks that match the investing strategy requirements of MACD and ALMA. Below are the MACD-ALMA trading rules used in this paper and is shown in equations (13), (14):(13) BuySignal:Histogramt=MACDt−Signalt>0andALMAt<P(O,C)t (14) SellSignal:FirstALMAt+m>P(O,C)tafterBuyt Where.(t + m) is the point after Buy t P(O,C)t are opening and closing points The calculation of MACD, signal line, histogram, annual return, and the annual rate of return is the same as in MACD. MACD-ALMA (12, 26, 9) and MACD-ALMA (4,22,3) were also examined. Fig. 3 shows the same sample stock with the corresponding buying and selling index values for MACD-ALMA (12, 26, 9) and MACD-ALMA (4,22,3).Fig. 3 Trading rules for MACD-ALMA (12, 26, 9) and MACD-ALMA (4,22,3) (Upper: Candlestick and ALMA Line; Middle: MACD and Signal Line of (12, 26, 9) Window; Lower: MACD and Signal Line of (4,22,3) Window). Fig. 3 3.3 Stage: 3: K-means algorithm K-means Algorithm was used to cluster the Philippine Stock Market. The input data used to evaluate clusters was the annual rate of return and average annual risk. The greatest number of possible clusters investigated in this study was 20. The proposed elbow method was utilized to find the best cluster. K-means algorithm was also used to determine the cluster centroid and cluster labels for each data set. Likewise, the cluster profile or characteristics were investigated and analyzed. Unsupervised learning models include the K-means clustering algorithm. Unsupervised models are used to learn from unlabeled or uncategorized data [50]. It searches for commonalities in the data set and responds to the presence or absence of such commonalities in each data point. A K-means clustering model starts with K centroids and categorizes data points that are close (similar) to the centroids as clusters [51]. Below is the step-by-step procedure of the K-means clustering algorithm. Input: clusters number. 1st Step: (Initialization): Generate initial centroid. 2nd Step: (Assignation): Data point assignment to the nearest cluster (the nearest centroid). 3rd Step: (Re-computation): Update centroid. 4th Step: Calculate differences between the old and new centroid of each cluster. If the difference is lower than the tolerance limit, then stop. Otherwise, return to the 2nd Step. 3.3.1 Input data: average annual risk and annual rate of return In financial terms, risk is defined as the possibility that the real profits from an outcome or investment will differ from the expected outcome or return. The danger of losing some or all of the initial investment is a risk. While it is true that stocks are the most volatile of all investments, investors should keep in mind that uncertainty is an inherent part of the investing process. This means that any investment entails some level of risk. Limiting and managing your risk would be a healthier approach. A maximum level of income or loss should be established, and when that level is achieved, calculated decisions should be taken. Additionally, an investor's equivalent annual return over a specific time period is used to compute an annualized rate of return. Equations 15–19 shows the calcuation of daily return, variance of annual daily return, annual return, standard deviation of annual daily return, and annual rate of return, respectively.(15) DRt=Pt−Pt−1Pt−1×100% (16) σDR2=∑i=1n(DRt−DRAve)2n−1 (17) RA=∑i=1MPSell−∑j=1NPBuy (18) σDR=σ2 (19) Ri=RARBS×100% Where.DRt = Daily return at t time period Pt = Stock Closing Price at actual t time period Pt-1 = Stock Closing Price at previous t time period DRAve = annual average daily return n = no. Of daily returns in a year σDR2 = variance of annual daily return σDR = standard deviation of annual daily return σ = standard deviation of stock per year RA = Annual Return Ri = Annual Rate of Return RBS = Closing index value where the first transaction occurs (Buy or Sell) M = number of sell signal N = number of a buy signal PBuy and P Sell are the closing index values on the days to execute buying and selling transactions, respectively. Fig. 4 shows an example of stock data plotted in candlestick form and corresponding daily returns for the entire year.Fig. 4 Sample stock data and daily return (Upper: Candlestick; Lower: Daily Return). Fig. 4 3.3.2 Min-max normalization In this paper, min-max normalization was used. The minimum value of each feature is transformed to a 0 value, the highest value is transformed to a 1, and the remaining values are converted to a decimal between 0 and 1. Normalizing the data so that they are all on the same scale means that each attribute is properly weighed before being used in the model [52]. The highest and smallest numerical values of each numeric value are determined, and the others are transformed correspondingly as shown in equation (20).(20) Y′=y+yy−y Where.Y’ = transformed value/pre-processed value y = observation value y min = minimum observation value, and y max = maximum observation value. The values in the dataset are reduced to {0, 1}. 3.3.3 Elbow method The elbow approach uses the sum of squared distance (SSE) to find an optimum value of k depending on the distance between data points and their allocated clusters [53]. We pick a k number where the SSE starts to flatten out and an inflection point appears. The method's name comes from the fact that this graph resembles an elbow when displayed. The goal is to have a low SSE k value. This paper uses the “knee locator” function for the elbow method to exactly determine the optimal number of clusters. 3.4 Stage: 4: evaluation of stock rate of return per cluster (asset allocation) After calculating the rate of return (Ri) using the trading rules discussed previously, the next thing is to select the possible number of companies per cluster to include in the portfolio. This paper set some rules for portfolio selection.1. Eliminate companies with a negative Ri 2. Eliminate companies with a window below MACD or ALMA window 3. Up to 10 companies per cluster. 10 is just an arbitrary number to simplify the analysis. Though it can be less than or more than 10. Based on author's own perspective and experience, 10 or more companies are difficult to manage and monitor. 4. Select companies with the highest Ri per cluster 3.5 Stage 5: portfolio optimization (mean-variance model) According to the mean-variance model, investors prefer the security with the higher return considering specific risks, or the one with the lower risk based on a specific expected return. It is accomplished by evaluating the amount of risk that investors are prepared to take in return for rewards [54]. Since one of the clustering attributes used was an average annual risk, this paper uses a lower risk based on a specific expected return criterion. Equation (21) shows the objective function while equations (22), (23) are the contraints. The model is formulated as a minimization problem:(21) Min(σP2)=∑i=1nσi2wi2+2∑i=1n−1∑j=i+1nwiwjσijρij Subject to:(22) RP=∑i=1nwiRi (23) ∑i=1nwi=1 Where.σp2 = denotes the variance of portfolio P wi = denotes the weight on asset i σi = denotes the standard deviation of asset i σij = denotes the covariance of asset i and asset j ρij = denotes the correlation between asset i and asset j RP = Expected portfolio return Ri = Annual Rate of Return 3.6 TAKMV simulation method 2018 (pre-COVID-19) and 2020 (during-COVID-19) data with 230 and 239 companies, respectively, were used to calculate each company's annual rate of return (Ri) based on the proposed technical investment strategies (MACD and MACD-ALMA) and trading rules. Ri may be a positive or negative value that indicates gain or loss or a “zero” value that indicates that the stock window is below the required window or there is no buying/selling point. Only the companies with positive Ri will be considered in the entire analysis. In parallel, average annual risks were also computed for each company. This resulted in an average annual risk and an annual rate of return per company per strategy. The average annual risk and an annual rate of return will be the attributes of the K-means clustering algorithm. The elbow Method was used to determine the optimal number of clusters. The resulting clusters will be the portfolio used in the optimization. To simplify the analysis, each portfolio was limited to 10 companies with the highest Ri. Each portfolio was simulated based on a mean-variance model that uses a lower risk based on a specific expected return criterion. 4 Results and discussion The data used were composed of 230 and 239 companies in the Philippines for 2018 (pre-COVID-19) and 2020 (during-COVID-19), respectively. This was then validated by applying the results and comparing them to 2019 and 2021 data, respectively. All simulations in the proposed method such as technical analysis, K-means clustering, and portfolio optimization were all conducted in the Matlab platform. 4.1 Technical analysis results The MACD (12, 26, 9), MACD (4,22,3), MACD-ALMA (12, 26, 9), and MACD-ALMA (4,22,3) were used to determine the annual rate of return (Ri) of individual companies per cluster group and highly dependent on the trading rules. The annual rate of return (Ri) may be a positive (negative) value that indicates gain (loss) or a “zero” value that indicates that the stock window is below the required MACD window or there is no buying/selling point. Only the companies with positive Ri will be considered in the entire analysis. Table 1 shows the summary of results for 2018 and 2020 data, respectively. Fig. 5 shows the graphical comparison of the number of companies with positive Ri for 2018 and 2020 data.Table 1 Technical analysis summary of 2018 and 2020 data. Table 1Technical Investment Strategy 2018 Data (230 Companies) 2020 Data (239 Companies) +Ri -Ri “0″ Ri +Ri -Ri “0″ Ri MACD (12,26,9) 154 71 5 86 146 7 MACD (4,22,3) 167 62 1 82 157 0 MACD-ALMA (12,26,9) 5 76 149 16 95 128 MACD-ALMA (4,22,3) 16 108 106 50 101 88 Fig. 5 No. Of companies with positive Ri in different technical investment strategies. Fig. 5 Based on Table 1, for 2018 Data, (pre-COVID-19) MACD strategy dominates the MACD-ALMA strategy regarding the number of companies with positive Ri. MACD (12, 26, 9) generates 66.96% (154/230) number of companies with positive Ri, while MACD (4,22,3), generates 72.61% (167/230) positive Ri. Meanwhile, During-COVID-19 condition (2020 data), the MACD strategy still dominates the MACD-ALMA regarding the number of companies with positive Ri. MACD (12, 26, 9) generates 35.98% (86/239). Of companies with positive Ri, while. MACD (4,22,3), generates 34.31% (82/239) number of companies with positive Ri. The MACD Strategy greatly affects its performance during COVID-19 conditions and receives a huge reduction of companies with positive Ri. From 66.96% down to 35.98%. For MACD (12, 26, 9), while 72.61% down to 34.31% number of companies with positive Ri for MACD (4,22,3). This only shows how the pandemic hit the stock market with regard to the effectiveness of the MACD strategy. However, the MACD-ALMA (12, 26, 9) and MACD-ALMA (4,22,3) generates 2.17% (5/230) and 6.96% (16/230) number of companies with positive Ri during Pre-Covid 19 (2018 data). During COVID-19 conditions (2020 data). , Both windows in MACD-ALMA Strategies surprisingly increase their performance. MACD-ALMA (12, 26, 9) generates 6.69% (16/239). Percent number of companies with positive Ri, while. MACD-ALMA (4,22,3) generates a 20.92% (50/239) number of companies with positive Ri. MACD (12, 26, 9) is superior compared to MACD (4,22,3) regarding the number of companies with positive Ri. MACD-ALMA (4,22,3) is superior compared to MACD-ALMA (12, 26, 9) regarding the number of companies with positive Ri. 4.2 K-means clustering results The 2018 and 2020 data were run in the K-means algorithm taking the average annual risk (σ) and an annual rate of return (Ri) as the clustering attribute. The Elbow Method was used to determine the best clusters for the given data. The cluster centroids would be the basis of the level of risk per cluster. For the 2018 data (230 companies), the simulation shows that the best clusters were 8, 8, 2, and 4 for MACD (12, 26, 9), MACD (4,22,3), MACD-ALMA (12, 26, 9), and MACD-ALMA (4,22,3), respectively. For 2020 data (239 companies), the best clusters were 7, 7, 4, and 6 for MACD (12, 26, 9), MACD (4,22,3), MACD-ALMA (12, 26, 9), and MACD-ALMA (4,22,3), respectively. Table 2, Table 3 show the summarized results of the Elbow Method for 2018 and 2020, respectively. Fig. 6, Fig. 7 shows the graphical representation of 2018 cluster assignment and centroid for MACD and MACD-ALMA, respectively, while Fig. 8, Fig. 9 shows the graphical representation of 2020 cluster assignment and centroid for MACD and MACD-ALMA, respectively.Table 2 Cluster centroid for 2018 data. Table 2MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster N + Ri σ(C) Ri(C) Cluster N + Ri σ(C) Ri(C) Cluster N + Ri σ(C) Ri(C) Cluster N + Ri σ(C) Ri(C) E1 40 2.05 101.70 E1 8 8.72 49.48 E1 3 1.62 4.27 E1* 1 1.97 58.86 E2 15 3.30 53.96 E2 17 5.34 93.19 E2 2 6.12 52.88 E2 4 3.61 6.90 E3 36 1.75 18.22 E3 25 3.45 61.87 – – – – E3 3 5.72 46.98 E4 20 3.83 134.77 E4 9 5.14 186.10 – – – – E4 8 1.61 4.26 E5 12 6.62 97.27 E5 26 1.82 93.66 – – – – – – – – E6 5 8.96 18.69 E6 22 2.16 139.76 – – – – – – – – E7 9 5.22 25.44 E7 11 4.53 21.07 – – – – – – – – E8 17 3.38 14.99 E8 49 1.93 22.64 – – – – – – – – Where: N + Ri - number of companies with positive Ri. σ(C) - Risk per Clustered Portfolio. Ri(C) - Clustered Portfolio Return. Note: E1 of MACD-ALMA (4,22,3) should be removed since only 1 company in a cluster. Table 3 Cluster centroid for 2020 data. Table 3MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster N + Ri σ(C) Ri(C) Cluster N + Ri σ(C) Ri(C) Cluster N + Ri σ(C) Ri(C) Cluster N + Ri σ(C) Ri(C) E1 5 10.15 45.22 E1 23 2.77 20.76 E1 4 2.02 9.91 E1 10 2.05 20.93 E2 7 6.43 181.00 E2 24 4.37 43.15 E2 5 3.89 38.93 E2 5 5.00 225.42 E3 17 4.50 20.51 E3 4 7.45 197.74 E3 3 5.32 48.40 E3 14 3.32 22.31 E4 22 3.41 117.33 E4 12 6.83 53.72 E4 4 4.16 15.20 E4 6 3.80 88.60 E5 23 2.85 20.14 E5 4 10.36 71.34 – – – – E5 5 7.24 30.63 E6 4 7.96 119.41 E6 7 4.82 137.81 – – – – E6 10 4.81 24.40 E7 8 6.74 33.39 E7 8 2.11 117.69 – – – – – – – – Where: N + Ri - number of companies with positive Ri. σ(C) - Risk per Clustered Portfolio. Ri(C) - Clustered Portfolio Return. Fig. 6 Cluster assignments and centroid for 2018 data (upper: Macd (12, 26, 9); lower: Macd (4,22,3)). Fig. 6 Fig. 7 Cluster assignments and centroid for 2018 data (upper: MACD-ALMA (12, 26, 9); lower: MACD-ALMA (4,22,3)). Fig. 7 Fig. 8 Cluster assignments and centroid for 2020 data (upper: Macd (12, 26, 9); lower: Macd (4,22,3)). Fig. 8 Fig. 9 Cluster assignments and centroid for 2020 data (upper: MACD-ALMA (12, 26, 9); lower: MACD-ALMA (4,22,3)). Fig. 9 4.3 Asset allocation results After k-means clustering, select the companies per cluster to include in the portfolio. Rules for asset allocation: In each different Technical Investment Strategy.1. The maximum allowable number of companies in a portfolio was set to 10 (2 ≤ N ≤ 10) 2. If there are more than 10 companies who have positive Ri (Return), select 10 companies with the maximum return (Ri) per cluster. 3. “N/A” means less than 2 companies Table 4, Table 5 shows that asset allocation for 2018 and 2020 data, respectively.Table 4 Asset allocation for 2018 data. Table 4MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster N + Ri AA Cluster N + Ri AA Cluster N + Ri AA Cluster N + Ri AA E1 40 10 E1 8 8 E1 3 3 E1 1 N/A E2 15 10 E2 17 10 E2 2 2 E2 4 4 E3 36 10 E3 25 10 – – – E3 3 3 E4 20 10 E4 9 9 – – – E4 8 8 E5 12 10 E5 26 10 – – – – – – E6 5 5 E6 22 10 – – – – – – E7 9 9 E7 11 10 – – – – – – E8 17 10 E8 49 10 – – – – – – Table 5 Asset allocation for 2020 data. Table 5MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster N + Ri AA Cluster N + Ri AA Cluster N + Ri AA Cluster N + Ri AA E1 5 5 E1 23 10 E1 4 4 E1 10 10 E2 7 7 E2 24 10 E2 5 5 E2 5 5 E3 17 10 E3 4 4 E3 3 3 E3 14 10 E4 22 10 E4 12 10 E4 4 4 E4 6 6 E5 23 10 E5 4 4 – – – E5 5 5 E6 4 4 E6 7 7 – – – E6 10 10 E7 8 8 E7 8 8 – – – – – – Where: N + Ri - number of companies with positive Ri, AA – asset allocation. Note: E1 of MACD-ALMA (4,22,3) should be removed since only 1 company in a cluster. 4.4 Portfolio optimization results The mean-variance portfolio optimization model minimizes the risk (σ p) given a specified expected return (R P) in each portfolio. The optimization model will then suggest a weight for each asset/company on that portfolio. This paper also assumes that companies are independent of each other and therefore set the correlation parameter (ρ) to zero. Since R P is subjectively determined by the investor, this paper examines different R P that resulted in different minimum risk levels (σ p). The upper and lower limit of R P used in this paper is the maximum and minimum values in a portfolio per cluster. The minimum risk (σp) between those ranges of R P will result in R P at a global minimum risk. Table 7, Table 9 show R P (in %) at global minimum risk for 2018 and 2020 data, respectively. The resulting weights per portfolio was shown in Table 8, Table 10 . Table 6 shows a Sample of portfolio optimization from MACD (12, 26, 9) of the Elbow Method, Cluster 4. Fig. 10 shows an R P of 146.8182 and has a Global minimum risk of 1.102035 (See Table 6, Table 7). This method selects the minimum risk that generates positive weights and is called “R P at global minimum risk” and negative weight has been eliminated in the simulations.Table 7 Optimal portfolio for 2018 data. Table 7MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster σp RP Cluster σp RP Cluster σp RP Cluster σp RP E1 0.48 116.11 E1 3.03 46.61 E1 0.86 2.83 E1 N/A N/A E2 0.90 64.00 E2 1.61 110.66 E2 4.29 57.64 E2 1.76 5.88 E3 0.60 28.61 E3 1.05 74.03 – – – E3 3.24 50.19 E4 1.10 146.82 E4 1.62 182.31 – – – E4 0.55 4.11 E5 1.93 99.36 E5 0.47 107.49 – – – – – – E6 3.87 19.87 E6 0.65 153.22 – – – – – – E7 1.71 25.57 E7 1.33 23.34 – – – – – – E8 1.01 20.48 E8 0.59 41.68 – – – – – – Table 9 Optimal portfolio for 2020 data. Table 9MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster σp RP Cluster σp RP Cluster σp RP Cluster σp RP E1 4.46 48.19 E1 0.78 34.97 E1 0.90 9.36 E1 0.61 20.55 E2 2.36 180.23 E2 1.31 64.18 E2 1.69 39.30 E2 1.99 227.59 E3 1.34 30.38 E3 3.56 200.60 E3 3.04 50.87 E3 1.01 29.59 E4 0.92 130.64 E4 2.12 59.98 E4 1.99 16.19 E4 1.51 85.74 E5 0.82 29.53 E5 5.12 68.31 – – – E5 3.22 28.51 E6 3.93 116.57 E6 1.77 136.15 – – – E6 1.50 25.68 E7 2.35 32.97 E7 0.68 114.12 – – – – – – Table 8 Optimal Portfolio weights for 2018 data. Table 8Strategy Cluster w1 w2 w3 w4 w5 w6 w7 w8 w9 w10 MACD (12,26,9) E1 0.0599 0.0900 0.0535 0.0854 0.0728 0.0656 0.2099 0.0750 0.2382 0.0498 E2 0.0674 0.3255 0.2315 0.1020 0.0504 0.0493 0.0353 0.0464 0.0484 0.0439 E3 0.0747 0.1155 0.1048 0.1129 0.0783 0.1212 0.0803 0.0598 0.1892 0.0632 E4 0.0369 0.2049 0.2076 0.0750 0.0933 0.0462 0.0776 0.0436 0.0462 0.1689 E5 0.0610 0.0829 0.0399 0.0867 0.1013 0.1747 0.0999 0.1782 0.1073 0.0680 E6 0.1811 0.1937 0.2701 0.2438 0.1113 – – – – – E7 0.1338 0.1313 0.0709 0.1022 0.1198 0.0950 0.1334 0.1323 0.0813 – E8 0.1045 0.1153 0.0956 0.0899 0.0766 0.1066 0.0565 0.1263 0.0815 0.1471 MACD (4,22,3) E1 0.0747 0.1571 0.1207 0.1492 0.1211 0.1477 0.1119 0.1177 – – E2 0.1214 0.1585 0.0707 0.1001 0.1426 0.0729 0.0687 0.0836 0.0696 0.1118 E3 0.0948 0.1109 0.0897 0.1485 0.1017 0.0691 0.1553 0.0892 0.0837 0.0571 E4 0.0961 0.0777 0.1081 0.1697 0.1629 0.0649 0.0583 0.1617 0.1007 – E5 0.0988 0.0641 0.1827 0.0564 0.0831 0.0854 0.1083 0.1828 0.1006 0.0378 E6 0.0736 0.0863 0.1120 0.0718 0.0961 0.1327 0.1377 0.0459 0.1398 0.1042 E7 0.1116 0.0860 0.0839 0.0883 0.1546 0.1238 0.0693 0.1351 0.0431 0.1042 E8 0.0889 0.0766 0.1571 0.0737 0.0761 0.1006 0.1414 0.0478 0.1776 0.0602 MACD-ALMA (12,26,9) E1 0.1715 0.2633 0.5651 – – – – – – – E2 0.5700 0.4300 – – – – – – – – MACD-ALMA (4,22,3) E2 0.2002 0.3217 0.2992 0.1789 – – – – – – E3 0.4260 0.3266 0.2473 – – – – – – – E4 0.0928 0.0994 0.2425 0.1414 0.1080 0.0749 0.1092 0.1318 – – Table 10 Optimal Portfolio weights for 2020 data. Table 10Strategy Cluster w1 w2 w3 w4 w5 w6 w7 w8 w9 w10 MACD (12,26,9) E1 0.2287 0.2702 0.1409 0.1899 0.1703 – – – – – E2 0.1312 0.1939 0.1080 0.0872 0.1984 0.1648 0.1166 – – – E3 0.0959 0.0917 0.1153 0.1192 0.0916 0.1272 0.0670 0.1190 0.1106 0.0625 E4 0.0740 0.1058 0.0538 0.0488 0.1627 0.0361 0.2081 0.0495 0.0505 0.2107 E5 0.0999 0.0535 0.0700 0.0703 0.0549 0.1491 0.0858 0.1513 0.0669 0.1983 E6 0.1783 0.2759 0.2590 0.2868 – – – – – – E7 0.1471 0.0984 0.0856 0.1407 0.1195 0.1494 0.1275 0.1319 – – MACD (4,22,3) E1 0.1387 0.1224 0.1199 0.0636 0.0703 0.1339 0.0687 0.1841 0.0558 0.0425 E2 0.1250 0.0892 0.1130 0.0661 0.0955 0.1232 0.1209 0.0885 0.0782 0.1003 E3 0.1704 0.4166 0.1996 0.2134 – – – – – – E4 0.0997 0.0801 0.0832 0.0723 0.1350 0.1450 0.0929 0.0976 0.1103 0.0838 E5 0.1842 0.2515 0.2959 0.2684 – – – – – – E6 0.1104 0.1075 0.1840 0.1586 0.1544 0.1983 0.0867 – – – E7 0.0526 0.1114 0.1124 0.1336 0.1870 0.0679 0.2670 0.0682 – – MACD-ALMA (12,26,9) E1 0.2663 0.1698 0.1002 0.4637 – – – – – – E2 0.1647 0.3239 0.1640 0.1444 0.2030 – – – – – E3 0.4105 0.3219 0.2676 – – – – – – – E4 0.3450 0.2299 0.1374 0.2877 – – – – – – MACD-ALMA (4,22,3) E1 0.0764 0.0901 0.1202 0.1716 0.0676 0.1465 0.0626 0.0550 0.0581 0.1517 E2 0.4424 0.1068 0.1150 0.1148 0.2210 – – – – – E3 0.0866 0.1289 0.0728 0.1154 0.0904 0.0739 0.1279 0.1086 0.0654 0.1299 E4 0.1335 0.1080 0.2237 0.2262 0.1625 0.1462 – – – – E5 0.1787 0.1929 0.1949 0.1851 0.2483 – – – – – E6 0.0996 0.1149 0.1196 0.1249 0.1023 0.0791 0.0882 0.1146 0.0782 0.0786 Table 6 Sample from MACD (12, 26, 9) of elbow 4. Table 6Rp σp 144.5974 1.14671 144.9675 1.13286 145.3377 1.121485 145.7078 1.112663 146.0779 1.106455 146.4481 1.102903 146.8182 1.102035 147.1883 1.103856 147.5584 1.108353 147.9286 1.115494 148.2987 1.125229 148.6688 1.13749 149.039 1.152198 149.4091 1.16926 149.7792 1.188574 150.1494 1.210033 150.5195 1.233524 150.8896 1.258935 151.2597 1.286151 151.6299 1.31506 Fig. 10 RP and sigma of MACD (12, 26, 9), elbow method 4 (upper: Positive and negative weights; lower: Positive weights). Fig. 10 Based on Table 7 (2018 data), R P at global minimum risk ranges are 19.87%–146.82%, 23.34%–182.31%, 2.83%–57.64%, and 4.11%–50.19% for MACD (12, 26, 9), MACD (4,22,3), MACD-ALMA (12, 26, 9), and MACD-ALMA (4,22,3), respectively. From Table 9 (2020 data), the ranges are 29.53%–180.23%, 34.97%–200.60%, 9.36%–50.87%, and 20.55%–227.59%, respectively. For the MACD strategy, the result shows that the E4 portfolio using MACD (4,22,3) has a maximum R P of 182.31% for the 2018 data while the E2 portfolio using MACD-ALMA (4,22,3) has a maximum R P of 227.59% for the 2020 data. This also shows that the MACD-ALMA strategy managed a high-risk market condition but can achieve maximum return. 4.5 Validation of results using the next Year's historical price To determine the performance of the TAKMV method, it was validated by applying the results and comparing them to the next year's historical price. In this case, 2018 results were compared to 2019 data, and 2020 results were compared to 2021 data. The comparison is applied to the same company per portfolio (simulation results and next year's data) for consistency. No validation was conducted between 2019 and 2020 since this was the transition of pre- and during-COVID-19 conditions. Table 11, Table 12 and Fig. 11, Fig. 12 show the comparison of R P applying the same weights for 2019 (2018 wt) and 2021 (2020 wt). %RA and % RB are the ratios of portfolio return for 2018–2019 and 2021–2022, respectively.Table 11 Validation of 2018 Results to 2019 Data (Pre-Covid condition). Table 11MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster RP(2018) RP(2019) %RA Cluster RP(2018) RP(2019) %RA Cluster RP(2018) RP(2019) %RA Cluster RP(2018) RP(2019) %RA E1 116.11 7.04 6.06% E1 46.61 55.65 119.42% E1 2.83 −0.32 −11.25% E1* N/A N/A E2 64.00 30.82 48.17% E2 110.66 31.63 28.59% E2 57.64 −12.04 −20.89% E2 5.88 −2.22 −37.67% E3 28.61 57.46 200.83% E3 74.03 45.70 61.73% – – – E3 50.19 −14.46 −28.81% E4 146.82 4.04 2.75% E4 182.31 52.55 28.83% – – – E4 4.11 −3.54 −86.19% E5 99.36 83.46 84.00% E5 107.49 5.52 5.14% – – – – – – E6 19.87 38.67 194.61% E6 153.22 36.32 23.71% – – – – – – E7 25.57 59.54 232.89% E7 23.34 56.23 240.97% – – – – – – E8 20.48 2.39 11.69% E8 41.68 24.74 59.36% – – – – – – %RA = (RP(2019)/RP(2018))100%. Table 12 Validation of 2020 Results to 2021 Data (Covid condition). Table 12MACD (12,26,9) MACD (4,22,3) MACD-ALMA (12,26,9) MACD-ALMA (4,22,3) Cluster RP(2018) RP(2019) %RA Cluster RP(2018) RP(2019) %RA Cluster RP(2018) RP(2019) %RA Cluster RP(2018) RP(2019) %RA E1 48.19 90.85 188.50% E1 34.97 5.69 16.27% E1 9.36 −1.46 −15.61% E1 20.55 −5.01 −24.40% E2 180.23 20.65 11.46% E2 64.18 31.85 49.62% E2 39.30 −0.80 −2.04% E2 227.59 −17.12 −7.52% E3 30.38 −6.80 −22.37% E3 200.60 44.64 22.25% E3 50.87 −2.55 −5.02% E3 29.59 −3.02 −10.22% E4 130.64 38.09 29.16% E4 59.98 76.84 128.12% E4 16.19 −2.84 −17.54% E4 85.74 −11.22 −13.09% E5 29.53 52.57 178.03% E5 68.31 43.22 63.28% – – – E5 28.51 −4.51 −15.83% E6 116.57 35.68 30.61% E6 136.15 8.21 6.03% – – – E6 25.68 −5.86 −22.80% E7 32.97 39.62 120.20% E7 114.12 19.78 17.33% – – – – – – %RB <svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="20.666667pt" height="16.000000pt" viewBox="0 0 20.666667 16.000000" preserveAspectRatio="xMidYMid meet"><metadata> Created by potrace 1.16, written by Peter Selinger 2001-2019 </metadata><g transform="translate(1.000000,15.000000) scale(0.019444,-0.019444)" fill="currentColor" stroke="none"><path d="M0 440 l0 -40 480 0 480 0 0 40 0 40 -480 0 -480 0 0 -40z M0 280 l0 -40 480 0 480 0 0 40 0 40 -480 0 -480 0 0 -40z"/></g></svg> (RP(2021)/RP(2020))100%. Fig. 11 Portfolio risk and return results comparison per clustered portfolio (upper: 2018 data; lower: 2019 data). Fig. 11 Fig. 12 Portfolio risk and return results comparison per clustered portfolio (upper: 2020 data; lower: 2021 data). Fig. 12 Based on Table 11, Table 12, the results show that the MACD strategy, both (12, 26, 9) and (4,22,3) is effective in comparison to the MACD-ALMA strategy in terms of next year's portfolio return (R P). It is noticeable that MACD-ALMA resulted in negative R P. For 2018 results (pre-Covid-19), the maximum %R A is 232.89% (E7) and 240.97% (E7) for MACD (12, 26, 9) and MACD (4,22,3), respectively, while for 2020 results (during-COVID-19), the maximum %R B is 188.50% (E1) and 128.12% (E4) for MACD (12, 26, 9) and MACD (4,22,3), respectively. 5 Conclusions The COVID-19 pandemic has been a worldwide health crisis. The Philippine stock market was seriously impacted by the global pandemic COVID-19. Retail investors continue to seek great ones in the damaged market. This paper uses three important analyses namely Technical Analysis, K-means Clustering, and Mean-Variance Portfolio Optimization Model (referred to as TAKMV Methodology) to determine possible portfolio and optimized portfolio returns while minimizing the expected risk. The optimal number of clusters was determined using the Elbow Method with average annual risk and an annual rate of return as an attribute. A total of 230 assets/companies and 239 assets/companies for 2018 (pre-COVID-19 condition) and 2020 (during-COVID-19 condition), respectively, were used as a dataset. MACD and hybrid strategy (MACD-ALMA) with different windows were compared and analyzed to determine their performance during normal and COVID-19 conditions. The application of technical analysis, clustering techniques, and portfolio optimization provides an efficient way to determine a pool of portfolios based on asset risk and return. Clustering and portfolio optimization will help to manage the portfolio selection and was based on the investor's risk criteria. The technical analysis result shows that MACD works efficiently in the pre-COVID-19 conditions while MACD-ALMA works efficiently during-COVID-19 conditions, regardless of the number of assets with a positive annual rate of return. Using the Elbow Method for K-means clustering shows that the optimal number of clusters for 2018 and 2020 data are different. Portfolio optimization result shows that optimal return can be obtained using the MACD strategy in a pre-COVID-19 condition and MACD-ALMA during-COVID-19 condition. This also shows that the MACD-ALMA strategy manages a high-risk market condition and can achieve maximum return during COVID-19 conditions. To determine the performance of the TAKMV method, it was validated by applying its results and comparing it to the next year's historical price. The 2018 results were compared to 2019 data and the 2020 results were compared to 2021 data. The comparison is applied to the same company per portfolio for consistency. No validation was conducted between 2019 and 2020 since this was the transition of pre- and during-COVID-19 conditions. Validation of results using next year's historical price shows that the MACD strategy is more effective compared to MACD-ALMA. Overall, Technical Analysis can help traders and investors to identify trends and patterns in the Philippine stock market during the pandemic. With the pandemic causing significant economic disruption, technical analysis can help traders to identify potential buying and selling opportunities based on market trends and patterns. While K-means clustering can be used to create a diversified portfolio of stocks that can help reduce risk during volatile market conditions. Additionally, the Mean-Variance Model can be used to develop investment strategies that maximize returns while minimizing risk. Which can be used to allocate funds across different sectors of the Philippine stock market based on market trends and risk factors. With the use of the TAKMV method, investors, students, and decision-makers can make more profitable investment decisions in the Philippine stock market using Technical Analysis, Machine Learning, and Mean-Variance Model. 5.1 Theoretical contribution This study provides theoretical contributions to the existing literature by using the three important analyses (Technical Analysis, K-Means Clustering, and Mean-Variance Portfolio Optimization Model) named TAKMV methodology to determine possible portfolio and optimized portfolio returns while minimizing the expected risk amidst the COVID-19 pandemic. First, it enhanced stock selection: The combination of Technical Analysis, K-means clustering, and Mean-Variance model can provide investors with a more comprehensive approach to stock selection. This can lead to more informed investment decisions and potentially higher returns. Second, it improved risk management: The Mean-Variance model can help investors to manage risk by selecting portfolios that optimize returns for a given level of risk. The use of K-means clustering and Technical Analysis can further enhance risk management by identifying stocks that are less correlated with each other and potentially less risky. Third, Technical Analysis can help investors to identify market trends and potential turning points. The use of K-means clustering can provide a more detailed analysis of market trends by grouping stocks with similar price movements. Fourth, it also improved portfolio diversification. The use of K-means clustering and the mean-variance model can help investors to create diversified portfolios that are less exposed to specific industries or stocks. This can reduce the risk of significant losses due to market or industry-specific events. And lastly, it identifies new investment opportunities. The use of K-means clustering, and Technical Analysis can help investors to identify new investment opportunities that may have been overlooked by traditional analysis methods. The integration of Technical Analysis, K-means clustering, and Mean-Variance model can provide a more comprehensive approach to stock selection and risk management in the Philippine stock market during the COVID-19 pandemic. The theoretical contributions of this approach include enhanced stock selection, improved risk management, a better understanding of market trends, improved portfolio diversification, and identification of new investment opportunities. These contributions can help investors to make more informed investment decisions, potentially leading to higher returns and reduced risk. 5.2 Practical implication This paper has implications for practitioners, decision-makers, and managers. Using the combined methods of technical analysis, K-means clustering, and mean-variance portfolio optimization model, this paper develops a cluster trading strategy that allows for building a diversified portfolio with technical analysis. This paper provides a step-by-step procedure for assessing, clustering, selecting, and optimizing portfolios while considering the risk and its return. This method can be used by engineers, managers, institutional, and retail investors to select the best stock portfolios; they will gain all the benefits of diversification in this study as this will help to protect an investor's portfolio from the systematic risk that could expose the portfolio to losses. In addition, broad market pullbacks such as the 2020 pandemic situation obviously can suffocate the entire portfolio without any good strategy employed. The proposed TAKMV Method will reduce the extreme downside losses in any market condition. Importantly, this method can be applied to other financial markets such as foreign exchange and cryptocurrency exchange, which are popular and trending among retail investors. 5.3 Limitations and future research This paper limits the assessment using previous data to test and validate the performance of the proposed TAKMV method. Return estimation with the use of forecasting tools and techniques can be considered in future research such as Holt-Winters, neural networks, or other forecasting techniques, and can be employed together with other evolutionary optimization methods. Also, the application of TAKMV to other financial markets such as foreign exchange and cryptocurrency exchange as well as applications of it to post-pandemic conditions. Lastly, other technical investment strategies can be explored and utilized in the TAKMV method. Author contributions Conceptualization, M.M.N., and M.N·Y.; methodology, M.M.N., M.N.Y. and J.V.T; validation, M.M.N., M.N·Y., and J.V.T; formal analysis, M.M.N., M.N.Y. and J.V.T.; investigation, M.M.N, and M.N·Y.; resources, M.M.N., M.N.Y. and J.V.T.; data curation, M.M.N., and M.N·Y.; writing—original draft preparation, M.M.N., M.N.Y. and J.V.T.; writing—review and editing, M.N·Y., Y.T.P. and J.V.T.; visualization, M.N.Y. and Y.T.P.; supervision, M.N·Y., Y.T.P. and J.V.T.; project administration, M.M.N., M.N·Y., and J.V.T.; funding acquisition, M.N·Y., and Y.T.P. All authors have read and agreed to the published version of the manuscript. Funding This research was funded by Mapúa University Directed Research for Innovation and Value Enhancement (DRIVE). Author contribution statement Maricar M. Navarro: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper. Michael Nayat Young: Conceived and designed the experiments, Analyzed and interpreted the data, Wrote the paper. Yogi Tri Prasetyo: Analyzed and interpreted the data; Wrote the paper. Jonathan V. Taylar: Analyzed and interpreted the data; Wrote the paper. Data availability statement Data will be made available on request. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References 1 Gherghina Ș.C.a. Joldeș C.C. Stock market reactions to COVID-19 pandemic outbreak: quantitative evidence from ARDL bounds tests and granger causality analysis Int. J. Environ. Res. Publ. Health 17 18 2020 6729 10.3390/ijerph17186729 2 Elsayed A. Abdelrhim M. The Effect of COVID-19 Spread on Egyptian Stock Market Sectors Available at: SSRN 3608734 2020 3 Al-Awadhi A.M. Alsaifi K. Ahmad A. Alhammadi S. Death and contagious infectious diseases: impact of the COVID-19 virus on stock market return J. Behav. Experimen. Finan. 27 2020 100326 10.1016/j.jbef.2020.100326 4 Kartal M.T. Depren S.K. Depren Ö. How main stock exchange indices react to covid-19 pandemic: daily evidence from east asian countries Global Econ. Rev. 50 1 2021 54 71 10.1080/1226508X.2020.1869055 5 Xu D. Canadian stock market volatility under COVID-19 Int. Rev. Econ. Finance 77 2022 159 169 6 Yiu M.S. Tsang A. Impact of COVID-19 on ASEAN5 stock markets Journal of Asia Pacific 2021 10.1080/13547860.2021.1947550 7 Ozkan O. Impact of COVID-19 on stock market efficiency: evidence from developed countries 2021 Research in International Business and Finance 10.1016/j.ribaf.2021.101445 8 Hatmanu M. Cautisanu C. The impact of COVID-19 pandemic on stock market: evidence from Romania Int. J. Environ. Res. Publ. Health 18 2021 9315 10.3390/ijerph18179315 9 Kartal M.T. Ertuğrul H.M. Ulussever T. The impacts of foreign portfolio flows and monetary policy responses on stock markets by considering COVID-19 pandemic: evidence from Turkey Borsa Istanbul Review 22 2022 12 19 10.1016/j.bir.2021.06.003 10 Li B. Hoi S.C. Sahoo D. Liu Z.-Y. Moving average reversion strategy for on-line portfolio selection Artif. Intell. 222 2015 104 123 11 Li T. Zhang W. Xu W. A fuzzy portfolio selection model with background risk Appl. Math. Comput. 256 2015 505 513 12 Markowitz H. Portfolio selection J. Finance 1952 77 91 10.1111/j.1540-6261.1952.tb01525.x 13 Sui Y. Hu J. Ma F. A possibilistic portfolio model with fuzzy liquidity constraint Complexity 2020 2020 1 10 14 Li B. Hoi S.C. Online portfolio selection: a survey ACM Comput. Surv. 46 2014 35 10.1145/2512962 15 Chang K.H. Young M.N. Behavioral stock portfolio optimization considering holding periods of B-stocks with short-selling Comput. Oper. Res. 112 2019 10.1016/j.cor.2019.104773 16 Moreno D. Petrakisab M. The impact on market outcomes of the portfolio selection of large equity investors Econ. Lett. 212 2022 10.1016/j.econlet.2022.110296 17 Rodríguez Y.E. Gómez J.M. Contreras J. Diversified behavioral portfolio as an alternative to modern portfolio theory N. Am. J. Econ. Finance 58 2021 101508 18 Harris R.D.F. Mazibas M. Portfolio optimization with behavioural preferences and investor memory Eur. J. Oper. Res. 296 2022 368 387 10.1016/j.ejor.2021.04.044 19 Bi J. Jin H. Meng Q. Behavioral mean-variance portfolio selection Eur. J. Oper. Res. 271 2018 644 663 10.1016/j.ejor.2018.05.065 20 Ayala J. Torres M.G. Noguera J.L.V. Vela F.G. Divina F. Technical analysis strategy optimization using a machine learning approach in stock market indices Knowl. Base Syst. 225 2021 107119 10.1016/j.knosys.2021.107119 21 Hafner C.M. Wang L. Dynamic Portfolio Selection with Sector-specific Regularization Econom. Statist. 2022 10.1016/j.ecosta.2022.01.001 22 Katsikis V.N. Mourtas S.D. Stanimirović P.S. Li S. Cao X. Time-varying mean–variance portfolio selection problem solving via LVI-PDNN Comput. Oper. Res. 138 2022 105582 23 Ghahtarani A. A new portfolio selection problem in bubble condition under uncertainty: application of Z-number theory and fuzzy neural network Expert Syst. Appl. 177 2021 10.1016/j.eswa.2021.114944 24 Zhang Y. Lin H. Yang X. Long W. Combining expert weights for online portfolio selection based on the gradient descent algorithm Knowl. Base Syst. 234 2021 107533 25 Gong X. Yu C. Min L. A cloud theory-based multi-objective portfolio selection model with variable risk appetite Expert Syst. Appl. 176 2021 114911 26 Ahmadi E. Jasemi M. Monplaisir L. Nabavi M.A. Mahmoodi A. Jam P.A. New efficient hybrid candlestick technical analysis model for stock market timing on the basis of the support vector machine and heuristic algorithms of imperialist competition and genetic Expert Syst. Appl. 94 2018 21 31 10.1016/j.eswa.2017.10.023 27 Chiang W.C. Enke D. Wu T. Wang R. An adaptive stock index trading decision support system Expert Syst. Appl. 59 2016 195 207 10.1016/j.eswa.2016.04.025 28 Weng B. Ahmed M.A. Megahed F.M. Stock market one-day ahead movement prediction using disparate data sources Expert Syst. Appl. 79 2017 153 163 29 Alhashel B.S. Almudhaf F.W. Hansz J.A. Can technical analysis generate superior returns in securitized property markets? Evidence from east asia markets Pac. Basin Finance J. 47 2018 92 108 10.1016/j.pacfin.2017.12.005 30 Henrique B.M. Sobreiro V.A. Kimura H. Stock price prediction using support vector regression on daily and up to the minute prices J. Finan. Data Sci. 4 3 2018 183 201 10.1016/j.jfds.2018.04.003 31 Patel J. Shah S. Thakkar P. Kotecha K. Predicting stock market index using fusion of machine learning techniques Expert Syst. Appl. 42 4 2015 2162 2172 10.1016/j.eswa.2014.10.031 32 Nakano M. Takahashi A. Takahashi S. Bitcoin technical trading with artificial neural network Phys. Stat. Mech. Appl. 510 2018 587 609 10.1016/j.physa.2018.07.017 33 Petrusheva N. Jordanoski I. Comparative analysis between the fundamental and technical analysis of stocks J. Process Manag. New Technol. 4 2 2016 26 31 34 Tsai C.F. Hsiao Y.C. Combining multiple feature selection methods for stock prediction: union, intersection, and multi-intersection approaches Decis. Support Syst. 50 1 2010 10.1016/j.dss.2010.08.028 35 Sedighi M. Jahangirnia H. Gharakhani M. Farahani Fard S. A novel hybrid model for stock price forecasting based on metaheuristics and support vector machine Data 4 2 2019 75 36 Picasso A. Merello S. Ma Y. Oneto L. Cambria E. Technical analysis and sentiment embeddings for market trend prediction Expert Syst. Appl. 135 2019 60 70 37 Ozkok F.O. Celik M. A hybrid validity index to determine K parameter value of k-means algorithm for time series clustering Int. J. Inf. Technol. Decis. Making 20 6 2021 1615 1636 38 Gubu L. Rosadi D. Abdurakhman A new approach for robust mean-variance portfolio selection using trimmed k-means clustering Indust. Engin. Manag. Syst. 2021 10.7232/iems.2021.20.4.782 39 Alwarid A. Sihabuddin A. An absolute differences K-means clustering approach on determining intervals to optimize fuzzy time series M arkov Chain model Int. J. Intellig. Engin. Syst. 20 4 2022 782 794 10.22266/IJIES2022.0228.04 40 Fang Z. Chiao C. Research on prediction and recommendation of financial stocks based on K-means clustering algorithm optimization J. Comput. Methods Sci. Eng. 2021 10.3233/JCM-204716 41 Seong N. Nam K. Predicting stock movements based on financial news with segmentation Expert Syst. Appl. 164 2021 113988 42 Gavira-Durón N. Gutierrez-Vargas O. Cruz-Aké S. Markov chain K-means cluster models and their use for companies' credit quality and default probability estimation Mathematics 9 2021 879 10.3390/math9080879 43 Bukhsuren E. Namsrai O.E. Methodology in securities portfolio selection from the stock exchange listed companies Smart Innov. Syst. Technol. 212 2021 123 132 10.1007/978-981-33-6757-9_17 44 Muangprathub J. Intarasit A. Boongasame L. Phaphoom N. Portfolio risk and return with a new simple moving average of price change ratio Wireless Pers. Commun. 115 2020 3137 3153 45 Marketwatch https://www.marketwatch.com 46 Appel G. The Moving Average Convergence-Divergence Trading Method. Signalert 1979 47 Sanel H. MACD analysis of weaknesses of the most powerful technical analysis tool Indepen. J. Manag. Prod. 7 2 2016 367 379 10.21511/imfi.17(4).2020.05 48 Kang B.K. Improving MACD technical analysis by optimizing parameters and modifying trading rules: evidence from the Japanese nikkei 225 futures market J. Risk Financ. Manag. 14 1 2021 37 10.3390/jrfm14010037 49 Aguirre A.A.A. Medina R.A.R. Méndez N.D.D. Machine learning applied in the stock market through the Moving Average Convergence Divergence (MACD) indicator Invest. Manag. Financ. Innovat. 17 4 2020 44 60 10.21511/imfi.17(4).2020.05 50 Tan P.-N. Steinbach M. Karpatne A. Kumar V. Introduction to Data Mining 2019 Pearson Education 51 Govindarajulu U. Bedi S. K-means for shared frailty models BMC Med. Res. Methodol. 22 1 2022 11 35021993 52 Sayli A. Akbulut C. Kosuta K. Multiple regression analysis system in machine learning and estimating effects of data Transformation&Min-max normalization J. Engin. Technol. Applied Sci. 3 3 2018 189 204 53 Omar T. Zohdy M. Rrushi J. Clustering application for data-driven prediction of health insurance premiums for people of different ages 2021 IEEE International Conference on Consumer Electronics (ICCE) 2021 54 Kizys R. Doering J. Juan A.A. Polat O. Calvet L. Panadero J. A simheuristic algorithm for the portfolio optimization problem with random returns and noisy covariances Comput. Oper. Res. 139 2022 105631
PMC010xxxxxx/PMC10287182.txt	==== Front Gen Hosp Psychiatry Gen Hosp Psychiatry General Hospital Psychiatry 0163-8343 1873-7714 Published by Elsevier Inc. S0163-8343(23)00104-4 10.1016/j.genhosppsych.2023.06.007 Article Women's perinatal depression: Anhedonia-related symptoms have increased in the COVID-19 pandemic Costa Raquel abc⁎ Pinto Tiago Miguel cd Conde Ana e Mesquita Ana d Motrico Emma f Figueiredo Bárbara d a EPIUnit, Instituto de Saúde Pública, Universidade do Porto, Rua das Taipas, 135, 4050-600 Porto, Portugal b Laboratório para a Investigação Integrativa e Translacional em Saúde Populacional (ITR), Porto, Portugal c Lusófona University, HEI-Lab: Digital Human-Environment Interaction Labs, Portugal d School of Psychology, University of Minho, Campus Gualtar, 4710-057 Braga, Portugal e INPP – Portucalense Institute for Human Development, Portucalense University, Porto, Portugal f Psychology Department, Universidad Loyola Andalucia, Avenida de las Universidades s/n, Dos Hermanas, Sevilla, Spain ⁎ Corresponding author at: EPIUnit, Instituto de Saúde Pública, Universidade do Porto, Rua das Taipas, 135, 4050-600 Porto, Portugal. 23 6 2023 23 6 2023 15 1 2023 15 6 2023 19 6 2023 © 2023 Published by Elsevier Inc. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background The prevalence of perinatal depression increased during the COVID-19 pandemic, which may be due to changes in the profile of specific depressive symptoms. Aims To analyze the impact of the COVID-19 pandemic on the (1) prevalence and severity of specific depressive symptoms; and on the (2) prevalence of clinically significant symptoms of depression during pregnancy and postpartum. Methods Pregnant and postpartum women recruited before (n = 2395) and during the COVID-19 pandemic (n = 1396) completed a sociodemographic and obstetric questionnaire and the Edinburgh Postnatal Depression Scale (EPDS). For each item, scores ≥1 and ≥ 2 were used to calculate the prevalence and severity of depressive symptoms, respectively. Results The prevalence and severity of symptoms of depression were significantly higher during the COVID-19 pandemic. The prevalence of specific symptoms increased by >30%, namely “being able to laugh and see the funny side of things” (pregnancy 32.6%, postpartum 40.6%), “looking forward with enjoyment to things” (pregnancy 37.2%, postpartum 47.2%); and “feelings of sadness/miserable” or “unhappiness leading to crying” during postpartum (34.2% and 30.2%, respectively). A substantial increase was observed in the severity of specific symptoms related to feelings that “things have been getting on top of me” during pregnancy and the postpartum period (19.4% and 31.6%, respectively); “feeling sad or miserable” during pregnancy (10.8%); and “feeling scared/panicky” during postpartum (21.4%). Conclusion Special attention should be paid to anhedonia-related symptoms of perinatal depression to ensure that they are adequately managed in present and future situations of crisis. Keywords COVID-19 pandemic Symptoms of depression Prevalence Severity Pregnancy Postpartum ==== Body pmc1 Introduction Mental health problems emerged as a major public health concern during the COVID-19 pandemic [1] and were considered an integral component of COVID-19 response plans by the World Health Organization (WHO) [2]. Nonetheless, out of 130 WHO member states, <1 in 5 countries allocated additional funding for mental health and psychosocial support, and >60% of antenatal or postnatal mental health services were disrupted [2]. At the same time, perinatal mental health problems increased, including clinically significant symptoms of depression [3]. 1.1 COVID-19 pandemic and perinatal symptoms of depression Systematic reviews published two to three years before the COVID-19 outbreak consistently estimated the prevalence of depressive symptoms to be 9.2% for the prenatal and 9.5% for the postpartum period in high-income countries [4], regardless of the tools used (symptom severity scales or diagnostic tools). In Portuguese cohorts, about 20.0% of pregnant women [5,6] and 15.0%–22.4% of postpartum women had clinically significant symptoms of depression [7,8]. The prevalence of perinatal symptoms of depression increased during the COVID-19 pandemic [3,9,10]. About 30% of women living in European and South American countries scored ≥13 on the Edinburgh Postnatal Depression Scale (EPDS), which indicates the presence of clinically significant symptoms of depression [10]. Specifically in Europe, a study conducted in five countries with pregnant and postpartum women reported lower prevalence of EPDS ≥13 in Switzerland (10.5%/10.4%); the Netherlands (11.5%/9.1%); and Norway (12.0%/14.6%), as compared to the UK (42.1%/42.3%) or Ireland (26.3%/24.3) [11]. The authors hypothesized that differences in prevalence rates could be explained at least partly by differences in social isolation; at the time of the study, social distancing restrictions were still in place in the UK and had recently been reduced in Ireland. 1.2 COVID-19 pandemic and specific symptoms of perinatal depression The prevalence of clinically significant symptoms of depression during the COVID-19 pandemic is alarming. It is essential to gain knowledge on the increase in the prevalence and severity of specific symptoms of depression as a result of the COVID-19 pandemic. This would enhance the understanding of the presentation of symptoms which will ensure that adequate interventions are developed to meet the women's specific needs. Exploring the presentation of perinatal depression symptoms during the COVID-19 pandemic is especially relevant, as differences have been reported in the literature in relation to depressive symptoms in the perinatal period, as compared to other life periods [12,13,14]. The Edinburgh Postnatal Depression Scale [15] is a validated instrument for screening clinically significant symptoms of depression, both in pregnant and postpartum women [16]. The EPDS provides information on symptoms of depression, categorized as symptoms of anhedonia (items 1 and 2: able to laugh/looking forward); anxiety (items 3 to 5: self-blame, anxious/worried, scared panicky); and depression (items 6 to 10: overwhelmed, difficulty sleeping, sad or miserable, unhappy/crying, self-harm) [17,18]. This scale provides detailed information on the symptoms of perinatal depression, which prevalence and severity may have increased during the COVID-19 pandemic. The results obtained may help improve the readiness of health systems to meet the mental health needs of these women, according to their individual circumstances. The aim of this study was to analyze the impact of the COVID-19 pandemic on the (1) prevalence and severity of specific symptoms of depression and on the (2) prevalence of clinically significant symptoms of depression during pregnancy and the postpartum period. This study will contribute to the literature by providing comparative data on the prevalence and severity of specific symptoms of depression before and during the COVID-19 pandemic in a large sample of women in the perinatal period. 2 Method 2.1 Procedure Data for the pre-pandemic period was collected from longitudinal cohorts of pregnant women recruited at public health centers in Northern Portugal (between 2004 and 2019). The aims and procedures of the study were explained to the participants. Written consent was obtained from pregnant women willing to participate. Data on expectant/postpartum women during the COVID-19 pandemic was collected through an online survey conducted between June 2020 and October 2020. All pregnant women (regardless of the stage of pregnancy) and mothers to infants younger than six months living in Portugal were eligible for inclusion. The aims and procedures of the study were explained to participants. Written consent was obtained prior to access being granted to the questionnaire (detailed procedures are described elsewhere; [19] reference omitted for blinded peer review). All the studies were conducted according to the Declaration of Helsinki. The study was approved by the local Ethics Committees of the participating sites. The final sample included 3791 women from the cohorts described (prior to the COVID-19 pandemic, n = 2395; during the COVID-19 pandemic, n = 1396). Participants filled in a socio-demographic and obstetric questionnaire and completed the EPDS [15] either during pregnancy or postpartum. 2.2 Measures 2.2.1 Socio-demographic and obstetric characteristics Information regarding age; country of birth (foreign vs. native); education (≤ 9 years of education vs. 10–12 years of education vs. higher education [bachelor, master's degree or postgraduate diploma]); marital status (married/cohabiting vs. single/separated/divorced/widow); employment status (employed vs. unemployed/student/housewife); history of mental health problems (yes vs. no); parity (nulliparous/primiparous vs. multiparous); and risk pregnancy (Yes vs. No; with yes including maternal age over 40 or gestational diabetes or hypertension or short cervix or other clinical condition) was collected via a self-reported questionnaire. 2.2.2 Symptoms of depression Symptoms of depression were assessed using the EPDS [15]. EPDS is a 10-item self-report scale rated on a four-point Likert-type scale (0 to 3) aimed at assessing the severity of symptoms of depression within the previous seven days. Total scores range from 0 to 30, with higher scores indicating a higher severity of symptoms of depression. The items and scoring system are as follows: “Item 1. I have been able to laugh and see the funny side of things; 0. As much as I always could, 1. Not quite so much now, 2. Definitely not so much now, 3. Not at all”; “Item 2. I have looked forward with enjoyment to things; 0. As much as I ever did, 1. Rather less than I used to, 2. Definitely less than I used to, 3. Hardly at all”; “Item 3. I have blamed myself unnecessarily when things went wrong; 0. No, never, 1. Not very often, 2. Yes, some of the time, 3. Yes, most of the time”; “Item 4. I have been anxious or worried for no good reason; 0. No, not at all, 1. Hardly ever, 2. Yes, sometimes, 3. Yes, very often”; “Item 5. I have felt scared or panicky for no very good reason; 0. No, not at all, 1. No, not much, 2. Yes, sometimes, 3. Yes, quite a lot”; “Item 6. Things have been getting on top of me; 0. No, I have been coping as well as ever, 1. No, most of the time I have coped quite well, 2. Yes, sometimes I haven't been coping as well as usual, 3. Yes, most of the time I haven't been able to cope”; “Item 7. I have been so unhappy that I have had difficulty sleeping; 0. No, not at all, 1. Not very often, 2. Yes, sometimes, 3. Yes, most of the time”; “Item 8. I have felt sad or miserable; 0 No, not at all, 1. Not very often, 2. Yes, quite often, 3. Yes, most of the time”; “Item 9. I have been so unhappy that I have been crying; 0. No, never, 1. Only occasionally, 2. Yes, quite often, 3. Yes, most of the time”; “Item 10. The thought of harming myself has occurred to me; 0. Never, 1. Hardly ever, 2. Sometimes, 3. Yes, quite often”. Two categorization processes were conducted to calculate the prevalence (absent vs. present; score = 0 vs. scores ≥1) and severity (non-severe vs. severe; scores ≤1 vs. scores ≥2) of each item. The Portuguese version of EPDS has good internal consistency [20,21]. In the cohorts assessed prior to the COVID-19 pandemic, Cronbach's α coefficient was 0.83 during pregnancy and 0.85 during the postpartum period. In the cohort assessed during the COVID-19 pandemic, Cronbach's α coefficient was 0.90 during pregnancy and 0.91 during the postpartum period. As recommended, a cut-off of 13 was used to identify women with clinically significant symptoms of depression [16]. 2.3 Statistical analyses Descriptive statistics were calculated to analyze the socio-demographic and clinical characteristics of participants. The estimated prevalence (scores ≥1) and severity (scores ≥2) of each depressive symptom were compared in the cohorts before and during the COVID-19 pandemic. Chi-square tests were performed to compare the prevalence of EPDS ≥13 in the cohorts before and during the COVID-19 pandemic. Two Multivariate Analyses of Covariance (MANCOVAs) and two Univariate Analyses of Covariance (UNIANCOVAs) were conducted to analyze the impact of the COVID-19 pandemic (before vs. during) on specific symptoms of depression during pregnancy and the postpartum period. As independent variables, the models included the period (before vs. during the COVID-19 pandemic) and the stage of pregnancy (early pregnancy –first and second trimester – vs. late pregnancy –third trimester) or stage of postpartum (early postpartum –until 12 weeks after childbirth– vs. late postpartum –>12 weeks after childbirth). The MANCOVA models included the 10 EPDS items as dependent variables, whereas the UNIANCOVA models included EPDS total score as the dependent variable. The models included potential sociodemographic, obstetric and health confounders (age, country of birth, education, marital status, employment status, history of mental health problems, parity, and risk pregnancy) as covariates. Two-step logistic regressions were performed to analyze the impact of the COVID-19 pandemic on the prevalence of clinically significant symptoms of depression (EPDS ≥13) during pregnancy and the postpartum period. The first step (enter method) was performed to control potential sociodemographic, obstetric and health confounders. As independent variables, the second step included the variables of the first step, along with the period (before vs. during the COVID-19 pandemic), and stage of pregnancy (early pregnancy –first and second trimester – vs. late pregnancy –third trimester) or stage of postpartum (early postpartum –until 12 weeks after childbirth– vs. late postpartum –>12 weeks after childbirth) and clinically significant symptoms of depression (EPDS ≥13) as dependent variable. Statistical significance was considered at p values < .05. Data were analysed with IBM SPSS 26.0 version (SPSS Inc., Chicago). 3 Results 3.1 Characteristics of participants Table 1 shows the characteristics of participants by perinatal period: pregnancy vs. postpartum. Most pregnant women were 25 to 35 years old (63.0%); primiparous (78.3%); lived with their partner (85.3%); and about half had higher education (53.6%). Near one in four had experienced a mental health problem ever in life; 20.4% had a risk pregnancy; 28.5% were unemployed; and 8.7% were foreign-born. Most postpartum women were 25 to 35 years old (63.9%), primiparous (66.6%), and lived with their partner (88.0%), and about half had a higher education (57.6%). Near one in four had had a mental health problem ever in life, 21.2% had a risk pregnancy, 18.7% were unemployed, and 8.8% were foreign-born.Table 1 Participants' characteristics according to perinatal period, before and during the COVID-19 pandemic. Table 1 Pregnancy Postpartum COVID-19 pandemic COVID-19 pandemic Before During Total Before During Total n = 1402 n (%) n = 748 n (%) N = 2150 N (%) p n = 993 n (%) n = 648 n (%) N = 1641 N (%) p Age ≤24 254 (18.5) 36 (5.1) 290 (14.0) < 0.001 156 (15.9) 24 (3.9) 180 (11.2) < 0.001 25–35 870 (63.5) 438 (62.2) 1308 (63.0) 642 (65.2) 382 (61.8) 1024 (63.9) ≥36 247 (18.0) 230 (32.7) 477 (23.0) 186 (18.9) 212 (34.3) 398 (24.8) Missing 1371 704 2075 984 618 1602 Country of birth Native 1252 (91.1) 590 (91.6) 1842 (91.3) 0.714 893 (90.8) 534 (92.1) 1427 (91.2) 0.374 Foreign 122 (8.9) 54 (8.4) 176 (8.7) 91 (9.2) 46 (7.9) 137 (8.8) Missing 28 104 132 984 580 1564 Education ≤ 9 years of education 356(26.0) 7 (1.0) 363(17.3) < 0.001 235(23.9) 8(1.3) 243(15.0) 10–12 years of education 443 (32.4) 167 (22.8) 610 (29.1) 300 (30.5) 145 (22.7) 445 (27.5) < 0.001 Higher education1 569 (41.6) 557 (76.2) 1126 (53.6) 447 (45.5) 486 (76.1) 933 (57.6) Missing 1368 731 2099 982 639 1621 Employment status Unemployed or other2 371 (27.1) 232 (31.1) 603 (28.5) 0.051 231 (23.5) 74 (11.5) 305 (18.7) < 0.001 Employed 997 (72.9) 513 (68,9) 1510 (71.5) 752 (76.5) 572 (88.5) 1324 (81.3) Missing 1368 745 2113 983 646 1629 Marital status Married/cohabiting 1100 (80.4) 690 (94.4) 1790 (85.3) < 0.001 813 (82.9) 613 (95.9) 1426 (88.0) < 0.001 Other situation3 268 (19.6) 41 (5.6) 309 (14.7) 168 (17.1) 26 (4.1) 194 (12.0) Missing 1368 731 2099 981 639 1620 History of mental health problems Yes4 326 (23.9) 207 (28.0) 533 (25.3) 0.039 233 (23.7) 180 (27.9) 413 (25.4) 0.059 No 1039 (76.1) 533 (72.0) 1572 (74.7) 750 (78.3) 466 (72.1) 1216 (74.6) Missing 1365 740 2105 983 646 1629 Parity Nulliparous/Primiparous 979 (70.3) 628 (88.1) 1607 (78.3) < 0.001 723 (73.2) 360 (56.3) 1083 (66.6) < 0.001 Multiparous 413 (29.7) 85 (11.9) 498 (23.7) 265 (26.8) 279 (43.7) 544 (33.4) Missing 1392 713 2105 988 639 1627 Risk pregnancy5 Yes 202(16.5) 199(27.0) 401(20.4) < 0.001 156(15.9) 178(29.9) 334(21.2) < 0.001 No 1025(83.5) 538(73.0) 1563(79.6) 824(84.1) 417(70.1) 1241(78.8) Missing 175 11 186 13 53 66 Notes. 1bachelor, master's degree or postgraduate diploma2Other includes student/housewife3;Other situation includes single/separated/divorced/widow4;Any mental health problem ever in life5;Risk pregnancy includes age over 40 or gestational diabetes or hypertension or short cervix or other clinical condition; COVID-19 = coronavirus disease. Both, pregnant and postpartum women assessed during the COVID-19 pandemic were more likely (ps < 0.001) to have clinically significant symptoms of depression (EPDS≥13; 24.3% and 28.2%, respectively), as compared to women assessed prior to the COVID-19 pandemic (9.0% and 6.9%, respectively). Pregnant women assessed during the COVID-19 pandemic were more likely to be older, live with their partner, have a high education, be primiparous, have a history of mental health problem, and have a risk pregnancy, as compared to pregnant women assessed prior to the COVID-19 pandemic. Postpartum women assessed during the COVID-19 pandemic were more likely to be older, live with their partner, have a higher education, be multiparous, and have a risk pregnancy. In addition, they were less likely to be unemployed, as compared to postpartum women assessed prior to the COVID-19 pandemic. 3.2 Impact of the COVID-19 pandemic on specific symptoms of depression (EPDS items and total) during pregnancy and the postpartum period Supplementary Fig. 1 displays the proportion of answer options selected per EPDS item by pregnant women assessed before vs. during the COVID-19 pandemic. The estimated prevalence of symptoms of depression in pregnant women (scores ≥1; see Fig. 1a) was higher during the COVID-19 pandemic for all items, except for items 3, 5, and 10, as compared to the pre-pandemic period. The items which prevalence increased the most (>30%) were item 1 (32.6%) and item 2 (37.2%) during the COVID-19 pandemic (49.9% and 53.3%, respectively), as compared to the pre-pandemic period (17.3% and 16.1%, respectively). The severity of symptoms of depression (scores ≥2; see Fig. 1b) was higher in all items except for item 10. The highest increases in severity were observed in items 6 (19.4%) and 8 (10.8%) during the COVID-19 pandemic (42.9% and 19.5%, respectively), as compared to the pre-pandemic period (23.5% and 8.7%, respectively).Fig. 1 Estimates and 95% CI around the probability of having a depressive symptom (score ≥ 1) before (black) and after (grey) or a severe depressive symptom (score ≥ 2) before (black) and after (grey) the COVID-19 pandemic - Pregnancy. Fig. 1 Supplementary Fig. 2 displays the proportion of answer options selected per EPDS item by postpartum women before vs. during the COVID-19 pandemic. The estimated prevalence of symptoms of depression in postpartum women (scores ≥1; see Fig. 2a) was higher during the COVID-19 pandemic for all EPDS items, as compared to the pre-pandemic period. A > 40% increase was observed in the estimated prevalence of items 1 (40.6%) and 2 (47.2%), followed by a > 30% increase in items 8 (34.2%) and 9 (30.2%) during the pandemic (56.8%, 60.0%, 68.5%, and 56.6%, respectively), as compared to the pre-pandemic period (16.2%; 12.8%, 34.3%, and 26.4% respectively). The severity of symptoms of depression (scores ≥2; see Fig. 2b) was higher in all items during the COVID-19 pandemic. The highest increases in severity were observed in items 5 (21.4%) and 6 (31.6%) during the COVID-19 pandemic (41.5% and 56.0%, respectively), as compared to the pre-pandemic (20.1% and 24.4%, respectively).Fig. 2 Estimates and 95% CI around the probability of having a depressive symptom (score ≥ 1) before (black) and after (grey) or a severe depressive symptom (score ≥ 2) before (black) and after (grey) the COVID-19 pandemic - Postpartum. Fig. 2 Significant multivariate effects of the COVID-19 pandemic were found on symptoms of depression (EPDS items) during pregnancy, Wilk's Lambda = 0.86, F(10, 1789) = 28.89, p < .001, ɳp 2 = 0.14. Significant univariate effects of the COVID-19 pandemic were found on EPDS total score and all items, except for items 3 and 10. Pregnant women assessed during the COVID-19 pandemic reported more symptoms of depression, as compared to pregnant women assessed prior to the COVID-19 pandemic (see Table 2 ). Significant multivariate effects of the COVID-19 pandemic were found on symptoms of depression (EPDS items) during postpartum, Wilk's Lambda = 0.77, F(10, 1455) = 44.66, p < .001, ɳp 2 = 0.24. Results revealed significant univariate effects of the COVID-19 pandemic on all EPDS items and the total score. Postpartum women assessed during the COVID-19 pandemic reported more symptoms of depression compared to postpartum women assessed prior to the COVID-19 pandemic (see Table 2).Table 2 The impact of COVID-19 pandemic on specific symptoms of depression (EPDS items and total) during pregnancy and the postpartum period, adjusting for sociodemographic and health confounders. Table 2 Pregnancy Postpartum COVID-19 pandemic COVID-19 pandemic Before n = 1402 During n = 748 Before n = 962 During n = 513 EPDS M SD M SD F M SD M SD F Item 1. I have been able to laugh and see the funny side of things 0.19 0.44 0.54 0.66 187.72* 0.17 0.42 0.71 0.71 211.28* Item 2. I have looked forward with enjoyment to things 0.20 0.52 0.63 0.69 185.08* 0.15 0.44 0.74 0.70 337.06* Item 3. I have blamed myself unnecessarily when things go wrong 1.15 0.82 1.15 0.88 1.47 1.10 0.80 1.40 0.92 32.55* Item 4. I have been anxious or worried for no good reason 1.35 0.88 1.42 0.86 11.55 1.14 0.83 1.48 0.88 40.01* Item 5. I have felt scared or panicky for no very good reason 1.06 0.85 1.13 0.91 7.21 0.83 0.78 1.26 0.91 63.40* Item 6. Things have been getting on top of me 0.92 0.79 1.26 0.90 54.25* 0.99 0.76 1.51 0.87 77.98* Item 7. I have been so unhappy that I have had difficulty sleeping 0.45 0.69 0.60 0.77 27.18* 0.32 0.58 0.59 0.74 50.95* Item 8. I have felt sad or miserable 0.51 0.69 0.77 0.81 54.32* 0.39 0.58 0.94 0.80 175.05* Item 9. I have been so unhappy that I have been crying 0.40 0.61 0.59 0.74 38.88* 0.29 0.52 0.71 0.73 128.17*** Item 10. The thought of harming myself has occurred to me 0.08 0.35 0.07 0.34 2.81 0.06 0.32 0.11 0.44 6.49* Total 6.30 4.28 8.20 5.74 75.62* 5.72 4.18 9.44 5.61 159.38* Notes. Response options, Item 1. 0. As much as I always could, 1. Not quite so much now, 2. Definitely not so much now, 3. Not at all; Item 2. 0. As much as I ever did, 1. Rather less than I used to, 2. Definitely less than I used to, 3. Hardly at all; Item 3. 0. No, never, 1. Not very often, 2. Yes, some of the time, 3. Yes, most of the time; Item 4. 0. No, not at all, 1. Hardly ever, 2. Yes, sometimes, 3. Yes, very often; Item 5. 0. No, not at all, 1. No, not much, 2. Yes, sometimes, 3. Yes, quite a lot; Item 6. 0. No, I have been coping as well as ever, 1. No, most of the time I have coped quite well, 2. Yes, sometimes I haven't been coping as well as usual, 3. Yes, most of the time I haven't been able to cope; Item 7. 0. No, not at all, 1. Not very often, 2. Yes, sometimes, 3. Yes, most of the time; Item 8. 0 No, not at all, 1. Not very often, 2. Yes, quite often, 3. Yes, most of the time; Item 9. 0. No, never, 1. Only occasionally, 2. Yes, quite often, 3. Yes, most of the time; Item 10. 0. Never, 1. Hardly ever, 2. Sometimes, 3. Yes, quite often. Models adjusted for age, country of birth, education, marital status, employment status, history of mental health problems, parity, and risk pregnancy. 3.3 Impact of the COVID-19 pandemic on the prevalence of clinically significant symptoms of depression (EPDS ≥ 13) during pregnancy and the postpartum period The first step of the regression model for the prevalence of clinically significant symptoms of depression (EPDS ≥13) during pregnancy included potential sociodemographic, obstetric and health confounders. The model was statistically significant (χ2(9) = 61.79, p < .001), explained 3% to 6% of the variance in the prevalence of clinically significant symptoms of depression during pregnancy (Cox & Snell R 2= 0.03, Nagelkerke R 2 = 0.06), and correctly classified 86.2% of cases. The period (before vs. during the COVID-19 pandemic) and stage of pregnancy (early vs. late) were added in the second step of the model. The second step of the regression model was statistically significant (χ2(11) = 154.33, p < .001), explained 8% to 15% of the variance in the prevalence of clinically significant symptoms of depression during pregnancy (Cox & Snell R 2 = 0.08, Nagelkerke R 2 = 0.15), and correctly classified 86.6% of cases. Pregnant women assessed during the COVID-19 pandemic had higher odds of reporting clinically significant symptoms of depression, as compared to pregnant women assessed prior to the COVID-19 pandemic. Being foreign or having a low education level or a history of mental health problems increased the odds of reporting clinically significant symptoms of depression during pregnancy (see Table 3 ).Table 3 The impact of timing (before vs during COVID-19 pandemic) on the prevalence of clinically significant symptoms of depression (EPDS ≥13) during pregnancy and the postpartum period, adjusting for sociodemographic, obstetric, and health confounders. Table 3Pregnancy β Wald p OR 95% CI Step 1 Not married/cohabiting1 (ref. married/cohabiting) 0.01 0.01 0.948 0.99 0.67–1.46 Risk pregnancy2 (ref. normal pregnancy) 0.25 2.10 0.147 0.78 0.56–1.09 Foreign (ref. native) 0.66 9.70 0.002 1.93 1.28–2.92 Unemployed or other3 (ref. employed) 0.48 10.12 0.001 1.62 1.20–2.18 ≤ 9 years of education (ref. higher education4) 0.21 0.94 0.332 0.81 0.53–1.24 10–12 years of education (ref. higher education4) 0.22 1.69 0.193 1.24 0.90–1.72 Any mental health problem ever in life (ref. no) 0.81 30.38 < 0.001 0.45 0.34–0.60 ≤ 24 years old (ref. >33 years old) 0.16 0.37 0.541 1.17 0.71–1.93 25–33 years old (ref. >33 years old) −0.18 1.10 0.294 0.84 0.60–1.17 Step 2 Not married/cohabiting1 (ref. married/cohabiting) 0.29 1.76 0.185 1.33 0.87–2.03 Risk pregnancy2 (ref. normal pregnancy) 0.04 0.06 0.814 0.96 0.68–1.36 Foreign (ref. native) 0.81 13.06 < 0.001 2.25 1.45–3.48 Unemployed or other3 (ref. employed) 0.28 3.22 0.073 1.33 0.97–1.81 ≤ 9 years of education (ref. higher education4) 0.82 10.30 < 0.001 2.27 1.38–3.75 10–12 years of education (ref. higher education4) 0.55 9.77 0.002 1.74 1.23–2.46 Any mental health problem ever in life (ref. no) 0.73 22.82 < 0.001 0.48 0.36–0.65 ≤ 24 years old (ref. >33 years old) 0.32 1.45 0.228 1.38 0.82–2.34 25–33 years old (ref. >33 years old) −0.06 0.12 0.725 0.94 0.67–1.33 During COVID-19 pandemic (ref. before) 1.60 81.35 < 0.001 0.20 0.14–0.29 Early pregnancy5 (ref. late pregnancy6) 0.36 4.91 0.027 0.70 0.51–0.96 Postpartum β Wald p OR 95% CI Step 1 Not married/cohabiting1 (ref. married/cohabiting) 0.06 0.06 0.807 0.94 0.57–1.56 Risk pregnancy2 (ref. normal pregnancy) 0.42 5.54 0.019 1.52 1.07–2.16 Foreign (ref. native) 0.54 4.84 0.028 1.71 1.06–2.77 Unemployed or other3 (ref. employed) 0.26 1.38 0.240 0.77 0.50–1.19 ≤ 9 years of education (ref. higher education4) 0.95 12.13 < 0.001 0.39 0.23–0.66 10–12 years of education (ref. higher education4) 0.43 4.99 0.026 0.65 0.44–0.95 Any mental health problem ever in life (ref. no) 0.95 33.28 < 0.001 0.39 0.28–0.53 ≤ 24 years old (ref. >33 years old) 0.54 2.70 0.100 1.72 0.90–3.27 25–33 years old (ref. >33 years old) 0.19 1.00 0.317 1.21 0.84–1.74 Step 2 Not married/cohabiting1 (ref. married/cohabiting) 0.35 1.56 0.212 1.42 0.82–2.44 Risk pregnancy2 (ref. normal pregnancy) 0.18 0.94 0.332 1.20 0.83–1.74 Foreign (ref. native) 0.70 7.03 0.008 2.01 1.20–3.36 Unemployed or other3 (ref. employed) 0.14 0.36 0.547 0.87 0.55–1.38 ≤ 9 years of education (ref. higher education4) 0.11 0.12 0.734 1.12 0.60–2.09 10–12 years of education (ref. higher education4) 0.13 0.43 0.513 0.87 0.59–1.31 Any mental health problem (ref. no) 0.85 24.04 < 0.001 0.43 0.30–0.60 ≤ 24 years old (ref. >33 years old) 0.84 5.71 0.017 2.32 1.16–4.61 25–33 years old (ref. >33 years old) 0.41 4.34 0.037 1.50 1.03–2.21 During COVID-19 pandemic (ref. before) 1.97 89.76 < 0.001 7.14 4.75–10.72 Early postpartum7 (ref. late postpartum8) 0.09 0.25 0.616 0.92 0.65–1.29 Notes. COVID-19, coronavirus disease; EPDS, Edinburgh Postnatal Depression Scale; OR, odds ratio; CI, Confidence interval1;Includes single/separated/divorced/widow2;Risk pregnancy includes age over 40 or gestational diabetes or hypertension or short cervix or other clinical condition3;Other includes student/housewife; 4bachelor, master's degree or postgraduate diploma; 5first and second pregnancy trimester; 6third pregnancy trimester; 7until 12 weeks after childbirth8; >12 weeks after childbirth. Models adjusted for age, country of birth, education, marital status, employment status, history of mental health problems, parity, and risk pregnancy. The first step of the regression model for the prevalence of clinically significant symptoms of depression (EPDS ≥13) during the postpartum period included potential sociodemographic, obstetric and health confounders. The model was statistically significant (χ2(9) = 54.02, p < .001), explained 4% to 7% of the variance in the prevalence of clinically significant symptoms of depression during the postpartum period (Cox & Snell R 2= 0.04, Nagelkerke R 2 = 0.07) and correctly classified 86.2% of cases. The period (before vs. during the COVID-19 pandemic) and stage of postpartum (early vs. late) were added in the second step. The second step was statistically significant (χ2(11) = 165.58, p < .001), explained 11% to 19% of the variance in the prevalence of clinically significant symptoms of depression during the postpartum period (Cox & Snell R 2 = 0.11, Nagelkerke R 2 = 0.19), and correctly classified 86.1% of cases. Postpartum women assessed during the COVID-19 pandemic had higher odds of reporting clinically significant symptoms of depression, as compared to postpartum women assessed prior to the COVID-19 pandemic. A foreign nationality, a history of mental health problems, and a young age increased the odds of reporting clinically significant symptoms of depression during the postpartum period (see Table 3). 4 Discussion 4.1 Main findings This study provides evidence that pregnant/postpartum women during the COVID-19 pandemic had higher odds of reporting clinically significant symptoms of depression, as compared to pregnant/postpartum women assessed prior to the COVID-19 pandemic. The increased prevalence of clinically significant symptoms of depression during the COVID-19 pandemic can be due to a significantly higher prevalence and severity of specific core depressive and anhedonia-related symptoms, as assessed on the EPDS. Additionally, being foreign or having a history of mental health problems increased the odds of having clinically significant symptoms of depression during pregnancy/postpartum. A low education only increased the odds of having clinically significant depressive symptoms during pregnancy, whereas a young age only increased the odds during the postpartum period. During the COVID-19 pandemic, the number and severity of symptoms of depression during pregnancy and the postpartum period was considerably higher, as compared to the pre-pandemic period. 4.2 Prevalence of clinically significant symptoms of depression Our data show that about one in four pregnant women obtained a score ≥ 13 on the EPDS and this prevalence was even higher in the postpartum period. Lower prevalence rates have been reported in Switzerland and the Netherlands, whereas higher prevalence rates have been reported in the UK, and similar in Ireland [11]. Importantly, the increase observed in the prevalence and severity of depressive symptoms during the pandemic was more remarkable during postpartum, as compared to the pregnancy period (21.3% vs. 15.3%, respectively). This finding suggests a higher burden of the COVID-19 pandemic on the mental health of women during the postpartum period. 4.3 Prevalence of symptoms of depression Important increases were observed in the prevalence of difficulties in being able to laugh and see the funny side of things (item 1, pregnancy 32.6%, postpartum 40.6%); looking forward with enjoyment to things (item 2, pregnancy 37.2%, postpartum 47.2%); feeling sad/miserable in the postpartum period (item 8, 34.2%); and unhappiness leading to crying in the postpartum period (item 9, 30.2%). The last two symptoms are core symptoms of depression, whereas the first two symptoms have been previously classified as anhedonia-related symptoms [17,18,22] and refer to the inability to feel pleasure in normally pleasurable activities. The increases in the prevalence of core symptoms of depression and anhedonia-related symptoms may be related to non-pharmaceutical interventions (including containment and closure policies) implemented by governments to contain the COVID-19 outbreak [23,24]. These measures had a huge impact on daily routines and prevented engagement in pleasurable activities, particularly those that involved social contact, which resulted in social isolation. In Portugal, within the study period, the Portuguese “lockdown style” was quite strict, with a high median stringency index (index varies from 0 to 100; 100 = strictest [23]) (71.30; P25-P75: 71.30–72.69; Min-Max: 63.43–74.54). This high stringency index indicates that social contact was substantially restricted in the population. Preventive measures included containment and closure policies such as school and workplace closure, cancellation of public events, restrictions on gatherings, closure of public transport, stay-at-home measures, restrictions on internal movements, or border restriction [23], which led to increased social isolation. The prevalence of anxiety-related symptoms (items 3 to 5: self-blame, anxious/worried, scared panicky) remained steadily. Those symptoms were the most prevalent in the pre-pandemic. During the COVID-19 pandemic, these symptoms remained the most prevalent along with the core depression symptom of overwhelmed (item 6), both during pregnancy and the postpartum period. This finding suggests that the symptoms that were already the most prevalent in women assessed prior to the COVID-19 pandemic may be less prone to variability resulting from contextual changes. 4.4 Severity of symptoms of depression The severity of symptoms of depression also increased significantly during the COVID-19 pandemic, especially concerning core-symptoms of depression of feeling overwhelmed –“things have been getting on top of me”– during pregnancy and the postpartum period (item 6, pregnancy 19.4%, postpartum 31.6%), and sadness during pregnancy –“I have felt sad or miserable” (item 8, 10.8%)–, as well as the anxiety-related symptom of feeling scared/panicky in the postpartum period (item 5, 21.4%). During pregnancy, severity increased most significantly in two core symptoms of depression (items 6 and 8), whereas in the postpartum period, the highest increases were observed in a core-depressive symptom and an anxiety-related symptom (items 5 and 6). The feeling of overwhelm during pregnancy and the postpartum might be associated with the significant proportion of Portuguese women reporting difficulties in accessing routine antenatal care (49.2%) and the reduced overall quality of maternal and neonatal care during the COVID-19 pandemic (47.3%) [25]. These concerns added to worries about the lack of social support (e.g., restrictions on visits) [26] resulting from non-pharmaceutical interventions (movement restrictions and stay-at-home requirements). These restrictions may have resulted in isolation and deprivation from the social support that is usually provided to women by relatives and friends around the postpartum period. A previous study showed that support from healthcare providers is associated with lower symptoms of depression [27]. Feeling scared in the postpartum period might be driven by increased worries about the infant and medical care, including the possibility that the infant will be infected with SARS-CoV-2. The proportion of pregnant women that had feelings of scariness or panic without a reason and of pregnant or postpartum women that had feelings of unnecessary blame or self-harm did not increase in the COVID-19 pandemic compared to before. The first two symptoms are anxiety-related components of depression [17,18,22]. These results suggest that the COVID-19 pandemic had a higher impact on symptoms of depression, as compared to anxiety-related symptoms in the perinatal period. Finally, the severity of anxiety-related symptoms did not increase during the COVID-19 pandemic. Those symptoms (e.g., self-blame, anxious/worried, scared panicky) were the most severe both before and during the COVID-19 pandemic, along with the core depressive symptom of feeling overwhelmed, both in pregnant and postpartum women. 4.5 Factors associated with the prevalence of clinically significant symptoms of perinatal depression during the COVID-19 pandemic Overall, pregnant/postpartum women assessed during the COVID-19 pandemic had higher odds of reporting clinically significant symptoms of depression, as compared to those assessed prior to the COVID-19 pandemic, which is consistent with previous evidence [3,9]. This is a major and serious public health concern, in the light that 60% of antenatal or postnatal mental health services were disrupted during the COVID-19 pandemic [2]. Untreated perinatal mental health problems have adverse consequences for children [28]. The finding that a young age, being foreign, a low education level or a history of mental health problems increase the risk for reporting clinically significant symptoms of depression during pregnancy and/or the postpartum period during the COVID-19 pandemic is consistent with previous reports [29]. Awareness should be raised about the mental health needs of the most vulnerable groups of women during the perinatal period. This would greatly contribute to the implementation of priority policies and promote the readiness of mental health services during the perinatal period in a context of crisis. 4.6 Strengths and limitations The EPDS was used to assess symptoms of depression during pregnancy and the postpartum period. Since it is one of the most widely used self-reported measures to assess perinatal symptoms of depression [16], it allows comparison with other studies. Considering the normative sociodemographic characteristics of the sample, generalization of findings to other populations should be dealt with caution. Significant differences were found between the cohort of women assessed during the COVID-19 pandemic and the cohort women assessed prior to the COVID-19 pandemic. However, these differences were controlled in the statistical analysis to examine the impact of the COVID-19 pandemic on perinatal symptoms of depression. 4.7 Implications for clinical practice and research A variety of reliable short versions of the EPDS have been developed [e.g. [30,31]] in order to increase the cost-effectiveness of mental health screening plans. The significant increase in the prevalence and severity of specific depressive symptoms during the COVID-19 pandemic raises concerns on the adequacy of using shortened EPDS versions during future contexts of crisis. For instance, items 5 and 6 were removed in some validated shortened versions of the EPDS [e.g. [31,32]]. However, these items represent symptoms which severity increases in situations of crisis, especially in the postpartum period. Clinicians should be aware that the use of shortened versions of the EPDS may lead to the underestimation of perinatal mental health problems. This may occur as short versions of the EPDS do not assess the symptoms which prevalence and severity increase the most during situations of crisis or when social isolation is required for any reason. Additionally, the administration of shortened versions prevents perinatal mental healthcare services from becoming aware of the occurrence of symptoms of depression in contexts of crisis or social isolation. This limitation results in health services failing to provide adequate screening, monitoring, and health care services, including strategies for reducing social isolation and improving social support, according to the specific needs of women during pregnancy or the postpartum period. We identified vulnerable groups that need the clinicians' special attention in terms of perinatal mental health. Maternal and neonatal health care services should be aware that women who are young, foreign or have a history of mental health problems or a low education level, have higher odds of developing mental health problems. Therefore, screening for mental health problems is of foremost importance to overcome inequities in health care. Funding for mental health and psychosocial support is scarce and antenatal or postnatal mental health services were disrupted during the COVID-19 pandemic [2]. At the same time our study showed that perinatal mental health worsened. If used wisely, the evidence provided in this study may contribute to improving the quality of mental healthcare by targeting and monitoring vulnerable groups using cost-effective tools. 5 Conclusion The COVID-19 pandemic had an impact on the prevalence and severity of specific symptoms of depression, as well as on the prevalence of clinically significant symptoms of depression during the perinatal period. These deleterious effects were mainly due to the significantly higher prevalence and severity of core depressive and anhedonia-related symptoms. Our results suggest that clinical attention should be focused on the specific symptoms of depression that these women experience, in order to improve screening and treatment in this and future pandemics. The use of appropriate tools that assess a broad range of symptoms and address vulnerable groups is key to improving mental health care during current and future situations of crisis. Financial disclosure statement The authors have no financial relationships relevant to this article to disclose. Ethical standards The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Funding sources This study was and supported by the Psychology Research Centre (UID/PSI/01662/2013), University of Minho, by the 10.13039/501100001871 Portuguese Foundation for Science and Technology and the 10.13039/501100005992 Portuguese Ministry of Education and Science through national funds and co-financed by FEDER through COMPETE2020 under the PT2020 Partnership Agreement, under the Grant No. POCI-01-0145-FEDER- 007653. This research was supported by the FEDER Funds through the Programa Operacional Factores de Competitividade (COMPETE) and by National Funds through FCT (10.13039/501100001871 Fundação para a Ciência e a Tecnologia ) under the Grant No. PTDC/SAU/SAP/116738/2010. Fundação Bial, under the project with the reference 157/12 and by the FCT– Fundação para a Ciência e a Tecnologia, I.P., under the projects PTDC/PSI-PCL/119152/2010, HEI-Lab R&D Unit UIDB/05380/2020, UIDB/04750/2020, and LA/P/0064/2020. It was supported by FEDER through COMPETE2020 under the PT2020 Partnership Agreement (POCI-01-0145-FEDER-007653). Raquel Costa was supported by the 10.13039/100012371 FSE and FCT under the Post-Doctoral Grant SFRH/BPD/117597/2016 [RC]. Tiago Miguel Pinto [TMP] was supported by the 10.13039/100012371 FSE and FCT under the individual grant SFRH/BD/115048/2016. Ana Conde was supported a doctoral grant for Science in Measure IV.3 and co-funded under the 2010 Science and Innovation Operational Program (POCI 2010) from Science and Technology Foundation, Government of the Portuguese Republic (Ref. SFRH/BD/13768/2003) [AC]. Ana Mesquita is supported from the Portuguese Foundation for Science and Technology (FCT) and from EU through the European Social Fund and from the Human Potential Operational Program - IF/00750/2015. This article is based upon work from COST Action CA18138 “Research Innovation and Sustainable Pan-European Network in Peripartum Depression Disorder” (Riseup-PPD), supported by 10.13039/501100000921 COST (10.13039/501100000921 European Cooperation in Science and Technology ; https://www.cost.eu/). The funders had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the article for publication. Author statement I declare that I participated in the design, execution, and analysis of the paper by Costa and colleagues entitled “Women's perinatal depression: Anhedonia-related symptoms have increased in the COVID-19 pandemic“, that I have seen and approved the final version and that it has neither been published nor submitted elsewhere. I also declare that I have no conflict of interest, other than any noted in the covering letter to the editor. Role of the funding source The funders had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the article for publication. Author contributions Costa and Pinto had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: RC TP AC BF; Acquisition, analysis, or interpretation of data: RC TP AC BF AM EM; Drafting of the manuscript: RC TP AC; Critical revision of the manuscript for important intellectual content: RC TP AC BF AM EM; Statistical analysis: RC TP; Obtained funding: RC AC BF AM. Declaration of Competing Interest None. Appendix A Supplementary data Supplementary material Image 1 Data availability Data will be made available on request. Acknowledgements We acknowledge the assistance of the psychologists and other professionals who were involved in the data collection. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.genhosppsych.2023.06.007. ==== Refs References 1 Vindegaard N. Benros M.E. COVID-19 pandemic and mental health consequences: systematic review of the current evidence Brain Behav Immun 89 2020 531 542 10.1016/j.bbi.2020.05.048 32485289 2 World Health Organization The impact of COVID-19 on mental, neurological and substance use services: results of a rapid assessment Geneva World Health Organization Licence: CC BY-NC-SA 3.0 2020 3 Chmielewska B. Barratt I. Townsend R. Kalafat E. van der Meulen J. Gurol-Urganci I. Effects of the COVID-19 pandemic on maternal and perinatal outcomes: a systematic review and meta-analysis Lancet Glob Health 9 6 2021 e759 e772 10.1016/S2214-109X(21)00079-6 33811827 4 Woody C.A. Ferrari A.J. Siskind D.J. Whiteford H.A. Harris M.G. A systematic review and meta-regression of the prevalence and incidence of perinatal depression J Affect Disord 219 2017 86 92 10.1016/j.jad.2017.05.003 28531848 5 Alves S. Fonseca A. Canavarro M.C. Pereira M. Dyadic coping and dyadic adjustment in couples with women with high symptoms of depression during pregnancy Journal of Reproductive & Infant Psychology 36 5 2018 504 518 10.1080/02646838.2018.1490496 30068221 6 Teixeira C. Figueiredo B. Conde A. Pacheco A. Costa R. Anxiety and depression during pregnancy in women and men J Affect Disord 119 1–3 2009 142 148 10.1016/j.jad.2009.03.005 19346001 7 Figueiredo B. Conde A. Anxiety and depression in women and men from early pregnancy to 3-months postpartum Arch Womens Ment Health 14 3 2011 247 255 10.1007/s00737-011-0217-3 21479759 8 Marques R. Monteiro F. Canavarro M.C. Fonseca A. The role of emotion regulation difficulties in the relationship between attachment representations and depressive and anxiety symptoms in the postpartum period J Affect Disord 238 2018 39 46 10.1016/j.jad.2018.05.013 29859386 9 Chrzan-Dętkoś M. Walczak-Kozłowska T. Lipowska M. The need for additional mental health support for women in the postpartum period in the times of epidemic crisis BMC Pregnancy Childbirth 21 1 2021 114 10.1186/s12884-021-03544-8 33557768 10 Mateus V. Cruz S. Costa R. Mesquita A. Christoforou A. Wilson C.A. Rates of depressive and anxiety symptoms in the perinatal period during the COVID-19 pandemic: comparisons between countries and with pre-pandemic data J Affect Disord 316 2022 245 253 10.1016/j.jad.2022.08.017 35964769 11 Ceulemans M. Foulon V. Ngo E. Panchaud A. Winterfeld U. Pomar L. Mental health status of pregnant and breastfeeding women during the COVID-19 pandemic-A multinational cross-sectional study Acta Obstet Gynecol Scand 100 7 2021 1219 1229 10.1111/aogs.14092 33475148 12 Hendrick V. Altshuler L. Strouse T. Grosser S. Postpartum and nonpostpartum depression: differences in presentation and response to pharmacologic treatment Depress Anxiety 11 2 2000 66 72 10.1002/(SICI)1520-6394(2000)11:2<66::AID-DA3>3.0.CO;2-D 10812531 13 Kumar R. Robson K.M. A prospective study of emotional disorders in childbearing women Br J Psychiatry 144 1 1984 35 47 10.1192/bjp.144.1.35 6692075 14 Pitt B. “Atypical” depression following childbirth Br J Psychiatry 114 516 1968 1325 1335 10.1192/bjp.114.516.1325 5750402 15 Cox J.L. Holden J.M. Sagovsky R. Detection of postnatal depression. Development of the 10-item Edinburgh postnatal depression scale Br J Psychiatry 150 6 1987 782 786 10.1192/bjp.150.6.782 3651732 16 Levis B. Negeri Z. Sun Y. Benedetti A. Thombs B.D. DEPRESsion Screening Data (DEPRESSD) EPDS Group Accuracy of the Edinburgh Postnatal Depression Scale (EPDS) for screening to detect major depression among pregnant and postpartum women: systematic review and meta-analysis of individual participant data BMJ (Clinical Research ed) 371 2020 m4022 10.1136/bmj.m4022 17 Lautarescu A. Victor S. Lau-Zhu A. Counsell S.J. Edwards A.D. Craig M.C. The factor structure of the Edinburgh Postnatal Depression Scale among perinatal high-risk and community samples in London Arch Womens Ment Health 2021 10.1007/s00737-021-01153-0 18 Tuohy A. McVey C. Subscales measuring symptoms of non-specific depression, anhedonia, and anxiety in the Edinburgh Postnatal Depression Scale Br J Clin Psychol 47 Pt 2 2008 153 169 10.1111/j.2044-8260.2008.tb00463.x 17761026 19 Omitted reference for blinded review. 20 Figueiredo B. Canário C. Tendais I. Pinto T.M. Kenny D.A. Field T. Couples' relationship affects mothers' and fathers' anxiety and depression trajectories over the transition to parenthood J Affect Disord 238 10 2018 204 212 10.1016/j.jad.2018.05.064 29886200 21 Figueiredo B. Conde A. Anxiety and depression symptoms in women and men from early pregnancy to 3-months postpartum: parity differences and effects J Affect Disord 132 1–2 2011 146 157 10.1016/j.jad.2011.02.007 21420178 22 Petrozzi A. Gagliardi L. Anxious and depressive components of Edinburgh Postnatal Depression Scale in maternal postpartum psychological problems J Perinat Med 41 4 2013 343 348 10.1515/jpm-2012-0258 23426862 23 Hale T. Angrist N. Goldszmidt R. Kira B. Petherick A. Phillips T. A global panel database of pandemic policies (Oxford COVID-19 government response tracker) Nat Hum Behav 5 4 2021 529 538 10.1038/s41562-021-01079-8 33686204 24 Kucharski A.J. Klepac P. Conlan A. Kissler S.M. Tang M.L. Fry H. Effectiveness of isolation, testing, contact tracing, and physical distancing on reducing transmission of SARS-CoV-2 in different settings: a mathematical modelling study Lancet Infect Dis 20 10 2020 1151 1160 10.1016/S1473-3099(20)30457-6 32559451 25 Lazzerini M. Covi B. Mariani I. Drglin Z. Arendt M. Hersoug Nedberg I. Quality of facility-based maternal and newborn care during the COVID-19 pandemic: a multicountry cross-sectional study in the WHO European region investigating service users' perspectives Lancet Regional Health – Europe. 3 2021 100268 10.1016/j.lanepe.2021.100268 26 Basu A. Kim H.H. Basaldua R. Choi K.W. Charron L. Kelsall N. A cross-national study of factors associated with women's perinatal mental health and wellbeing during the COVID-19 pandemic PloS One 16 4 2021 e0249780 10.1371/journal.pone.0249780 27 Ostacoli L. Cosma S. Bevilacqua F. Berchialla P. Bovetti M. Carosso A.R. Psychosocial factors associated with postpartum psychological distress during the Covid-19 pandemic: a cross-sectional study BMC Pregnancy Childbirth 20 1 2020 703 10.1186/s12884-020-03399-5 33208115 28 Gentile S. Untreated depression during pregnancy: short- and long-term effects in offspring. A systematic review Neuroscience 342 2017 154 166 10.1016/j.neuroscience.2015.09.001 26343292 29 Liu C.H. Erdei C. Mittal L. Risk factors for depression, anxiety, and PTSD symptoms in perinatal women during the COVID-19 pandemic Psychiatry Res 295 2021 113552 10.1016/j.psychres.2020.113552 30 Choi S.K. Kim J.J. Park Y.G. Ko H.S. Park I.Y. Shin J.C. The simplified Edinburgh Postnatal Depression Scale (EPDS) for antenatal depression: is it a valid measure for pre-screening? Int J Med Sci 9 1 2012 40 46 10.7150/ijms.9.40 22211088 31 Harel D. Levis B. Ishihara M. Levis A.W. Vigod S.N. Howard L.M. Shortening the Edinburgh postnatal depression scale using optimal test assembly methods: development of the EPDS-Dep-5 Acta Psychiatr Scand 143 4 2021 348 362 10.1111/acps.13272 33354768 32 Eberhard-Gran M. Eskild A. Samuelsen S.O. Tambs K. A short matrix-version of the Edinburgh Depression Scale Acta Psychiatr Scand 116 3 2007 195 200 10.1111/j.1600-0447.2006.00934.x 17655561
PMC010xxxxxx/PMC10287183.txt	==== Front Antiviral Res Antiviral Res Antiviral Research 0166-3542 1872-9096 Elsevier B.V. S0166-3542(23)00133-X 10.1016/j.antiviral.2023.105655 105655 Article Cathepsin inhibitors nitroxoline and its derivatives inhibit SARS-CoV-2 infection Milan Bonotto Rafaela a1 Mitrović Ana bc1 Sosič Izidor c Martinez-Orellana Pamela a Dattola Federica a Gobec Stanislav c Kos Janko bc∗∗ Marcello Alessandro a∗ a Laboratory of Molecular Virology, The International Centre for Genetic Engineering and Biotechnology (ICGEB), Padriciano, 99, 34149, Trieste, Italy b Department of Biotechnology, Jožef Stefan Institute, Jamova 39, 1000, Ljubljana, Slovenia c Faculty of Pharmacy, University of Ljubljana, Aškerčeva cesta 7, 1000, Ljubljana, Slovenia ∗ Corresponding author. Laboratory of Molecular Virology, International Centre for Genetic Engineering and Biotechnology (ICGEB), Padriciano, 99, 34149, Trieste, Italy. ∗∗ Corresponding author. Faculty of Pharmacy, University of Ljubljana, Aškerčeva cesta 7, 1000, Ljubljana, Slovenia, . 1 Contributed equally to this work. 23 6 2023 23 6 2023 10565526 1 2023 13 6 2023 14 6 2023 © 2023 Elsevier B.V. All rights reserved. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The severity of the SARS-CoV-2 pandemic and the recurring (re)emergence of viruses prompted the development of new therapeutic approaches that target viral and host factors crucial for viral infection. Among them, host peptidases cathepsins B and L have been described as essential enzymes during SARS-CoV-2 entry. In this study, we evaluated the effect of potent selective cathepsin inhibitors as antiviral agents. We demonstrated that selective cathepsin B inhibitors, such as the antimicrobial agent nitroxoline and its derivatives, impair SARS-CoV-2 infection in vitro. Antiviral activity observed at early stage of virus entry was cell-type dependent and correlated well with the intracellular content and enzymatic function of cathepsins B or L. Furthermore, tested inhibitors were effective against the ancestral SARS-CoV-2 D614 as well as against the more recent BA.1_4 (Omicron). Taken together, our results highlight the important role of host cysteine cathepsin B in SARS-CoV-2 virus entry and show that cathepsin-specific inhibitors, such as nitroxoline and its derivatives, could be used to treat COVID-19. Finally, these results also suggest that nitroxoline has potential to be further explored as repurposed drug in antiviral therapy. Keywords SARS-CoV-2 COVID-19 Coronavirus Cathepsin Nitroxoline Drug repurposing Inhibition ==== Body pmc1 Introduction Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is the aetiologic agent of Coronavirus Disease (COVID-19). Since its first appearance at the end of 2019, several successive waves of infections have had a tremendous impact on the lives of millions of people worldwide, causing the most severe pandemics in recent history (CoVariants, 2021, 2022). Notwithstanding, the availability of effective vaccines and monoclonal antibodies, the ongoing evolution of the virus continuously works to evade the immune response. Specific antiviral drugs have been developed, such as the clinically approved molnupiravir and remdesivir, which target the viral polymerase, and the paxlovid, which targets viral proteases (Gottlieb et al., 2022; Jayk Bernal et al., 2022; Kokic et al., 2021; Marzi et al., 2022). However, continued efforts are needed to increase the availability of broad-spectrum antivirals that can act against different virus strains for future emerging or re-emerging epidemics (Dolgin, 2021; Liu et al., 2020; Zakaria et al., 2018). Host cell peptidases are critical in various steps of virion entry into host cells (Pišlar et al., 2020). Entry of SARS-CoV-2 occurs by binding of the Spike (S) protein to the cellular receptor angiotensin converting enzyme 2 (ACE2) and activation for membrane fusion by the transmembrane peptidase/serine subfamily member 2 (TMPRSS2), or by other transmembrane serine peptidases at the cell surface. Alternatively, the virus enters by receptor-mediated endocytosis, in which host endosomal/lysosomal cysteine peptidases, such as cathepsins B (CatB) and L (CatL), are involved in S protein activation (Evans and Liu, 2021; Pišlar et al., 2020; Shang et al., 2020). The main function of CatL and CatB is intracellular protein catabolism in endosomal/lysosomal compartments (Turk et al., 2000). However, a number of more specific functions have been identified, often associated with pathological processes, including cancer, neurodegeneration, and viral infections (Simmons et al., 2013). The involvement of cysteine cathepsins in virion entry has been previously described for several other non-enveloped and enveloped viruses (Evans and Liu, 2021; Pišlar et al., 2020; Shang et al., 2020) and confirmed also for SARS-CoV-2 (Hoffmann et al., 2020). Among host cysteine peptidases, CatL is the most commonly associated with viral glycoprotein activation due to its ability to process the SARS-CoV-2 S protein and related proteins of other human CoVs (Gomes et al., 2020; Kawase et al., 2009; Liu et al., 2020; Shirato et al., 2013; Simmons et al., 2005). CatL has been identified as an essential lysosomal peptidase for virion entry into cells stably expressing recombinant human ACE2 (Ou et al., 2020). CatB has been shown to play a role in Ebola virus (Chandran et al., 2005), Nipah virus (Diederich et al., 2008), and feline CoV (Regan et al., 2008) entry through catalytic activation of viral membrane glycoproteins. Investigation of the role of CatB and CatL in various pathologies resulted in development of a large number of inhibitors (Dana and Pathak, 2020; Kos et al., 2014). We evaluated the well-established antimicrobial agent nitroxoline, identified by our group as a potent selective inhibitor of CatB (Mirković et al., 2011, 2015), for its activity against SARS-CoV-2. In this study, we show that nitroxoline and its derivatives, which are specific for CatB, potently prevent SARS-CoV-2 entry into the host cells. Antiviral activity was dependent on cell type and correlated well with the intracellular amount and activity of the targeted cathepsin. Thus, we demonstrate the critical role of the host CatB in SARS-CoV-2 infection and suggest that nitroxoline can be repurposed as potential therapeutic agents for clinical intervention against SARS-CoV-2. 2 Materials and methods 2.1 Cells and virus Vero E6 cells (ATCC-1586), Huh7 cells expressing hACE2 (Huh7-hACE2) (Milani et al., 2021), and human lung adenocarcinoma cells Calu-3 (ATCC HTB-55) were cultured in Dulbecco's modified Eagle's medium (DMEM, ThermoFisher, Paisley, UK) supplemented with 10% foetal bovine serum (FBS, ThermoFisher, Paisley, UK). All cell lines were obtained from the American Type Culture Collection (ATCC). Human Nasal Epithelial cells (HNEpC), PromoCell cat number c-12620, were grown in Airway Epithelial cell Growth Medium (ready-to-use) PromoCell cat number C-21060. The same protocol was used for Human Pulmonary Alveolar Epithelial Cells (HPAEpC) Catalog #3200 ScienCell Research Laboratories growing in complete PneumaCult-ALI-S cat number #05051 Stemcell technologies. Working strains of SARS-CoV-2 ICGEB-FVG_5 (Licastro et al., 2020) (ancestral strain with D614G mutation) and the Omicron variant SARS-CoV-2 BA.1_4 isolated in Trieste, Italy, were routinely propagated and titrated as described elsewhere (Milani et al., 2021). 2.2 Inhibitors Hydroxychloroquine, nitroxoline, E−64 and E−64d were purchased from Sigma-Aldrich (St. Luis, USA). Nitroxoline derivatives compounds 3 (Mirković et al., 2011) and 17 (Sosič et al., 2013) were synthesised as reported. Their purity was determined by 1H NMR and 13C NMR spectroscopy and by CHN elemental analysis and tested for specificity against other cysteine cathepsins (Mirković et al., 2011; Sosič et al., 2013). CLIK-148 was kindly provided by prof. Nobuhiko Katunuma (The University of Tokushima, Tokushima, Japan). Compound GCV-5 (Mitrović et al., 2022) was obtained from MolPort (Riga, Latvia). The structures are shown in Fig. S1. The compounds were dissolved in DMSO, expect for hydroxychloroquine (HCQ), which was prepared in water, and stored at −20 °C until use. Physicochemical properties of compounds 3 and 17, such as efflux ratio, albumin binding, logP and solubility were determined using QikProp (Schrödinger Suite, 2020–2, Schrödinger, LLC, New York, NY, 2020) and are presented in the Supplementary Information. 2.3 High content assay (HCA) Drugs were tested for antiviral activity with a previously optimized protocol on Huh7 cells expressing hACE2 (Milani et al., 2021). briefly, cells were treated and simultaneously infected with virus for 24 hours on 96-wells plates and processed for immunofluorescence with antibodies against SARS-CoV-2 Spike or Nucleocapsid. Images were automatically acquired, and the total number of cells and the number of infected cells were analysed by hogh-contnt imaging. Extended protocols are available in Supplementary Methods. 2.4 Virological methods Vero E6 cells were used for plaque assay using 1.5% carboxymethylcellulose (CMC, Sigma-Aldrich, St Louis, USA) as previously described (Rajasekharan et al., 2021). To calculate viral yields, infected cells’ medium was collected after 24 h and the viral titre for each concentration was determined by the plaque assay. Viability assay was performed using the Alamar Blue reagent according to manufacturer specifications (Invitrogen, Waltham, MA, USA). Time of addition studies (TOA) were performed on infected Huh7-hACE2 as described (Milan Bonotto et al., 2022). Briefly, drugs were added 3 hours before infection, or during infection, and washed away or added after 1, 3, 5 hours post-infection and in all cases kept until harvest after 24 h. Binding of the virus to cells was conducted at 4 °C for two hours in the presence of the drug as described (Milan Bonotto et al., 2022). To measure entry cells were washed after the binding step and further incubate at 37 °C for 3 h before harvest. Extended protocols are available in Supplementary Methods. 2.5 Protein preparation, Western blot analysis and ELISA Cells were detached from culture flasks, lysed, and processed for Western blotting as previously described (Mitrović et al., 2016). ELISA was performed as previously described (Kos et al., 2005). Extended protocols are available in Supplementary Methods. 2.6 Determination of cathepsin activity CatB activity was measured using specific fluorogenic substrates as described in detail in extended protocols available in Supplementary Methods. 2.7 Statistical analysis The sigmoidal dose-response curve function (variable slope) was used to calculate the half maximal effective (EC50) and cytotoxic (CC50) concentrations was calculated using GraphPad Prism version 7 (GraphPad Software, San Diego, USA). The ratio between CC50 and EC50 determines the selective index (SI). To calculate the statistical significance (P < 0.05) of infected and DMSO treated samples a two-way analysis of variance (ANOVA) with Sidak's test, a multiple comparison test, was conducted. All data were plotted and analysed using GraphPad Prism software version 7. 3 Results 3.1 Screening of a panel of cathepsin inhibitors against SARS-CoV-2 Screening of cathepsin inhibitors was performed using a previously developed High Content assay (HCA) (Milani et al., 2021). The control hydroxychloroquine (HCQ) was active with an inhibitory concentration EC50 of 4.75 ± 0.35 μM. Nitroxoline and compound 17 inhibited more than 90% of SARS-CoV-2 infection at the highest concentration tested (100 μM) with an EC50 of 2 ± 0.003 μM and 2.8 ± 0.048 μM, respectively (Table 1 , Fig. 1 ). Compound 3, that inhibits both CatB and CatL (our unpublished data), showed a higher EC50 value (4.5 ± 2.1 μM). Surprisingly, the general cathepsin epoxy-succinyl irreversible inhibitor E−64 inhibited the virus only at high concentrations (>70 μM), whereas its cell-permeable analogue E−64d showed EC50 = 1.9 ± 0.10 μM (Table 1). However, neither E−64 nor E−64d reached >90% inhibition (Fig. S2). Compound CLIK-148, a specific CatL inhibitor, provided EC50 value of 8.75 ± 6.2 μM and reached a maximum of 64% antiviral activity at the highest concentration tested. In contrast, GCV-5, another CatL inhibitor, failed to provide EC50 values and showed no antiviral activity against SARS-CoV-2. All compounds were not toxic to cells under conditions tested, except for GCV-5. These results demonstrate that CatB inhibitors are most potent against SARS-CoV-2 in Huh7-hACE2 cells.Table 1 Dose-response results from cathepsin inhibitors in Huh7-hACE2 infected with SARS-CoV2 D614G. Table 1 EC50a μM (mean ± SD b) EC90c μM (mean ± SD) CC50d μM Highest % of Virus Inhibition Nitroxoline 2.1 ± 0.003 9.9 ± 8 >100 99 Compound 17 2.8 ± 0.04 6.4 ± 0.3 >100 98 Compound 03 4.5 ± 2.1 11.9 ± 0.9 >100 80 E64 73 ± 8.3 n.a >100 61 E64-d 1.9 ± 0.10 n.a >100 79 CLIK 148 8.75 ± 6.2 n.a >100 64 GCV-5 n.a* n.a* n.a* n.a* Hydroxychloroquine 4.7 ± 0.35 18.9 ± 2.3 >100 98 n.a not assessable. a Half maximal effective concentration from values obtained in HCA. b Standard deviation. c Maximal effective concentration from values obtain in HCA. d Half maximal cytotoxic concentration from values obtained in HCA. e Reference compound, positive control. Fig. 1 Nitroxoline, compound 17 and compound 3 efficiently inhibit SARS-CoV2 D614G infection in Huh7-hACE2 cells. (A) Representative images from High Content assay (HCA) showing infection of SARS-CoV-2 D614G in Huh7-hACE2 at 24 hpi. Cells were infected and treated with the inhibitors simultaneously for 24 h. Then cells were fixed and processed for immunofluorescence with the anti-nucleocapsid antibody (green fluorescence). Nuclei were stained with DAPI (red fluorescence). (B) Huh7-hACE2 were treated with 2-fold dilutions of compound and simultaneously infected with SARS-CoV-2 D614G at MOI 0.1. Hydroxychloroquine (HCQ) was used as the reference compound (positive control). Cells were processed for high-content image analysis as described. The percentage of inhibition (black dots) was normalized by the average infection ratio of cells treated with 1% DMSO (vehicle). The percentage of nuclei (blue triangles) was normalized with the total number of cells. Fig. 1 3.2 Cathepsin inhibitors show cell-dependent activity against SARS-CoV-2 Cathepsins are host peptidases and their expression and activity varies depending on the cell type (Yadati et al., 2020). SARS-CoV-2 replicates efficiently in lung- and kidney-derived cells, such as human adenocarcinoma Calu-3 cells and Vero E6 cells (Cagno, 2020). To determine whether the selected inhibitors retained their activity in the aforementioned cell lines, we tested nitroxoline, compound 17, compound 3, and GCV-5 in Calu-3 cells and Vero E6 cells using the viral yield reduction assay. The results were compared with those obtained in Huh7-hACE2 cells (Table 2 ). Both CatB inhibitors, nitroxoline and compound 17, showed a decrease of SARS-CoV-2 infection in Calu-3 and Vero E6 cells (Fig. 2 ). Nitroxoline showed an EC50 value of 18.6 ± 10 μM for Vero E6 cells and an EC50 value of 23.3 ± 4.3 μM for Calu-3 cells, which was 10-fold higher than in Huh7-hACE2 (EC50 1.9 ± 1.0 μM). Similarly, compound 17, a more potent and selective inhibitor of CatB, showed very similar activity in Vero E6 and Calu-3 cells with EC50 values of 13.5 ± 0.6 μM and 14.4 ± 2.7 μM, respectively, however the activity was still lower compared to that of Huh7-hACE2 cells (EC50 2.3 ± 0.4 μM) (Fig. 2).Table 2 Antiviral activity of cathepsin inhibitors at different cells lines. Table 2 Calu-3a Huh7-hACE2 Vero E6 EC50b μM (mean ± SD b) EC90c μM (mean ± SD) CC50d μM EC50 μM (mean ± SD) EC90 μM (mean ± SD) CC50 μM EC50 μM (mean ± SD) EC90 μM (mean ± SD) CC50 μM Nitroxoline 23.3 ± 4.3 n.a >100 1.9 ± 1.0 3.5 ± 0.4 >100 18.6 ± 10 52.4 ± 3.5 >100 Compound 17 14.4 ± 2.7 46.8 ± 4.4 >100 2.3 ± 0.4 4.0 ± 0.3 >100 13.5 ± 0.6 24.4 ± 0.0 >100 Compound 3 12.45 ± 0.07 18.8 ± 6.0 >100 10.7 ± 2.0 18.0 ± 7.5 >100 37.7 ± 10 n.a* >100 GCV-5 15.5 ± 13 50.0 ± 0.0 >100 n.a* n.a* >100 21.9 ± 2.2 27.1 ± 1.2 >100 n.a: not assessable. cStandard deviation. a Cell type used to measure compound activity by the virus yield reduction assay. b Half maximal effective concentration obtained by the virus yield reduction assay. c Maximal effective concentration obtained by the virus yield reduction assay. d Half maximal cytotoxicity concentration obtained by the viability assay Alarmar Blue. Fig. 2 Activity of cathepsin inhibitors against SARS-CoV-2 D614G in different cells. Cells were treated with 2-fold dilutions of the inhibitors within a range of concentrations of 50 to 0.78 μM. The percentage of inhibition of virus yield was quantified by plaque assay. The black, blue, and red curves represent inhibition in Huh7-hACE2, Calu-3 and Vero E6 cells. Fig. 2 Compound 3 inhibited SARS-CoV-2 infection in Vero E6 cells (Fig. 3 A) less effectively than in Calu-3 cells (Fig. 3B), with EC50 values of 37.7 ± 10 μM and 12.4 ± 0.07, respectively (Table 2). Compound 3 is a less selective CatB inhibitor that targets other cathepsins including CatL and CatH, (our unpublished data) (Mirković et al., 2011; Mitrović et al., 2016). EC50 values for nitroxoline were comparable in Calu-3 and Vero E6 cells, although an order of magnitude higher than in Huh7-ACE2 cells. While for compound 3 EC50 values for Huh7-hACE2 and Calu-3 cells were at the same range and were higher for Vero E6 cells Table 2). Additionally, to evaluate the contribution of CatL in these cells, we used CatL inhibitor GCV-5. The results showed poor EC50 values for GCV-5. However, the compound still reduced by 100% the infection in Vero E6 cells and Calu-3 cells, whereas not even 50% reduction of the infection could be achieved in Huh7-hACE2 (Fig. 2). Cell cytotoxicity was not detected at these concentrations (Fig. S3). These data confirm that sensitivity to cathepsin inhibitors during SARS-CoV-2 infection depends on the cell type.Fig. 3 Expression and activity levels of CatB and CatL in different cell lines. (A) CatB and CatL protein levels in whole-cell lysates of Vero E6, Huh7-hACE2, and Calu-3 cells as determined by Western blot analysis. Histograms show relative protein levels of CatB and CatL, with the contribution of each form indicated. Data are presented as mean ± SD. (B) CatB and CatL protein levels in whole cell lysates as determined by ELISA. Data are given as mean ± SD. (C) Activity of CatB and CatL in whole cell lysates assessed by enzyme kinetics using fluorogenic substrates. For specific assessment of CatL activity, CatB specific inhibitor CA-074 (10 μM) was added to exclude contribution of CatB to the degradation of the Z-Phe-Arg-AMC substrate. Data are presented as mean ± SD. P < 0.01. Fig. 3 Nitroxoline, a drug with repurposing potential, was also tested in human primary cells derived from the upper or lower respiratory tract (Supplementary Fig. S4). In both HNEpC and HPAEpC antiviral activity was observed in the low micromolar (EC50 ∼5μM) with cytotoxicity at lower concentrations compared to cell lines (CC50 > 50 μM). 3.3 Expression and activity levels of CatB and CatL in different cells To explain the antiviral effect of the inhibitors on selected cell lines, the protein levels and activity of CatB and CatL were determined in Vero E6, Huh7-hACE2, and Calu-3 cells. As shown by Western blot analysis (Fig. 3A) and ELISA (Fig. 3B), all cell lines contain CatB. Protein level of CatB was the highest in Calu-3 cells, followed by Huh7-hACE2 and Vero E6 cells. In all cell lines, CatB is present in all three forms: pro-CatB, active single-chain CatB and double-chain CatB. In line with this, the highest endo- and exopeptidase activity of CatB was observed in Calu-3 cells (Fig. 3C). In Huh7-hACE2 cells, lower endo- and exopeptidase activity of CatB correlated with lower protein levels, whereas in Vero E6 cells, higher activity of CatB was observed than would have been expected from the protein levels (Fig. 3C). The enzyme kinetics results show that CatB predominantly acts as endopeptidase in Vero E6 cells. The highest protein levels as well as the highest activity for CatL were observed in Vero E6 cells, compared to other two cell lines (Fig. 3). The fluorogenic substrate Z-Phe-Arg-AMC was used to measure CatL activity. However, as this substrate is not specific for CatL but can also be degraded by CatB, the potent CatB-specific inhibitor CA-074 was added to the cell lysates in the assay to distinguish between substrate degradation due to CatL or due to CatB (Fig. 3C). In Huh7-hACE2 and Calu-3 cells, in which higher protein levels and higher activity of CatB were detected, substrate degradation was almost completely abolished by the addition of CA-074, suggesting that Z-Phe-Arg-AMC substrate was degraded predominantly by CatB in these cells. In contrast, addition of CA-074 to Vero E6 cells’ lysates did not reduce substrate degradation, indicating that CatL activity is responsible for substrate degradation. Considering the high levels of CatL protein and activity in Vero E6 cells compared to CatB, CatL might contribute in part to CatB substrate degradation. This could explain the apparent high activity of CatB observed in Vero E6 cells, which does not match protein concentration. These results are consistent with the antiviral effect of CatL inhibitors in Vero E6 cells, whereas their impact in other cell lines is weaker (Fig. 2). These data suggest that SARS-CoV-2 may take advantage of different cathepsins for cell entry depending on their expression and activity. 3.4 The antiviral activity of cathepsin inhibitors is limited to early stages of viral infection Cathepsins are responsible for processing and cleavage of viral S protein in endosomes during entry, allowing the virus to release the nucleocapsid into the cytoplasm (Jackson et al., 2022; Padmanabhan et al., 2020; Pišlar et al., 2020; Schornberg et al., 2006). To assess the mode of action of cathepsin inhibitors, a time of addition experiment was performed. Huh7-hACE2 cells were treated at different times before or during infection (−3 h, 0 h, +1 h, +3 h, +5 h, and +18 h) (Fig. 4 A). The compounds that showed the highest antiviral activity in Huh7-hACE2 (Fig. 1), such as nitroxoline, compound 17 and compound 3, were selected to perform the time of addition studies and used at a final concentration of 10 μM.Fig. 4 Evaluation of the effect of cathepsin inhibitors at different stages of SARS-CoV2 D614G infection in Huh7-hACE2 cells. (A) Diagram of the experimental setup showing the time points when the compounds were added at a concentration of 10 μM, SARS-CoV-2 was inoculated at time 0 for 1 hour at MOI 0.1. (B) Viral yields for nitroxoline, compound 17 and compound 3. DMSO-treated samples were used as control. Bars represent mean ± SD of two replicates. P ≤ 0.05, P ≤ 0.01, P ≤ 0.001, **P ≤ 0.0001 (two-way ANOVA, followed by Sidak's test). Fig. 4 All compounds significantly impaired viral infection when they were added shortly after infection (+1 h) and for up to 5 h after infection, consistent with cathepsins’ activity in the early stages of infection (Fig. 4B). Compound 3 showed the impact also when it was added before infection and at the time of infection. Furthermore, compound 17 impaired viral infection when it was added at the time of infection and also maintained antiviral activity at a later stage of infection as it significantly impaired viral infection also 18 h post-infection. The dosage regimen for nitroxoline treatment of urinal infections maintains active plasma concentrations of 10 μM (Mrhar et al., 1979; Naber et al., 2014; Wagenlehner et al., 2014; Wijma et al., 2018), which matches the EC90 values obtained thus ensuring the correct dosage also for the inhibition of SARS-CoV-2 infection. A binding assay was performed to distinguish between virus adsorption and the later endocytosis and fusion steps. As shown in Fig. 5 A, virus absorption at 4 °C was not affected by the compounds (Fig. 5A). Conversely, the entry assay showed that nitroxoline, compounds 3 and 17, inhibited this step similarly to HCQ, a known inhibitor of virus endocytosis (Fig. 5B). These data confirm the inhibitory effect of inhibitors during virus entry.Fig. 5 Effect of nitroxoline, compound 17, and compound 3 on the binding or endocytosis/fusion of SARS-CoV-2. (A) Binding assay monitored on Huh7-hACE2 treated with the viral inoculum (SARS-CoV-2 at MOI 2) and 20 μM cathepsin inhibitors, 20 μM HCQ (positive control) or 1% DMSO as a negative control for 2 h at 4 °C. Cells were then washed with ice-cold PBS and extracted to determine viral RNA by real time (RT)qPCR. The housekeeping gene GAPDH was used as a reference. Viral genomes are expressed as percentage of relative expression (double-delta Ct method) compared with the negative control (DMSO without transduction). Bars represent mean ± SD from two replicates of independent experiment. P ≤ 0.033, P ≤ 0.002, P ≤ 0.001 (one-way ANOVA followed by Dunnett's test). (B) Entry assay monitored on Huh7-hACE2 treated with the viral inoculum (SARS-CoV-2 at MOI 2) at 4 °C for 2 h. After 2 h, cells were washed with PBS and supplemented with fresh medium containing 2% FBS and 20 μM cathepsin inhibitors, 20 μM HQC or 1% DMSO. Cells were maintained at 37 °C for 3 h. Cells were then processed as described above (Fig. 5A). Fig. 5 3.5 Activity of CatB and CatL inhibitors against the SARS-CoV2 Omicron variant Since the end of 2021, the SARS-CoV-2 Omicron variant B.1 and its sub-lineages are prevalent worldwide (2022; WHO, 2021). Previous work has shown that the SARS-CoV-2 B.1 variant enters cells via the endocytosis pathway, independent of the TMPRRS2 protease. In this context, cathepsins play an important role in SARS-CoV-2 infection (Meng et al., 2022; Willett et al., 2022; Zhao et al., 2022). Therefore, we evaluated the inhibitory effect of cathepsin inhibitors against SARS-CoV-2 Omicron variant B.1. Effect of nitroxoline, compound 17, compound 3, GCV-5, E−64 and its cell permeable derivative E−64d was evaluated using HCA on Huh7-hACE2. After 72 h, infection was measured by assessing nucleocapsid structural protein (N) expression (Fig. 6 A). Nitroxoline, compound 17, and compound 3 showed an EC50 values 4.1 ± 0.7 μM, 6.6 ± 0.4 μM and 5.4 ± 0.8 μM, respectively (Table 3 ). Interestingly general cathepsin inhibitors, such as E−64 and E−64d also demonstrated a dose-response antiviral activity against this viral variant with EC50 values 4.3 ± 1.2 μM and 1.1 ± 0,44 μM, respectively. However, the specific CatL inhibitor, GCV-5, was confirmed cytotoxic in these cells. These data indicate that the cathepsin inhibitors are also effective against the Omicron variant and suggest that the virus evolved to modify the use of cathepsins to promote cell entry.Fig. 6 Cathepsin inhibitors are effective against the SARS-CoV-2 Omicron variant. (A) Representative images from High Content assay (HCA) showing infection of SARS-CoV-2 Omicron in Huh7-hACE2 at 24 h. Cells were infected and treated with the inhibitors simultaneously for 24 h. Then cells were fixed and processed for immunofluorescence with the anti-nucleocapsid antibody (green fluorescence). Nuclei were stained with DAPI (red fluorescence). Plates from HCA were imaged, and image analysis was performed using the High Content System microscope. (B) Dose-response of cathepsin inhibitors against SARS-CoV-2 Omicron in Huh7-hACE2. Hydroxychloroquine (HQC) was used as a reference compound. Plates from HCA were acquired, and image analysis was performed using the High Content System microscope. The percentage of inhibition (black dots) was calculated from image analysis after normalization with the average infection ratio of cells treated with 1% DMSO. The percentage of nuclei (blue triangles) was normalized by the total number of cells. Fig. 6 Table 3 Dose-response results from cathepsin inhibitors in Huh7-hACE2 infected with SARS-CoV-2 Omicron. Table 3 EC50a μM (mean ± SD b) EC90c μM (mean ± SD) CC50d μM Highest % of Virus Inhibition Nitroxoline 4.1 ± 0.7 6.4 ± 0.6 >100 100 Compound 17 6.6 ± 0.4 14.7 ± 9.8 >100 98 Compound 03 5.4 ± 0.8 6.8 ± 0.4 >100 100 E64 4.3 ± 1.16 14.9 ± 0.0 >100 91 E64-d 1.1 ± 0.44 n.a >100 83 GVC-5 n.a* n.a* n.a* n.a* Hidroxychloroquinee 2.6 ± 0.36 7.7 ± 1.2 >100 99 *n.a not assessable. a Half maximal effective concentration from values obtained in HCA. b Standard deviation. c Maximal effective concentration from values obtained in HCA. d Half maximal cytotoxic concentration from values obtained in HCA. e Reference compound, positive control. 4 Discussion Drugs that target viral and host factors essential for virus entry, replication, and spread are in need to overcome new pandemic threats (Chitsike and Duerksen-Hughes, 2021; Fang, 2022). Cathepsins, have been identified as important host factors involved in the endosome-dependent entry pathway of viruses, including SARS-CoV-2 (Evans and Liu, 2021; Hoffmann et al., 2020; Pišlar et al., 2020; Shang et al., 2020). Cathepsins are druggable enzymes and we identified the antibiotic nitroxoline as a potent, selective and reversible inhibitor of CatB (Kos et al., 2014). Pharmacokinetic and toxicity data are available for nitroxoline (Mrhar et al., 1979; Naber et al., 2014; Wagenlehner et al., 2014; Wijma et al., 2018), which is administered orally for the treatment of urinary infection. Nitroxoline and its derivative compound 17, with improved potency and selectivity for CatB (Mitrović et al., 2017; Sosič et al., 2013), reduced SARS-CoV-2 D614G infection by more than 90% in Huh7-hACE2 (Table 1, Fig. 1). Another nitroxoline derivative, compound 3, which has a larger structural element that decreases its selectivity and potency for inhibiting CatB activity and increases inhibitory activity against CatL (Mirković et al., 2011; Mitrović et al., 2016), also showed high antiviral activity with EC50 values of less than 10 μM (Table 1, Fig. 1). In contrast, treatment with inhibitors selective for CatL, compounds CLIK-148 (Katunuma et al., 1999) and GCV-5 (Mitrović et al., 2022), had weaker antiviral activity (CLIK-148) or failed (GCV-5) to reduce SARS-CoV-2 infection (Table 1). These results highlight the role of CatB in the endosomal-dependent pathway of SARS-CoV-2 infection in Huh7-hACE2 cells. Targeted cathepsins are mainly involved in endosomal-dependent cell entry and less in further steps of viral replication and propagation (Fig. 4) (Jackson et al., 2022; Padmanabhan et al., 2020; Pišlar et al., 2020; Schornberg et al., 2006). This was also confirmed by the binding assay, where none of the inhibitors affected viral adsorption, while all of them significantly inhibited endocytosis step during viral entry (Fig. 5). Only compound 17 retained some antiviral activity also at later stages of infection indicating that other functions cannot be completely excluded (Fig. 4). The role of either CatB or CatL in S protein processing during SARS-CoV-2 entry is being actively investigated. Studies consistent with our findings confirmed the importance of CatB during SARS-CoV-2 infection. The work of Hashimoto suggests that CatB plays an important role in endocytosis-dependent infection reinforced by the antiviral activity of CatB inhibitor CA-074Me (Hashimoto et al., 2021). Another study performed on lung tissue from cancer patients infected with SARS-CoV-2 showed that CatB expression increased after virus infection and is associated with a hyperinflammatory response and poor prognosis (Ding et al., 2022).Cat B inhibitors could therefore have a dual effect, impairing viral uptake and excessive inflammation, as suggested for Cat L inhibitors (Mellott et al., 2021; Wu et al., 2020; Zhao et al., 2021). In contrast, other studies show that CatL plays an essential role in SARS-CoV-2 endosome-dependent entry into Hek293T-hACE2 cells (Ou et al., 2020; Zhou et al., 2015, 2016). For example, treatment with a CatL inhibitor reduced entry of SARS-CoV-2 by more than 76%, in these cells, while no effect on virus entry was observed after treatment with a CatB inhibitor (Ou et al., 2020). CatL, rather than CatB, was reported to cleave SARS-CoV-2 Spike, at pH 5.5 (Mellott et al., 2021). However, at this low pH the endopeptidase activity of CatB is sub-optimal and CatB-mediated Spike cleavage has been previously reported (Bollavaram et al., 2021; Jaimes et al., 2020). In addition, most of the reports showing little effect for CatB were performed in HEK-293-hACE2 or Vero E6 cell lines (Ashhurst et al., 2022; Mediouni et al., 2022; Ou et al., 2020, 2021). We therefore compared the different cell models showing that the antiviral activity of the cathepsin inhibitors correlated with the protein levels and activity of both cathepsins in each cell line used in the assay (Fig. 3). For nitroxoline and its two derivatives, compound 17 and compound 3, the strongest antiviral effect was observed in the Huh7-hACE2 cell line containing a high content of active CatB (Fig. 3). Addition of a potent irreversible CatB specific inhibitor CA-074 to discriminate between CatB and CatL activity showed that CatB dominated over CatL in this cell line (Fig. 3). In the other two cell lines used, Calu-3 and Vero E6, nitroxoline and compound 17 were found to be approximately 10-fold and 5-fold less potent, respectively, compared with Huh7-hACE2 cells (Fig. 2). The differences in inhibitor antiviral efficacy could be explained by the differences in protein levels and activity of CatB and CatL. In Calu-3 cells, protein levels and activity of CatB were higher than in Huh7-hACE2 cells. Because EC50 values are kinetic parameters, dependent on the experimental settings, including the concentration of enzyme and substrate in the assay (Copeland, 2005), a higher concentration of CatB inhibitor is required in Calu-3 to achieve the same inhibitory effect as in Huh7-hACE2 cells. Moreover, Calu-3 cells predominantly support the membrane fusion entry pathway due to their high endogenous expression of TMPRSS2 (Hoffmann et al., 2020; Ou et al., 2021; Saccon et al., 2021), which can additionally reduce the overall effect of CatB inhibitors. In contrast, in Vero E6 cells, the protein and activity levels of CatL are much higher than CatB indicating that CatL plays the major role during SARS-CoV-2 Spike cleavage and consequently cell entry and this could be a reason for the lower effect of CatB inhibitors in this host cell type compared to others. Moreover, the addition of the CatB specific inhibitor CA-074 in this cell line had no additional effect on CatL substrate degradation. Together, these data explain the lower effect of the CatB inhibitors nitroxoline and compound 17 on inhibition of SARS-CoV-2 infection in Vero E6 cells. These results may also explain the lack of effect of CatB inhibitors reported in some other studies that used this cell model (Ashhurst et al., 2022; Mediouni et al., 2022; Ou et al., 2021). The inhibition of SARS-CoV-2 infection by compound 3 did not correlate so well with the protein levels and activity of both CatB and CatL compared to nitroxoline and compound 17. In this regard it should be noted that compound 3 is an equal inhibitor of both, CatB and CatL, but less potent compared to nitroxoline and compound 17. To confirm our hypothesis about the correlation of cathepsin expression and compound efficacy in inhibiting SARS-CoV-2 infection, we next tested the effect of the specific CatL inhibitor GCV-5. In line with other studies, this compound is active in Vero E6 cells infected with SARS-CoV-2 (Ashhurst et al., 2022; Mellott et al., 2021). The Omicron variant of SARS-CoV-2 was sensitive to CatB inhibitors (Fig. 6). Omicron enters the cells mainly via the endosomal pathway, and is more sensitive to host cathepsins compared to previous variants (Benlarbi et al., 2022; Du et al., 2022; Hu et al., 2022; Hui et al., 2022; Meng et al., 2022; Willett et al., 2022). Indeed, E−64d was more effective also in our case (Table 3). Mutations within the Omicron S protein that alter conformation and interactions with the receptor, could affect utilization of different cathepsins for its entry thus modifying sensitivity to cathepsin inhibitors (Du et al., 2022; Hu et al., 2022; Hui et al., 2022; Imai et al., 2023; Willett et al., 2022). Selection of the proteolytic profile needed for virus entry into host cells is another level of adaptation of SARS-CoV-2 virus to its frequent genetic changes. Furthermore, differential cathepsin expression in tissues may dictate viral tropism and pathological consequences, as well as opportunities for specific inhibitors to impact on the consequences of infection on other tissues. For example, expression analysis shows that nasal epithelial cells are enriched for ACE2 not matched by TMPRSS2 levels, which seem to be compensated by CatL/CatB (Sungnak et al., 2020). In the human vasculature the expression pattern of CatB/CatL varies with CatB being mainly expressed in the brain vasculature and CatL predominantly in the peripheral vasculature, notably in the absence of TMPRSS2 expression in both cases (Ghobrial et al. BiorXiv doi.org/10.1101/2020.10.10.334664). Intriguingly, CatB is also expressed in microglia thus providing an interesting CatB-dependent pathway of entry that could account for long-term neuropathological consequences of COVID-19. 5 Conclusion The results presented here highlight the important role of host cysteine CatB in SARS-CoV-2 virus entry and demonstrate that, in addition to CatL inhibitors, specific CatB inhibitors, such as nitroxoline and its derivatives, significantly impair SARS-CoV-2 infection. As they solely inhibit CatB endopeptidase activity, involved in virus entry and inflammation, they would less likely affect physiological function of CatB based on exopeptidase activity. Furthermore, our data suggest that the antiviral activity of the inhibitors correlates with the amount of targeted cathepsin in host cells and depends on cell type and viral adaptions. Overall, the use of cathepsin-specific inhibitors could serve as a reliable therapeutic strategy against SARS-CoV-2 infection. Other polypharmacological effects of Cat inhibitors on SARS-CoV-2 induced inflammation could be additional advantages of the approach that deserve to be explored in more detail. To note, nitroxoline would act as repurposed drug in this context being clinically used as an antibiotic. Further studies would be required to explore the clinical use of nitroxoline in the context of SARS-CoV-2 infection, including dosage in vivo, route of administration and side effects. Funding This work was supported by the 10.13039/501100004329 Slovenian Research Agency , grant numbers P4-0127 to J.K., J3-3071 to AnM, and P1-0208 to SG. Work in AM laboratory related to SARS-CoV-2 is supported by grants from SNAM Foundation, Generali SpA, Beneficentia Stiftung, CARIPLO (INNATE-CoV) and the #FarmaCovid crowdfunding initiative. The funders had no role in the study design, data collection and analysis, or preparation of the manuscript. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: The authors declare the international application number PCT/EP2021/083960 “8-hydroxyquinoline cysteine protease inhibitors for use in the prevention and/or treatment of a coronavirus disease”. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Supplementary data The following are the Supplementary data to this article.Multimedia component 1 Multimedia component 1 Multimedia component 2 Multimedia component 2 Data availability Data will be made available on request. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.antiviral.2023.105655. ==== Refs References Ashhurst A.S. Tang A.H. Fajtová P. Yoon M.C. Aggarwal A. Bedding M.J. Stoye A. Beretta L. Pwee D. Drelich A. Skinner D. Li L. Meek T.D. McKerrow J.H. Hook V. Tseng C.-T. Larance M. Turville S. Gerwick W.H. O'Donoghue A.J. Payne R.J. Potent anti-SARS-CoV-2 activity by the natural product gallinamide A and analogues via inhibition of cathepsin L J. Med. Chem. 65 2022 2956 2970 34730959 Benlarbi M. Laroche G. Fink C. Fu K. Mulloy R.P. Phan A. Ariana A. Stewart C.M. Prevost J. Beaudoin-Bussieres G. Daniel R. Bo Y. El Ferri O. Yockell-Lelievre J. Stanford W.L. Giguere P.M. Mubareka S. Finzi A. Dekaban G.A. Dikeakos J.D. Cote M. Identification and differential usage of a host metalloproteinase entry pathway by SARS-CoV-2 Delta and Omicron iScience 25 2022 105316 Bollavaram K. Leeman T.H. Lee M.W. Kulkarni A. Upshaw S.G. Yang J. Song H. Platt M.O. Multiple sites on SARS-CoV-2 spike protein are susceptible to proteolysis by cathepsins B, K, L, S, and V Protein Sci. 30 2021 1131 1143 33786919 Cagno V. SARS-CoV-2 cellular tropism The Lancet Microbe 1 2020 e2 e3 32835319 Chandran K. Sullivan N.J. Felbor U. Whelan S.P. Cunningham J.M. Endosomal proteolysis of the Ebola virus glycoprotein is necessary for infection Science (New York, N.Y.) 308 2005 1643 1645 15831716 Chitsike L. Duerksen-Hughes P. Keep out! SARS-CoV-2 entry inhibitors: their role and utility as COVID-19 therapeutics Virol. J. 18 2021 154-154 Copeland R.A. Evaluation of Enzyme Inhibitors in Drug Discovery: a Guide for Medicinal Chemists and Pharmacologists 2005 Wiley-Interscience New York CoVariants https://covariants.org 2022 December 2022 Dana D. Pathak S.K. A Review of Small Molecule Inhibitors and Functional Probes of Human Cathepsin L 2020 Diederich S. Thiel L. Maisner A. Role of endocytosis and cathepsin-mediated activation in Nipah virus entry Virology 375 2008 391 400 18342904 Ding X. Ye N. Qiu M. Guo H. Li J. Zhou X. Yang M. Xi J. Liang Y. Gong Y. Li J. Cathepsin B is a potential therapeutic target for coronavirus disease 2019 patients with lung adenocarcinoma Chem. Biol. Interact. 353 2022 109796-109796 Dolgin E. The race for antiviral drugs to beat COVID — and the next pandemic Nature 592 2021 340 343 33854246 Du X. Tang H. Gao L. Wu Z. Meng F. Yan R. Qiao S. An J. Wang C. Qin F.X.-F. Omicron adopts a different strategy from Delta and other variants to adapt to host Signal Transduct. Targeted Ther. 7 2022 45-45 Evans J.P. Liu S.-L. Role of host factors in SARS-CoV-2 entry J. Biol. Chem. 297 2021 100847-100847 Fang F.F. As the virus evolves, so too must we: a drug developer's perspective Virol. J. 19 2022 159-159 Gomes C.P. Fernandes D.E. Casimiro F. da Mata G.F. Passos M.T. Varela P. Mastroianni-Kirsztajn G. Pesquero J.B. Cathepsin L in COVID-19: from pharmacological evidences to genetics Front. Cell. Infect. Microbiol. 10 2020 Gottlieb R.L. Vaca C.E. Paredes R. Mera J. Webb B.J. Perez G. Oguchi G. Ryan P. Nielsen B.U. Brown M. Hidalgo A. Sachdeva Y. Mittal S. Osiyemi O. Skarbinski J. Juneja K. Hyland R.H. Osinusi A. Chen S. Camus G. Abdelghany M. Davies S. Behenna-Renton N. Duff F. Marty F.M. Katz M.J. Ginde A.A. Brown S.M. Schiffer J.T. Hill J.A. Early remdesivir to prevent progression to severe covid-19 in outpatients N. Engl. J. Med. 386 2022 305 315 34937145 Hoffmann M. Kleine-Weber H. Schroeder S. Krüger N. Herrler T. Erichsen S. Schiergens T.S. Herrler G. Wu N.-H.H. Nitsche A. Müller M.A. Drosten C. Pöhlmann S. SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 and is blocked by a clinically proven protease inhibitor Cell 181 2020 271 280.e278 32142651 Hu B. Chan J.F.-W. Liu H. Liu Y. Chai Y. Shi J. Shuai H. Hou Y. Huang X. Yuen T.T.-T. Yoon C. Zhu T. Zhang J. Li W. Zhang A.J. Zhou J. Yuan S. Zhang B.-Z. Yuen K.-Y. Chu H. Spike mutations contributing to the altered entry preference of SARS-CoV-2 omicron BA.1 and BA.2 Emerg. Microb. Infect. 11 2022 2275 2287 Hui K.P.Y. Ho J.C.W. Cheung M.-c. Ng K.-c. Ching R.H.H. Lai K.-l. Kam T.T. Gu H. Sit K.-Y. Hsin M.K.Y. Au T.W.K. Poon L.L.M. Peiris M. Nicholls J.M. Chan M.C.W. SARS-CoV-2 Omicron variant replication in human bronchus and lung ex vivo Nature 603 2022 715 720 35104836 Imai M. Ito M. Kiso M. Yamayoshi S. Uraki R. Fukushi S. Watanabe S. Suzuki T. Maeda K. Sakai-Tagawa Y. Iwatsuki-Horimoto K. Halfmann P.J. Kawaoka Y. Efficacy of antiviral agents against omicron subvariants BQ.1.1 and XBB N. Engl. J. Med. 388 2023 89 91 36476720 Jackson C.B. Farzan M. Chen B. Choe H. Mechanisms of SARS-CoV-2 entry into cells Nat. Rev. Mol. Cell Biol. 23 2022 3 20 34611326 Jaimes J.A. Millet J.K. Whittaker G.R. Proteolytic cleavage of the SARS-CoV-2 spike protein and the role of the novel S1/S2 site iScience 23 2020 101212 Jayk Bernal A. Gomes da Silva M.M. Musungaie D.B. Kovalchuk E. Gonzalez A. Delos Reyes V. Martín-Quirós A. Caraco Y. Williams-Diaz A. Brown M.L. Du J. Pedley A. Assaid C. Strizki J. Grobler J.A. Shamsuddin H.H. Tipping R. Wan H. Paschke A. Butterton J.R. Johnson M.G. De Anda C. Molnupiravir for oral treatment of covid-19 in nonhospitalized patients N. Engl. J. Med. 386 2022 509 520 34914868 Katunuma N. Murata E. Kakegawa H. Matsui A. Tsuzuki H. Tsuge H. Turk D. Turk V. Fukushima M. Tada Y. Asao T. Structure based development of novel specific inhibitors for cathepsin L and cathepsin S in vitro and in vivo FEBS (Fed. Eur. Biochem. Soc.) Lett. 458 1999 6 10 Kawase M. Shirato K. Matsuyama S. Taguchi F. Protease-mediated entry via the endosome of human coronavirus 229E J. Virol. 83 2009 712 721 18971274 Kokic G. Hillen H.S. Tegunov D. Dienemann C. Seitz F. Schmitzova J. Farnung L. Siewert A. Höbartner C. Cramer P. Mechanism of SARS-CoV-2 polymerase stalling by remdesivir Nat. Commun. 12 2021 279-279 Kos J. Mitrović A. Mirković B. The current stage of cathepsin B inhibitors as potential anticancer agents Future Med. Chem. 6 2014 1355 1371 25163003 Kos J. Sekirnik A. Premzl A. Bergant V.Z. Langerholc T. Turk B. Werle B. Golouh R. Repnik U. Jeras M. Turk V. Carboxypeptidases cathepsins X and B display distinct protein profile in human cells and tissues Exp. Cell Res. 306 2005 103 113 15878337 Licastro D. Rajasekharan S. Dal Monego S. Segat L. D'Agaro P. Marcello A. Isolation and full-length genome characterization of SARS-CoV-2 from COVID-19 cases in northern Italy J. Virol. 94 2020 Liu T. Luo S. Libby P. Shi G.-P. Cathepsin L-selective inhibitors: a potentially promising treatment for COVID-19 patients Pharmacol. Therapeut. 213 2020 107587-107587 Marzi M. Vakil M.K. Bahmanyar M. Zarenezhad E. Paxlovid: mechanism of action, synthesis, and in silico study BioMed Res. Int. 2022 2022 1 16 Mediouni S. Mou H. Otsuka Y. Jablonski J.A. Adcock R.S. Batra L. Chung D.-H. Rood C. de Vera I.M.S. Rahaim R. Jr. Ullah S. Yu X. Getmanenko Y.A. Kennedy N.M. Wang C. Nguyen T.-T. Hull M. Chen E. Bannister T.D. Baillargeon P. Scampavia L. Farzan M. Valente S.T. Spicer T.P. Identification of potent small molecule inhibitors of SARS-CoV-2 entry SLAS Discov. 27 2022 8 19 35058179 Mellott D.M. Tseng C.T. Drelich A. Fajtova P. Chenna B.C. Kostomiris D.H. Hsu J. Zhu J. Taylor Z.W. Kocurek K.I. Tat V. Katzfuss A. Li L. Giardini M.A. Skinner D. Hirata K. Yoon M.C. Beck S. Carlin A.F. Clark A.E. Beretta L. Maneval D. Hook V. Frueh F. Hurst B.L. Wang H. Raushel F.M. O'Donoghue A.J. de Siqueira-Neto J.L. Meek T.D. McKerrow J.H. A clinical-stage cysteine protease inhibitor blocks SARS-CoV-2 infection of human and monkey cells ACS Chem. Biol. 16 2021 642 650 33787221 Meng B. Abdullahi A. Ferreira I.A.T.M. Goonawardane N. Saito A. Kimura I. Yamasoba D. Gerber P.P. Fatihi S. Rathore S. Zepeda S.K. Papa G. Kemp S.A. Ikeda T. Toyoda M. Tan T.S. Kuramochi J. Mitsunaga S. Ueno T. Shirakawa K. Takaori-Kondo A. Brevini T. Mallery D.L. Charles O.J. Baker S. Dougan G. Hess C. Kingston N. Lehner P.J. Lyons P.A. Matheson N.J. Ouwehand W.H. Saunders C. Summers C. Thaventhiran J.E.D. Toshner M. Weekes M.P. Maxwell P. Shaw A. Bucke A. Calder J. Canna L. Domingo J. Elmer A. Fuller S. Harris J. Hewitt S. Kennet J. Jose S. Kourampa J. Meadows A. O'Brien C. Price J. Publico C. Rastall R. Ribeiro C. Rowlands J. Ruffolo V. Tordesillas H. Bullman B. Dunmore B.J. Gräf S. Hodgson J. Huang C. Hunter K. Jones E. Legchenko E. Matara C. Martin J. Mescia F. O'Donnell C. Pointon L. Shih J. Sutcliffe R. Tilly T. Treacy C. Tong Z. Wood J. Wylot M. Betancourt A. Bower G. Cossetti C. De Sa A. Epping M. Fawke S. Gleadall N. Grenfell R. Hinch A. Jackson S. Jarvis I. Krishna B. Nice F. Omarjee O. Perera M. Potts M. Richoz N. Romashova V. Stefanucci L. Strezlecki M. Turner L. De Bie E.M.D.D. Bunclark K. Josipovic M. Mackay M. Butcher H. Caputo D. Chandler M. Chinnery P. Clapham-Riley D. Dewhurst E. Fernandez C. Furlong A. Graves B. Gray J. Hein S. Ivers T. Le Gresley E. Linger R. Kasanicki M. King R. Kingston N. Meloy S. Moulton A. Muldoon F. Ovington N. Papadia S. Penkett C.J. Phelan I. Ranganath V. Paraschiv R. Sage A. Sambrook J. Scholtes I. Schon K. Stark H. Stirrups K.E. Townsend P. Walker N. Webster J. Butlertanaka E.P. Tanaka Y.L. Ito J. Uriu K. Kosugi Y. Suganami M. Oide A. Yokoyama M. Chiba M. Motozono C. Nasser H. Shimizu R. Kitazato K. Hasebe H. Irie T. Nakagawa S. Wu J. Takahashi M. Fukuhara T. Shimizu K. Tsushima K. Kubo H. Kazuma Y. Nomura R. Horisawa Y. Nagata K. Kawai Y. Yanagida Y. Tashiro Y. Tokunaga K. Ozono S. Kawabata R. Morizako N. Sadamasu K. Asakura H. Nagashima M. Yoshimura K. Cárdenas P. Muñoz E. Barragan V. Márquez S. Prado-Vivar B. Becerra-Wong M. Caravajal M. Trueba G. Rojas-Silva P. Grunauer M. Gutierrez B. Guadalupe J.J. Fernández-Cadena J.C. Andrade-Molina D. Baldeon M. Pinos A. Bowen J.E. Joshi A. Walls A.C. Jackson L. Martin D. Smith K.G.C. Bradley J. Briggs J.A.G. Choi J. Madissoon E. Meyer K.B. Mlcochova P. Ceron-Gutierrez L. Doffinger R. Teichmann S.A. Fisher A.J. Pizzuto M.S. de Marco A. Corti D. Hosmillo M. Lee J.H. James L.C. Thukral L. Veesler D. Sigal A. Sampaziotis F. Goodfellow I.G. Matheson N.J. Sato K. Gupta R.K. Altered TMPRSS2 usage by SARS-CoV-2 Omicron impacts infectivity and fusogenicity Nature 603 2022 706 714 35104837 Milan Bonotto R. Boni F. Milani M. Chaves-Sanjuan A. Franze S. Selmin F. Felicetti T. Bolognesi M. Konstantinidou S. Poggianella M. Marquez C.L. Dattola F. Zoppe M. Manfroni G. Mastrangelo E. Marcello A. Virucidal activity of the pyridobenzothiazolone derivative HeE1-17Y against enveloped RNA viruses Viruses 14 2022 Milani M. Donalisio M. Bonotto R.M. Schneider E. Arduino I. Boni F. Lembo D. Marcello A. Mastrangelo E. Combined in silico and in vitro approaches identified the antipsychotic drug lurasidone and the antiviral drug elbasvir as SARS-CoV2 and HCoV-OC43 inhibitors Antivir. Res. 189 2021 105055 Mirković B. Markelc B. Butinar M. Mitrović A. Sosič I. Gobec S. Vasiljeva O. Turk B. Čemažar M. Serša G. Kos J. Nitroxoline impairs tumor progression in vitro and in vivo by regulating cathepsin B activity Oncotarget 6 2015 19027 19042 25848918 Mirković B. Renko M. Turk S. Sosič I. Jevnikar Z. Obermajer N. Turk D. Gobec S. Kos J. Novel mechanism of cathepsin B inhibition by antibiotic nitroxoline and related compounds ChemMedChem 6 2011 1351 1356 21598397 Mitrović A. Mirković B. Sosič I. Gobec S. Kos J. Inhibition of endopeptidase and exopeptidase activity of cathepsin B impairs extracellular matrix degradation and tumour invasion Biol. Chem. 397 2016 165 174 26565553 Mitrović A. Senjor E. Jukić M. Bolčina L. Prunk M. Proj M. Nanut M.P. Gobec S. Kos J. New inhibitors of cathepsin V impair tumor cell proliferation and elastin degradation and increase immune cell cytotoxicity Comput. Struct. Biotechnol. J. 20 2022 4667 4687 36147668 Mitrović A. Sosič I. Kos Š. Tratar U.L. Breznik B. Kranjc S. Mirković B. Gobec S. Lah T. Serša G. Kos J. Addition of 2-(ethylamino)acetonitrile group to nitroxoline results in significantly improved anti-tumor activityin vitroandin vivo Oncotarget 8 2017 59136 59147 28938624 Mrhar A. Kopitar Z. Kozjek F. Presl V. Karba R. Clinical pharmacokinetics of nitroxoline Int. J. Clin. Pharmacol. Biopharm. 17 1979 476 481 118941 Naber K.G. Niggemann H. Stein G. Stein G. Review of the literature and individual patients' data meta-analysis on efficacy and tolerance of nitroxoline in the treatment of uncomplicated urinary tract infections BMC Infect. Dis. 14 2014 628 25427651 Ou T. Mou H. Zhang L. Ojha A. Choe H. Farzan M. Hydroxychloroquine-mediated inhibition of SARS-CoV-2 entry is attenuated by TMPRSS2 PLoS Pathog. 17 2021 Ou X. Liu Y. Lei X. Li P. Mi D. Ren L. Guo L. Guo R. Chen T. Hu J. Xiang Z. Mu Z. Chen X. Chen J. Hu K. Jin Q. Wang J. Qian Z. Characterization of spike glycoprotein of SARS-CoV-2 on virus entry and its immune cross-reactivity with SARS-CoV Nat. Commun. 11 2020 Padmanabhan P. Desikan R. Dixit N.M. Targeting TMPRSS2 and Cathepsin B/L together may be synergistic against SARS-CoV-2 infection PLoS Comput. Biol. 16 2020 Pišlar A. Mitrovic A. Sabotič J. Fonovic U.P. Nanut M.P. Jakoš T. Senjor E. Kos J. The role of cysteine peptidases in coronavirus cell entry and replication: the therapeutic potential of cathepsin inhibitors PLoS Pathog. 16 2020 1 23 Rajasekharan S. Milan Bonotto R. Nascimento Alves L. Kazungu Y. Poggianella M. Martinez-Orellana P. Skoko N. Polez S. Marcello A. Inhibitors of protein glycosylation are active against the coronavirus severe acute respiratory syndrome coronavirus SARS-CoV-2 Viruses 13 2021 Regan A.D. Shraybman R. Cohen R.D. Whittaker G.R. Differential role for low pH and cathepsin-mediated cleavage of the viral spike protein during entry of serotype II feline coronaviruses Vet. Microbiol. 132 2008 235 248 18606506 Saccon E. Chen X. Mikaeloff F. Rodriguez J.E. Szekely L. Vinhas B.S. Krishnan S. Byrareddy S.N. Frisan T. Végvári Á. Mirazimi A. Neogi U. Gupta S. Cell-type-resolved quantitative proteomics map of interferon response against SARS-CoV-2 iScience 24 2021 102420-102420 Schornberg K. Matsuyama S. Kabsch K. Delos S. Bouton A. White J. Role of endosomal cathepsins in entry mediated by the ebola virus glycoprotein J. Virol. 80 2006 4174 4178 16571833 Shang J. Wan Y. Luo C. Ye G. Geng Q. Auerbach A. Li F. Cell entry mechanisms of SARS-CoV-2 Proc. Natl. Acad. Sci. USA 117 2020 11727 11734 32376634 Shirato K. Kawase M. Matsuyama S. Middle East respiratory syndrome coronavirus infection mediated by the transmembrane serine protease TMPRSS2 J. Virol. 87 2013 12552 12561 24027332 Simmons G. Gosalia D.N. Rennekamp A.J. Reeves J.D. Diamond S.L. Bates P. Inhibitors of cathepsin L prevent severe acute respiratory syndrome coronavirus entry Proc. Natl. Acad. Sci. U.S.A. 102 2005 11876 11881 16081529 Simmons G. Zmora P. Gierer S. Heurich A. Pöhlmann S. Proteolytic activation of the SARS-coronavirus spike protein: cutting enzymes at the cutting edge of antiviral research Antivir. Res. 100 2013 605 614 24121034 Sosič I. Mirković B. Arenz K. Štefane B. Kos J. Gobec S. Development of new cathepsin B inhibitors: combining bioisosteric replacements and structure-based design to explore the structure-activity relationships of nitroxoline derivatives J. Med. Chem. 56 2013 521 533 23252745 Sungnak W. Huang N. Becavin C. Berg M. Queen R. Litvinukova M. Talavera-Lopez C. Maatz H. Reichart D. Sampaziotis F. Worlock K.B. Yoshida M. Barnes J.L. Network H.C.A.L.B. SARS-CoV-2 entry factors are highly expressed in nasal epithelial cells together with innate immune genes Nat. Med. 26 2020 681 687 32327758 Turk B. Turk D. Turk V. Lysosomal Cysteine Proteases: More than Scavengers 2000 98 111 Wagenlehner F.M. Munch F. Pilatz A. Barmann B. Weidner W. Wagenlehner C.M. Straubinger M. Blenk H. Pfister W. Kresken M. Naber K.G. Urinary concentrations and antibacterial activities of nitroxoline at 250 milligrams versus trimethoprim at 200 milligrams against uropathogens in healthy volunteers Antimicrob. Agents Chemother. 58 2014 713 721 24217699 WHO WHO Coronavirus (COVID-19) Dashboard (accessed December 2021) https://covid19.who.int 2021 Wijma R.A. Huttner A. Koch B.C.P. Mouton J.W. Muller A.E. Review of the pharmacokinetic properties of nitrofurantoin and nitroxoline J. Antimicrob. Chemother. 73 2018 2916 2926 30184207 Willett B.J. Grove J. MacLean O.A. Wilkie C. De Lorenzo G. Furnon W. Cantoni D. Scott S. Logan N. Ashraf S. Manali M. Szemiel A. Cowton V. Vink E. Harvey W.T. Davis C. Asamaphan P. Smollett K.L. Tong L. Orton R.J. Hughes J. Holland P. Silva V. Pascall D.J. Puxty K. da Silva Filipe A. Yebra G. Shaaban S. Holden M.T.G. Pinto R.M. Gunson R.N. Templeton K.E. Murcia P.R. Patel A.A.H. Klenerman P. Dunachie S. Dunachie S. Klenerman P. Barnes E. Brown A. Adele S. Kronsteiner B. Murray S.M. Abraham P. Deeks A. Ansari M.A. de Silva T.I. Turtle L. Moore S. Austin J. Richter A. Duncan C. Payne R. Ash A. Koshy C. Kele B. Cutino-Moguel T. Fairley D.J. McKenna J.P. Curran T. Adams H. Fraser C. Bonsall D. Fryer H. Lythgoe K. Thomson L. Golubchik T. Murray A. Singleton D. Beckwith S.M. Mantzouratou A. Barrow M. Buchan S.L. Reynolds N. Warne B. Maksimovic J. Spellman K. McCluggage K. John M. Beer R. Afifi S. Morgan S. Mack A. Marchbank A. Price A. Morriss A. Bresner C. Kitchen C. Merrick I. Southgate J. Guest M. Jones O. Munn R. Connor T.R. Whalley T. Workman T. Fuller W. Patel A.A.H. Patel B. Nebbia G. Edgeworth J. Snell L.B. Batra R. Charalampous T. Beckett A.H. Shelest E. Robson S.C. Underwood A.P. Taylor B.E.W. Yeats C.A. Aanensen D.M. Abudahab K. Menegazzo M. Joseph A. Clark G. Howson-Wells H.C. Berry L. Khakh M. Lister M.M. Boswell T. Fleming V.M. Holmes C.W. McMurray C.L. Shaw J. Tang J.W. Fallon K. Odedra M. Willford N.J. Bird P.W. Helmer T. Williams L.-A. Sheriff N. Campbell S. Raviprakash V. Blakey V. Moore C.C. Sang F. Debebe J. Carlile M. Loose M.W. Holmes N. Wright V. Torok M.E. Hamilton W.L. Carabelli A.M. Jermy A. Blane B. Churcher C.M. Ludden C. Aggarwal D. Westwick E. Brooks E. McManus G.M. Galai K. Smith K.S.K. Smith K.S.K. Cox M. Fragakis M. Maxwell P. Judges S. Peacock S.J. Feltwell T. Kenyon A. Eldirdiri S. Davis T. Taylor J.F. Tan N.K. Zarebski A.E. Gutierrez B. Raghwani J. du Plessis L. Kraemer M.U.G. Pybus O.G. Francois S. Attwood S.W. Vasylyeva T.I. Jahun A.S. Goodfellow I.G. Georgana I. Pinckert M.L. Hosmillo M. Izuagbe R. Chaudhry Y. Ryan F. Lowe H.H.L. Moses S. Bedford L. Cargill J.S. Hughes W. Moore J. Stonehouse S. Shah D. Lee J.C.D. Brown J.R. Harris K.A. Atkinson L. Storey N. Spyer M.J. Flaviani F. Alcolea-Medina A. Sehmi J. Ramble J. Ohemeng-Kumi N. Smith P. Bertolusso B. Thomas C. Vernet G. Lynch J. Moore N. Cortes N. Williams R.R.J. Kidd S.P. Levett L.J. Pusok M. Grant P.R. Kirk S. Chatterton W. Xu-McCrae L. Smith D.L. Young G.R. Bashton M. Kitchman K. Gajee K. Eastick K. Lillie P.J. Burns P.J. Everson W. Cox A. Holmes A.H. Bolt F. Price J.R. Pond M. Randell P.A. Madona P. Mookerjee S. Volz E.M. Geidelberg L. Ragonnet-Cronin M. Boyd O. Johnson R. Pope C.F. Witney A.A. Monahan I.M. Laing K.G. Smollett K.L. McNally A. McMurray C.L. Stockton J. Quick J. Loman N.J. Poplawski R. Nicholls S. Rowe W. Castigador A. Macnaughton E. Bouzidi K.E. Sudhanva M. Lampejo T. Martinez Nunez R.T. Breen C. Sluga G. Withell K.T. Machin N.W. George R.P. Ahmad S.S.Y. Pritchard D.T. Binns D. Wong N. James V. Williams C.C.C.A.C. Illingworth C.J. Jackson C. de Angelis D. Pascall D.J. Mukaddas A. Broos A. da Silva Filipe A. Mair D. Robertson D.L. Wright D.W. Thomson E.C. Starinskij I. Tsatsani I. Shepherd J.G. Nichols J. Hughes J. Nomikou K. Tong L. Orton R.J. Vattipally S. Harvey W.T. Sanderson R. O'Brien S. Rushton S. Perkins J. Blacow R. Gunson R.N. Gallagher A. Wastnedge E. Templeton K.E. McHugh M.P. Dewar R. Cotton S. Coupland L. Stanley R. Dervisevic S. Spurgin L.G. Smith L. Graham C. Padgett D. Barton E. Scott G. Cross A. Mirfenderesky M. Swindells E. Greenaway J. Denton-Smith R. Turnbull R. Idle G. Cole K. Hollis A. Nelson A. McCann C.M. Henderson J.H. Crown M.R. Yew W.C. Stanley W. Duckworth N. Clarke P. Walsh S. Sloan T.J. Bicknell K. Impey R. Wyllie S. Elliott S. Glaysher S. Bradley D.T. Killough N.F. Wyatt T. Bosworth A. Vipond B.B. Pearson C. Allara E. Robinson E. Pymont H.M. Osman H. Muir P. Hopes R. Hutchings S. Curran M.D. Parmar S. Thornton A. Lackenby A. Bishop C. Bibby D. Lee D. Gallagher E. Dabrera G. Harrison I. Hubb J. Twohig K.A. Chand M. Ellaby N. Manesis N. Myers R. Platt S. Mbisa T. Chalker V. Yebra G. Holden M.T.G. Shaaban S. Rooke S. Birchley A. Adams A. Davies A. Gaskin A. Gatica-Wilcox B. McKerr C. Moore C.C. Williams C.C.C.A.C. Williams C.C.C.A.C. Heyburn D. De Lacy E. Hilvers E. Downing F. Pugh G. Jones H. Asad H. Coombes J. Hey J. Powell J. Watkins J. Evans J.M. Fina L. Gifford L. Gilbert L. Graham L. Perry M. Morgan M. Bull M. Pacchiarini N. Craine N. Corden S. Kumziene-Summerhayes S. Rey S. Taylor S. Cottrell S. Jones S. Edwards S. Annett T. Trotter A.J. Mather A.E. Aydin A. Page A.J. Baker D.J. Foster-Nyarko E. Kay G.L. O'Grady J. de Oliveira Martins L. Meadows L. Alikhan N.-F. Prosolek S.J. Rudder S. Le-Viet T. Casey A. Ratcliffe L. Singh A. Mariappan A. Baxter C. Radulescu C. Simpson D.A. Lavin D. Rogan F. Miskelly J. Fuchs M. Tang M. Carvalho S.F. Bridgett S. Skvortsov T. Molnar Z. Ramadan N.A. Knight B.A. Jones C.R. Auckland C. Morcrette H. Poyner J. Irish-Tavares D. Witele E. Hart J. Mahungu T.W. Haque T. Bourgeois Y. Fearn C. Cook K.F. Loveson K.F. Goudarzi S. Evans C. Partridge D.G. Johnson K. Yavus M. Raza M. Mower C. Baker P. Essex S. Bonner S. Murray L.J. Watson L.K. Liggett S. Lawton A.I. Lyons R.A. Payne B.A.I. Eltringham G. Collins J. Waugh S. Burton-Fanning S. Taha Y. Jeanes C. Gomes A.N. Murray D.R. Kimuli M. Dobie D. Ashfield P. Best A. Percival B. Moles-Garcia E. Ashford F. Mirza J. Crawford L. Mayhew M. Cumley N. Megram O. Frampton D. Heaney J. Byott M. Houlihan C. Williams C.C.C.A.C. Nastouli E. Lowe H.H.L. Hartley J.A. Breuer J. Maftei L. Ensell L. Cotic M. Mondani M. Driscoll M. Bayzid N. Williams R.R.J. Roy S. Mahanama A.I.K. Samaraweera B. Wilson-Davies E. Pelosi E. Umpleby H. Wheeler H. Prieto J.A. Saeed K. Harvey M. Jeremiah S. Silviera S. Aplin S. Sass T. Macklin B. Crudgington D. Sheridan L.A. Cogger B.J. Malone C.S. Munemo F. Huckson H. Lewis J. Easton L.J. Mutingwende M. Erkiert M.J. Hassan-Ibrahim M.O. Chaloner N.J. Podplomyk O. Randell P.A. Nicodemi R. Lowdon S. Somassa T. Richter A. Beggs A. Hesketh A.R. Smith C.P. Bucca G. Ruis C. Cormie C. Higginson E.E. Young J. Dias J. Kermack L.M. Maes M. Gupta R.K. Forrest S. Girgis S.T. Davidson R.K. O'Toole Á. Rambaut A. Jackson B. Balcazar C.E. Maloney D. Scher E. McCrone J.T. Williamson K.A. Gallagher M.D. Medd N. Colquhoun R. Stanton T.D. Williams T. Hill V. Jeffries A.R. Temperton B. Sambles C.M. Studholme D.J. Warwick-Dugdale J. Jackson L.M. Michelsen M.L. Manley R. Michell S.L. Darby A.C. Lucaci A.O. Nelson C. Wierzbicki C. Vamos E.E. Webster H.J. Jackson K.A. Rainbow L. Hughes M. Whitehead M. Gemmell M. Iturriza-Gomara M. Eccles R. Gregory R. Haldenby S.T. Paterson S. Angyal A. Keeley A.J. Foulkes B.H. Lindsey B.B. Wang D. Hornsby H.R. Green L.R. Pohare M. Gallis M. Parker M.D. Whiteley M. Smith N. Wolverson P. Zhang P. Hansford S.E. Hsu S.N. Louka S.F. de Silva T.I. Freeman T.M. Mori M. Park E.J. Hill J.D. Dey J. Ball J. Chappell J.G. McClure P.C. Byaruhanga T. Fanaie A. Yaze G. Hilson R.A. Trebes A. Green A. Buck D. MacIntyre-Cockett G. Todd J.A. Bassett A.R. Whitwham A. Langford C.F. Rajan D. Kwiatkowski D. Harrison E.M. Bronner I.F. Tovar-Corona J.M. Liddle J. Durham J. Bellis K.L. Lewis K. Aigrain L. Redshaw N.M. Davies R.M. Moll R.J. McCarthy S.A. Lensing S.V. Leonard S. Farr B.W. Scott C. Beaver C. Ariani C.V. Weldon D. Jackson D.K. Betteridge E. Tonkin-Hill G. Johnston I. Martincorena I. Bonfield J. Barrett J.C. Sillitoe J. Keatley J.-P. Oliver K. James K. Shirley L. Prestwood L. Foulser L. Gourtovaia M. Dorman M.J. Quail M.A. Spencer Chapman M.H. Park N.R. Livett R. Amato R. Kay S. Goodwin S. Thurston S.A.J. Rajatileka S. Gonçalves S. Lo S. Sanderson T. Maclean A. Goldstein E.J. Ferguson L. Tomb R. Catalan J. Jones N. Haughney J. Robertson D.L. Palmarini M. Ray S. Thomson E.C. SARS-CoV-2 Omicron is an immune escape variant with an altered cell entry pathway Nat. Microbiol. 7 2022 1161 1179 35798890 Wu M. Chen Y. Xia H. Wang C. Tan C.Y. Cai X. Liu Y. Ji F. Xiong P. Liu R. Guan Y. Duan Y. Kuang D. Xu S. Cai H. Xia Q. Yang D. Wang M.W. Chiu I.M. Cheng C. Ahern P.P. Liu L. Wang G. Surana N.K. Xia T. Kasper D.L. Transcriptional and proteomic insights into the host response in fatal COVID-19 cases Proc. Natl. Acad. Sci. U. S. A. 117 2020 28336 28343 33082228 Yadati T. Houben T. Bitorina A. Shiri-Sverdlov R. The ins and outs of cathepsins: physiological function and role in disease management Cells 9 2020 1679-1679 Zakaria M.K. Carletti T. Marcello A. Cellular targets for the treatment of flavivirus infections Front. Cell. Infect. Microbiol. 8 2018 398 30483483 Zhao H. Lu L. Peng Z. Chen L.-L. Meng X. Zhang C. Ip J.D. Chan W.-M. Chu A.W.-H. Chan K.-H. Jin D.-Y. Chen H. Yuen K.-Y. To K.K.-W. SARS-CoV-2 Omicron variant shows less efficient replication and fusion activity when compared with Delta variant in TMPRSS2-expressed cells Emerg. Microb. Infect. 11 2022 277 283 Zhao M.M. Yang W.L. Yang F.Y. Zhang L. Huang W.J. Hou W. Fan C.F. Jin R.H. Feng Y.M. Wang Y.C. Yang J.K. Cathepsin L plays a key role in SARS-CoV-2 infection in humans and humanized mice and is a promising target for new drug development Signal Transduct. Targeted Ther. 6 2021 134 Zhou N. Pan T. Zhang J. Li Q. Zhang X. Bai C. Huang F. Peng T. Zhang J. Liu C. Tao L. Zhang H. Glycopeptide antibiotics potently inhibit cathepsin l in the late endosome/lysosome and block the entry of ebola virus, middle east respiratory syndrome coronavirus (MERS-CoV), and severe acute respiratory syndrome coronavirus (SARS-CoV) J. Biol. Chem. 291 2016 9218 9232 26953343 Zhou Y. Vedantham P. Lu K. Agudelo J. Carrion R. Nunneley J.W. Barnard D. Pöhlmann S. McKerrow J.H. Renslo A.R. Simmons G. Protease inhibitors targeting coronavirus and filovirus entry Antivir. Res. 116 2015 76 84 25666761
PMC010xxxxxx/PMC10287185.txt	==== Front J King Saud Univ Comput Inf Sci J King Saud Univ Comput Inf Sci Journal of King Saud University. Computer and Information Sciences 1319-1578 2213-1248 The Author(s). Published by Elsevier B.V. on behalf of King Saud University. S1319-1578(23)00183-0 10.1016/j.jksuci.2023.101629 101629 Article A Novel Secure Authentication Protocol for e-Health Records in Cloud with a New Key Generation Method and Minimized Key Exchange Mohinder Singh B. Natarajan Jaisankar ⁎ School of Computer Science and Engineering, Vellore Institute of Technology, Vellore 632014, India ⁎ Corresponding author. 23 6 2023 23 6 2023 10162922 2 2023 25 5 2023 16 6 2023 © 2023 The Author(s) 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In wake of covid19, many countries are shifting their paper-based health record management from manual processes to digital ones. The major benefit of digital health record is that data can be easily shared. As health data is sensitive, more security is to be provided to gain the trust of stakeholders. In this paper, a novel secure authentication protocol is planned for digitalizing personal health record that will be used by the user. While transacting data, a key is used to secure it. Many protocols used elliptic curve cryptography. In this proposed protocol, at an initial stage, an asymmetric and quantum-resistant crypto-algorithm, Kyber is used. In further stages, symmetric crypto-algorithm, Advanced Encryption Standard in Galois/Counter mode (AES-GCM) is used to secure transferred data. For every session, a new key is generated for secure transactions. The more interesting fact in this protocol is that transactions are secured without exchanging actual key and also minimized the key exchange. This protocol not only verified the authenticity of user but also checked rightful citizenship of user. This protocol is analyzed for various security traits using ProVerif tool and provided better results relating to security provisioning, cost of storage, and computation as opposed to related protocols. Keywords Authentication protocol eHealth records Post-quantum crypto algorithm Symmetric crypto-algorithm Asymmetric crypto-algorithm minimized key exchange ==== Body pmc1 1. Introduction In this fast-developing digital world, most data is being digitized. Many transactions are carried out digitally. This is same in case of healthcare also. The health data of patients are being digitized to give better health benefits. The advent of these digital records made the patients hassle free from maintaining documents on paper. In one-way, it reduced the usage of paper which in-turn is environment friendly. On other hand, healthcare providers take quick decisions in emergencies by accessing particular patient’s medical record. Globally, many medical organizations have digitalized their paper-based patients’ data. These digital patients’ records are collectively called electronic health records (EHR). These EHRs are maintained using cloud. As these records are maintained in the cloud, patients’ data can be exchanged among many stakeholders such as patients, doctors, nurses, and so on very easily. In many fields like Healthcare (Jiang et al., 2017, Nikou et al., 2020), Cloud (Akbarzadeh et al., 2019), BigData (Shen et al., 2021, Saheb and Izadi, 2019), and so on, the security as well as privacy of digital data are major concern. To secure health data, it becomes important that access to health data should be available only to those who are authenticated strictly (Latha and Sheela, 2019). Otherwise, sensitive data of the patient may get leaked and lead to bad consequences. To secure the data, as initial step, authentication of data requestor is verified. If passed, requestor will be allowed to gain access to eHealth care system. For this, the stakeholder should login into eHealth care system. Prior to login, registration of the user is required. In this course of authentication provisioning, many researchers have proposed different authentication schemes. Like Khan et al. and Kumar et al. (Khan et al., 2022, Kumar et al., 2019), researchers proposed authentication protocols to secure patients’ data. In first place, authentication protocols used different authentication factors categorized under knowledge factor, inherence factor, and possession factor. The factors that depend purely on memory come under knowledge factor; biometrics of individual under inherence; and third-party or external device that helps in authentication is categorized under possession factor. The authentication system that involved two factors was considered most appropriate. These systems not only provide strong authentication but also have very small computation as well as communication overhead. Most widely known authentication strategy, password-based authentication, is exposed to flaws since passwords tend to be short and easy to remember. Using same password for many sites also increases risk of dictionary attacks or attacks by guessing. To share messages securely between client and server, symmetric and asymmetric encryption techniques were used. The most commonly used symmetric techniques are Data Encryption Standard (DES), Blowfish, and Advanced Encryption Standard (AES). Similarly, asymmetric methods like Rivest-Shamir-Adleman (RSA) and Elliptic Curve Cryptosystem (ECC) were used (Mustacoglu et al., 2020). The common challenge for researchers is to frame an authentication scheme that is highly secure with high user usability. But on major side, EHRs are maintained by the third-party authority (TPA). The patient’s data is not under control of corresponding patient and the patient should trust TPA. But, the major concern about these digital records is their security. The TPA may reveal sensitive data of patients knowingly or unknowingly. It is very difficult to gain trust of patient and make them store their health data in the cloud. To gain the trust of patient, patient-centric personal health record (PHR) system emerged as a solution. In general, a PHR system should have some basic features like recommended in (Kim and Johnson, 2002, Park and Yoon, 2020):• Patients should have complete control over their records; • Health record should contain complete health information of patient; • Patient should decide who else can access record and to what level of access; • Patient should be able to view access details of the record. In PHR system, patient will own and maintain personal record. To access the record, patient or any other stakeholder should have proper authentication (Liu and Chung, 2017). The main objective of many researchers was to guarantee the security of PHRs. 1.1 Motivation and contribution Many researchers have proposed many authentication protocols for this eHealth care system. Most of these proposed systems involved TPA who controls the patients’ data. As the data is sensitive, involvement of a third-party should be reduced. In few authentication systems, TPA is not involved. But, these protocols were prone to attacks like stolen-device attack, password guessing attacks, and so on. Also, a delay in receiving One-Time Password (OTP) exists, which hinders authentication verification process. Till now, eHealth care system was designed exclusively for patients. In this proposed paper, following contributions are made• An eHealth care system model is designed for all the rightful citizens. It means that every citizen will have a health record; • A model is designed to guarantee secure communication that thwarted the attacks like password-guessing attacks, and stolen-device attacks; • A model in which exchange of keys is very much minimized thus reducing communication overhead; • Key that is exchanged is not actual key thus ensuring security against key-leakage attack; • Security analysis was performed to prove that proposed protocol withstands several threats; • Performance and security features of proposed protocol are paralleled with prevailing protocols in like environment. This scheme is not only for patients but also for all rightful citizens. So, instead of the word ‘patient’, ‘citizen’ is used wherever needed. 2 2. Related Work Many healthcare applications have emerged in domain of cloud environment (Kumar et al., 2019, Li et al., 2018, Chandrakar et al., 2020, Kumari et al., 2020). As the cloud is involved in healthcare system, and because of third-party involvement, many security problems arise (Selvam and Renjit, 2021). Protocols that were introduced to secure applications involved cryptography methods (Chandrakar and Om, 2018, Ali and Pal, 2018, Kumar et al., 2018). But, these protocols were prone to known attacks. As most of the transactions were done using an insecure channel, many protocols were designed to secure user identity (Shen et al., 2018, Amin et al., 2018). It could be inferred that preserving patients’ anonymity from attackers is a great concern. Among various public-key crypto techniques, ECC provides greater security for small key sizes. Many authentication protocols used ECC for better security (Mahmood et al., 2018, Khatoon et al., 2019, Chaudhry et al., 2020, Alzahrani et al., 2020, Sowjanya et al., 2020, Kirsal Ever, 2019, Chen et al., 2020, Mo et al., 2020). Even then, there exist some security flaws. In work by Qiu et al. (Qiu et al., 2018), the suggested scheme has a flaw in anonymity of user. It was not resistant to various attacks. Attacker was able to access message that transacted in login as well as user verification phase. To preserve patient anonymity, Qi and Chen (Qi and Chen, 2019) suggested an authentication model based on biometrics for a multi-server setup. In 2021, Wang et al. (Wang et al., 2021) offered and also verified security of a proficient privacy-preserving user authentication model along with forward secrecy. A three-factor authentication model was proposed in lieu of user for wearable devices by Jiang et al. (Jiang et al., 2019). In 2021, Sahoo et al. (Sahoo et al., 2021) proposed a three-factor authentication protocol and claimed to resist stolen device, password guessing, and insider attacks. But, protocol was vulnerable to insider attacks. In 2021, Son et al. (Son et al., 2021) proposed a lightweight authentication protocol as it used mostly hash for calculations. The user id, password, and other more valuable data were stored in smart card of user. Without smart card i.e. if card is lost, user does not get authenticated even if the user is legitimate one. In work by Mohit (Mohit, 2021), biometrics of user along with mobile device was used. Biometrics is a unique biological feature that is not replaceable. The biometric of a user is a physical or behavioral entity that may change under unavoidable conditions. For e.g., a user, whose job requires rigorous physical work will have change in fingerprint values. Chandrakar et al. (Chandrakar et al., 2020) has given a protocol for cloud-based healthcare system. But, this was prone to unlinkability, non-repudiation, and impersonation attacks. In this, the upload phase for doctor, patient, and specialist were introduced. But, authentication process of user was not conferred. Moreover, in upload phase, patient's mobile received an OTP as a part of the process. If device is lost then patient has to re-register a new device to receive OTP. Also, sometimes patient has to face a delay in receiving OTP due to network problems. Kaur et al. (Kaur et al., 2022) used ECC for better security. Here also, OTP was sent to mail and mobile as a part of authentication process. But, this protocol was vulnerable to replay attacks. In registration phase of work by Shamshad et al. (Shamshad et al., 2022), the identity of patient was sent to server in its original form. This was easily revealed to attacker. Also, detail of user was stored in a smart card without which the patient cannot login in to the system. If smart card is lost, then until the patient gets a new smart card, healthcare system cannot be accessed. It was determined that in most of protocols, patient has to depend on an external device such as smart card to get authenticated. The genuineness of a patient who is to be registered is not verified. There is no process defined to check whether a particular patient is registered already or not. In a healthcare system, there cannot be a duplicate record of a particular patient. 3 3. Preliminaries 3.1 Aadhaar Server It is a system where details of the citizen are stored and a unique number like a Social security number (SSN) is generated and allotted to that citizen. Here, unique number is referred to as Aadhaar number. 3.2 3.2. Polynomial rings The ZX/Xn+1 ring is represented by R, whereas Rq represents the ring ZqX/Xn+1. Here n=2n′ such that Xn+1 is the 2n′-th is cyclotomic polynomial. 3.3 3.3. Vectors Vector can also be denoted as a matrix. Vectors are generally column-based vectors. For vector v[m], an equivalent matrix is Mat[m][n]. The m in v is the starting index; m, and n in Mat are row and column indexes. If Mat is a matrix then MatT is transpose(Mat). Similarly, v is vector and vT is transpose(v). A Number-Theoretic Transform (NTT) is considered to be an effective approach to do multiplications in Rq. 3.4 Kyber Kyber was first introduced as public key encryption (PKE) algorithm (Bos et al., 2018). Basically, Kyber is constructed based on MLWE, a variant of Learning with Errors (LWE) problem. LWE is a contemporary lattice-based cryptography (Regev, 2009). An m-dimensional integer lattice(L) is given in Eq.(1) as(1) L=L(BV)=∑j=1mkjbvj∈Z where, m < n; n is maximum number of dimensions BV=bv1,bv2,...,bvn is set of basis vectors bv=(a1,a2,...,an) is tuple of real numbers i.e., aj ∈ R. MLWE was presented in (Brakerski et al., 2014). It was the combination of standard LWE and RLWE that resulted in m-dimensional polynomial ring Rpm . An m-dimensional integer ring Zp was used in LWE. In RLWE, a 1-dimensional polynomial ring Rp=Zp/xd+1 was used. The PKE process of Kyber resembles that of RLWE-based Lyubashevsky, Peikert, and Regev’s LPR system in (Lyubashevsky et al., 2010). Secret key, sk, is a vector chosen from a matrix Ar in NTT domain i.e., Ar^∈Rqk∗k. The vector of errors, ve belongs to Rqk. The public key, f^=Ar^.sk^+ve^ where sk^ , ve^ are the NTT representations of sk, and ve respectively. In the process of encrypting a message msg, Ar^ is transposed. Then sa, sb are chosen from Rqk. Similarly, sc is chosen from Rq. Then, f^ and transpose(Ar^) is converted from NTT to standard order as f and transpose(Ar). A pair (a, b) is generated and given in Eq.(2) and Eq.(3), where fcf (2) a=ArT.sa+sb (3) Then, using a and b, part cipher texts cp1 and cp2 are generated respectively. The combined cp1 and cp2 gives full cipher text cp. To decrypt, the secret key sk and the pair (a, b) are used. Eq.(4) is calculated to generate plain text msg.(4) b-skT.a=(veT.sa-skT.sb+sc+msg).⌊2/q⌉mod+2 At first, Kyber was Indistinguishability Under Chosen-Password Attack (IND-CPA)-secure. Later, it was transformed to Indistinguishability Under Adaptive Chosen-Ciphertext Attack (IND-CCA2)-secure key-encapsulation mechanism (KEM) using moderately modified Fujisaki–Okamoto (FO) transform presented in (Fujisaki and Okamoto, 2013). The Kyber-KEM uses key derivation function (KDF) upon Kyber-PKE encryption and decryption functions to have the shared key. KEM encapsulation and decapsulation processes were given in (Avanzi et al., 2021, Eriksson, 2020). 3.5 Galois/Counter Mode-Advanced Encryption Standard (GCM-AES) In proposed protocol, GCM (Dworkin, 2007), a mode of operating AES is used. AES is a symmetric key encryption algorithm. This was accepted and detailed in (NIST, 2001). GCM-AES is used at places where symmetric encryption is needed. The GCM encryption was defined in (Dworkin, 2007) given in Eq.(5) as(5) GCM-AEK(IV.P,A)=(C,T) The GCM decryption was defined in (Dworkin, 2007) given in Eq.(6) as(6) GCM-ADK(IV,C,A,T)=PorFAIL where, IV is initial vector, Cipher text C, A is Additional authenticated data, and Authenticated code T. The encryption is done using key K. In decryption, code T’ is calculated. If T=T’ then the function outputs plain text P, otherwise FAIL message. 4 Notations The notations used in this paper are represented in Table 1 .TABLE 1 Some related notations. CS Client System AS Authentication Server ADS Aadhaar Server CHIS Citizen Health Information Server PHC Primary Healthcare Center Cu Citizen user nm name of citizen v, v' virtual id of citizen p passcode of citizen c communication address of citizen hs hashed passcode SE(…) Symmetric Encryption function SD(…) Symmetric Decryption function insert(…) Data insertion function derive(..) key derivation function hid, hid', hidc, hidn health identification number tl timer limit tud,tpd timer differences otac one time authentication code msg, msg' Message key, key' derived key bio,bio' biometrics of citizen otvid, otvid' one time verification id 5 4. Proposed Protocol In proposed protocol, eHealth care system is modeled for all the rightful citizens. Every citizen will have a health record that holds every single health detail. As health data is more sensitive data, more security is provided through proposed protocol. At first, the citizen has to register with the system. In the case of a minor citizen, the parent or authorized guardian will control the health record. Upon successful registration, a login process to the system can be made. After successful login, user can access their medical record. In this protocol, Initially, in login phase, for secure transaction, KEM algorithm Kyber (Bos et al., 2018) is used. Also, Kyber is quantum-resistant (Bos et al., 2018). Later for further transactions in the session, GCM-AES is used as symmetric encryption algorithm. The key used in GCM-AES is generated using a new key derivation method. This is discussed in section 4.1. In this proposed protocol, at first, the citizen will register some basic details through the application user interface of Citizen Health Information Management System (CHIMS). Upon proper registration, the citizen will get a hid along with a code. The citizen will visit nearest PHC and give the hid,v to Nodal officer(No) in order to complete registration process. No will check the rightfulness of citizen. If yes, then registration process is completed. To access the health record, the citizen will make a successful login and mutually authenticate with the CHIS. The overall process of this protocol is secure, efficient and depicted in figure Fig. 1 . The servers in this proposed protocol are considered to be secure. The following are different phases of proposed protocol.FIG. 1 Overall architecture of proposed protocol. 5.1 Key Derivation Method In key generation process, key is derived from ohscode using otac. The ohscode is calculated as a part of kohscode calculation. The key generated here is used as the key for symmetric encryption and decryption of the data that are being transferred. This key is only known to the server and the corresponding Cu. This key will change for each session. This is because the key depends on ohscode and otac which changes for every session. The process of key derivation is depicted in Algorithm 1.Algorithm 1 Key Derivation Method Input: otac, ohscode Output: keyInitialize k ←“” Divide ohscode into 4 equal blocks, say Bi where i=0,1,2,3 for i ←0 to 3 do for j←0 to length(otac)-1 do if otac[j] in {0,1,…,9} then key ←key + Bi[otac[j]] otherwise if otac[j] in {a,b,…,z} then key ←key + Bi[Ascii(otac[j])-97] otherwise if otac[j] in {A,B,…,Z} then key ←key + Bi[31-(Ascii(otac[j])-65)] return key The key used to encrypt and decrypt the data to be transferred between the server and the user is different from the key used among the server for secure storage of the data. 5.2 Registration Phase In registration phase, the first time user will register. This phase has two sub-phases. In sub-phase 1, the user provides basic details such as name (nm), virtual Aadhaar number (v), passcode (scode), and communication address(c) from a client system (CS). The CS then calculates hscode=h(scode) and sends the encrypted details to an authentication server (AS). The AS decrypts the details, retrieves a health ID (hidc) from CHIS, generates a new health ID (hidn), and stores encrypted information (SE) along with a one-time verification ID (otvid) at AS. The encrypted hidn and otvid are then sent back to CS, where they are decrypted. In sub-phase 2, to complete the registration process, user has to visit any registered primary healthcare center (PHC). The nodal officer at PHC gets the hidn and virtual id (v) from Cu and sends it to AS in encrypted format through CHIMS. The AS decrypts the information. Based on the received v, the AS checks for the details with the Aadhaar server (ADS). If successful, the biometric details are matched with Cu and checked with the otvid. If success, then one time authentication code (otac) is generated and along with the nm,v,c are shared to AS, CHIS in encrypted format. The otac is also shared with Cu. Sub-phase 1At client system (CS), the user (Cu) should give the name (nm), virtual Aadhaar number (v) and passcode (scode), communication address (c); At CS, the hscode=h(scode) is calculated. Send Enc(nm, v, hscode, c) to authentication server (AS); At AS, Denc(Enc(nm, v, hscode, c)) is calculated. Retrieve current health id (hidc) from CHIS. Then next health id (hidn) is generated and allotted to Cu. Randomly, a one-time verification id (otvid) is generated. The SE(hidn, nm, v, hs, c,otvid) is stored at AS; At AS, Enc(hidn,otvid) is calculated and sent to CS; At CS, Denc(Enc(hidn,otvid)). The process flow of sub-phase 1 is given in figure Fig. 2 FIG. 2 Registration phase:Sub-phase-1. Sub-phase 2Cu should visit a nearest PHC; Cu gives the hidn to No. No logs-in into the CHIMS. If hidn is found and registration is partial, then the No will proceed the process; In the system, if hid'=hid then, if AS(v)=ADS(v') then, if bio=bio' then, if otvid= otvid' generate otac. Give otac to Cu. Also, update Cu details with new otac at AS. Send hidn,nm,v,c to CHIS. The process flow of sub-phase 2 is given in figure Fig. 3 FIG. 3 Registration phase:Sub-phase-2. 5.3 Login cum Authentication phase In this phase, at CS, Cu enters the hid, scode and otac. The ohscode= h(h(scode),otac) is calculated. Then the key is derived from ohscode using otac. Calculate kohscode=h(ohscode,key). Send Enc(pid,kohscode) to AS. The Denc(Enc(pid, kohscode)) extracts the hid and kohscode. Parallelly, the AS calculates the kohscode', key' locally; If hid=hid' then, if kohscode=kohscode' then, allow access of CHIS to Cu and also generate otacn. Encrypt otacn as SE(otacn) using key'. Send this to the Cu; Also, send SE(hid,kohscode', key') to CHIS. At CHIS, SD(SE(hid,kohscode', key')) gives the hid,kohscode', key'; Cu sends SE(kohscode, msg, tu1) using key to CHIS. CHIS retrieves kohscode,msg,tu1 and finds tud=tu1-tu2 where tu2 is the timer value when the message is reached at CHIS. CHIS sends SE(msg,tp1,tud) using key' to Cu. Cu retrieves msg,tp1 and finds tpd=tp1-tp2 where tp2 is the timer when the message is reached at CS. If tud==tpd and if msg==msg' then the receiver is the intended one. The td is maintained for further transactions. The process flow of login and authentication are depicted in figures Fig. 4 and Fig. 5 respectively.FIG. 4 Login process flow. FIG. 5 Authentication process flow. 6 5. Results The proposed protocol is analyzed both in terms of security and performance. The result is compared with related protocols and the inferences are discussed. 6.1 Security Analysis 6.1.1 Informal analysis 6.1.1.1 Anonymity and privacy In the proposed protocol, the Cu details are sent in an encrypted format. The AS receives the details and decrypts them. The passcode sent is in a hashed format. The Cu details are stored in AS in an encrypted format. Moreover, most of the transactions are done in a secure and encrypted format. The attacker will not be able to get the details of the Cu. 6.1.1.2 No key stolen attack via eavesdropping The key that is used in secure transaction of the data during a session is not actually shared or transferred between the user and the server. Rather, the key is generated at client, and AS using the ohscode, otac of a particular user. This key also changes for every new session, because the otac changes for every new session. So, the key cannot be stolen as it is not transferred. 6.1.1.3 Mutual Authentication The Cu and the CHIS are mutually authenticated by sharing the details that are encrypted using the key. This key is only known to Cu, AS, and CHIS. The interesting fact is that the key is not transferred among them. Rather, the key is generated at Cu, AS and CHIS in the same way. The attacker in no way can get the key. 6.1.1.4 Secure data storage The data when transferred between user and server, it is secured using GCM-AES using key. When the data is stored in CHIS, CHIS uses its own key (ko) to secure the data. Whenever a request for the data from an authentic user is made, at first CHIS will decrypt the data using ko and then encrypt the data with key. Then the encrypted data is send to the requestor. The requestor will decrypt the data. As mentioned earlier, the key is only known to the user and server. 6.1.1.5 Replay attack For every new session, at CS a new kohscode is calculated and its encrypted format is sent to the AS. kohscode is also calculated at AS. If kohscode=kohscode’ then Cu will move to the next process. To generate the kohscode, the attacker needs to get the passcode, otac which are not possible. Also, the actual key is not exchanged. To counter this attack, the countdown timer is also used. 6.1.1.6 Device stolen attack In this protocol, dependency on a third-party device like a smart card or mobile phone is not involved. So, there is no scope for a device stolen attack. 6.1.1.7 No clock synchronization The timestamp is not involved in this proposed protocol. Rather, for each session, countdown timers such as tu and tp are used. A maximum timer limit tl is also used. So, clock synchronization is not necessary. The security analysis of some attacks on related protocols was compared and given in Table 2 .TABLE 2 Security analysis of some attacks on proposed protocol compared with the related protocols. Prop. (Shamshad et al., 2022) (Islam and Khan, 2014) (Chaudhry et al., 2015) (Qiu et al., 2018) (Wei et al., 2012) (Tu et al., 2015) Provide anonymity and privacy 1 1 1 1 1 0 - Thwarts privilege insider attack 1 1 1 1 1 1 1 Prevents offline password-guess attack 1 1 0 0 0 0 1 Thwart replay attack 1 1 1 1 1 0 1 Thwart man-in-the-middle attack 1 1 0 0 1 0 - Provide mutual authentication 1 1 0 1 0 0 1 Thwart clock synchronization problem 1 1 0 0 0 0 - Thwart smart card stolen attack 1 1 0 0 0 0 0 Stops device dependency problem 1 0 0 0 0 0 0 6.1.2 Formal analysis The protocol algorithm is formally analyzed using the ProVerif tool. The version used here is of 2.02 which is a plugin for the Eclipse 4.7.3a version. ProVerif is a tool for automatic verification of security protocols (Blanchet et al.,2021). In ProVerif, protocols are modeled and verified using the process calculus given at (Blanchet et al.,2021). This is a variant of applied π-calculus. To support the modeling of crypto-functions used in security protocols, the pi-calculus is used along with the rich term algebra (Delaune et al., 2008). The proposed protocol successfully passed ProVerif test and there are no possible threats found from the attacker. Also, mutual authentication of Cu and CHIS provided by protocol is proved. The secrecy of the Cu is maintained using this protocol. 6.2 Performance analysis In this part, performance of suggested protocol is compared with related protocols. The computational cost of proposed protocol is calculated by considering execution time of one-way hash function, asymmetric as well as symmetric encryption, and decryption functions. The system specifications for the analysis of protocols are shown in table 3. Also, the execution time of the crypto-function such as hash function (Tth) is 0.00146ms, asymmetric encryption/decryption (Tencn/Tdecn) is 0.01817ms/0.02310ms, and symmetric encryption/decryption (Tse/Tsd) is 0.01363ms/0.01020ms. These are based on the specifications in Table 3 .TABLE 3 System Specifications. Element Specification System DELL Generation Core i5 RAM 4 GB Processor 2.30 GHz OS Windows 7 The estimated time of computational cost that is calculated for various protocols is given in Table 4 . The computational cost of the proposed protocol is calculated as 7Th+3Tencn+3Tdecn+9Tse+8Tsd+5T\|\| ≈ 0.3381ms. Also, communication cost for the same is shown in table 4.TABLE 4 Computational, communication, and storage cost of proposed and related protocols. Scheme Total Computation cost (milliseconds) Communication cost (bits) Storage cost (bits) Proposed 0.3381 2816 1162 (Shamshad et al., 2022) 0.4858 2240 1216 (Islam and Khan, 2014) 0.6884 3040 1376 (Chaudhry et al., 2015) 0.7404 2240 1440 (Qiu et al., 2018) 0.5468 1664 1568 (Wei et al., 2012) 0.3432 2755 1152 (Tu et al., 2015) 0.6486 1920 416 The comparisons that are made in Table 4 are plotted as a graph and depicted in figure Fig. 6 FIG. 6 Computational cost comparison of proposed and related protocols in milliseconds. The protocols were also checked for multiple iterations. The computational cost of the protocols at respective iterations are calculated and plotted as a graph in figure Fig. 7 .FIG. 7 Computational cost comparison of proposed and related protocols in milliseconds over multiple iterations. 7 6. Discussions Though the communication cost of the proposed protocol is higher than (Qiu et al., 2018, Shamshad et al., 2022, Chaudhry et al., 2015, Wei et al., 2012, Tu et al., 2015), the storage cost is lesser than (Qiu et al., 2018, Shamshad et al., 2022, Islam and Khan, 2014, Chaudhry et al., 2015) and only 10 bits more than (Tu et al., 2015). The cost of computation for suggested protocol is 43.69%, 103.61%, 118.99%, 61.74%, and 91.82% less than (Shamshad et al., 2022, Islam and Khan, 2014, Chaudhry et al., 2015, Qiu et al., 2018, Tu et al., 2015) respectively. The computational cost of (Wei et al., 2012) seems to be same as that of the proposed protocol but is slightly 1.50% greater than the proposed protocol. It is determined from the comparisons made above that security and performance analysis of proposed protocol is better than other related protocols. 8 7. Conclusion The proposed protocol is a novel secure and quantum-resistant authentication protocol modeled for the rightful citizens to handle their eHealth records. This employs a combination between asymmetric and symmetric cryptographic algorithm using kyber and GCM-AES in for secure data transfer. The overhead of key exchange is minimized to much extent. The new key generation method proposed here is a unique and efficient one. This protocol is well-tested for many security traits and performed better than their relative protocols. Even though the communication cost is high than the relative schemes, the computation cost is better than its counterpart. This proposed protocol provides better security to the eHealth records of the citizens which enables the citizens to be part of health record digitization. As the future work, the authorization to access the data will be allied to the proposed protocol to provide higher level security to the data. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References Akbarzadeh A. Bayat M. Zahednejad B. Payandeh A. Aref M.R. A lightweight hierarchical authentication scheme for internet of things J. Ambient Intell. Humaniz. Comput. 10 2019 2607 2619 10.1007/s12652-018-0937-6 Ali R. Pal A.K. Cryptanalysis and Biometric-Based Enhancement of a Remote User Authentication Scheme for E-Healthcare System Arab. J. Sci. Eng. 43 2018 7837 7852 10.1007/s13369-018-3220-4 Alzahrani B.A. Chaudhry S.A. Barnawi A. Al-Barakati A. Alsharif M.H. A Privacy Preserving Authentication Scheme for Roaming in IoT-Based Wireless Mobile Networks Symmetry (Basel). 12 2020 10.3390/sym12020287 Amin R. Islam S.K.H. Biswas G.P. Khan M.K. Kumar N. A robust and anonymous patient monitoring system using wireless medical sensor networks Futur. Gener. Comput. Syst. 80 2018 483 495 10.1016/j.future.2016.05.032 Avanzi, R., Bos, J., Ducas, L., Kiltz, E., Lepoint, T., Lyubashevsky, V., Schanck, J.M., Schwabe, P., Seiler, G., Stehlé, D., 2021. CRYSTALS-Kyber Algorithm Specifications And Supporting Documentation (version 3.01). Blanchet, B., Smyth, B., Cheval, V., Sylvestre, M., n.d. ProVerif 2.04: Automatic Cryptographic Protocol Verifier, User Manual and Tutorial (2021). Bos, J., Ducas, L., Kiltz, E., Lepoint, T., Lyubashevsky, V., Schanck, J.M., Schwabe, P., Seiler, G., Stehlé, D., 2018. CRYSTALS - Kyber: A CCA-Secure Module-Lattice-Based KEM. 2018 IEEE Eur. Symp. Secur. Priv. https://doi.org/10.1109/EuroSP.2018.00032 Brakerski Z. Gentry C. Vaikuntanathan V. (Leveled) fully homomorphic encryption without bootstrapping ACM Trans. Comput. Theory 6 2014 10.1145/2633600 Chandrakar P. Om H. An extended ECC-based anonymity-preserving 3-factor remote authentication scheme usable in TMIS Int. J. Commun. Syst. 31 2018 e3540 Chandrakar P. Sinha S. Ali R. Cloud-based authenticated protocol for healthcare monitoring system J. Ambient Intell. Humaniz. Comput. 11 2020 3431 3447 10.1007/s12652-019-01537-2 Chaudhry S.A. Naqvi H. Shon T. Sher M. Farash M.S. Cryptanalysis and Improvement of an Improved Two Factor Authentication Protocol for Telecare Medical Information Systems J. Med. Syst. 39 2015 66 10.1007/s10916-015-0244-0 25912427 Chaudhry S.A. Shon T. Al-Turjman F. Alsharif M.H. Correcting design flaws: An improved and cloud assisted key agreement scheme in cyber physical systems Comput. Commun. 153 2020 527 537 10.1016/j.comcom.2020.02.025 Chen C.-L. Huang P.-T. Deng Y.-Y. Chen H.-C. Wang Y.-C. A secure electronic medical record authorization system for smart device application in cloud computing environments Human-centric Comput. Inf. Sci. 10 2020 21 10.1186/s13673-020-00221-1 Delaune S. Ryan M. Smyth B. Automatic Verification of Privacy Properties in the Applied pi Calculus Karabulut Y. Mitchell J. Herrmann P. Jensen C.D. Trust Management II 2008 Springer US, Boston, MA 263 278 Dworkin M. Recommendation for Block Cipher Modes of Operation: Galois/Counter Mode (GCM) and GMAC Gaithersburg, MD. 2007 10.6028/NIST.SP.800-38D Eriksson H. Implementing and Evaluating the Quantum Resistant Cryptographic Scheme Kyber on a Smart Card 2020 Linköping University Fujisaki E. Okamoto T. Secure integration of asymmetric and symmetric encryption schemes J. Cryptol. 26 2013 80 101 10.1007/S00145-011-9114-1/METRICS Islam S.K.H. Khan M.K. Cryptanalysis and improvement of authentication and key agreement protocols for telecare medicine information systems J. Med. Syst. 38 2014 135 10.1007/s10916-014-0135-9 25190590 Jiang Q. Ma J. Yang C. Ma X. Shen J. Chaudhry S.A. Efficient end-to-end authentication protocol for wearable health monitoring systems Comput. Electr. Eng. 63 2017 182 195 10.1016/j.compeleceng.2017.03.016 Jiang Q. Qian Y. Ma J. Ma X. Cheng Q. Wei F. User centric three-factor authentication protocol for cloud-assisted wearable devices Int. J. Commun. Syst. 32 2019 e3900 kaur, S., kaur, G., Shabaz, M., A Secure Two-Factor Authentication Framework in Cloud Computing Secur. Commun. Networks 2022 2022 7540891 10.1155/2022/7540891 Khan A.A. Kumar V. Ahmad M. An elliptic curve cryptography based mutual authentication scheme for smart grid communications using biometric approach. J. King Saud Univ. - Comput Inf. Sci. 34 2022 698 705 10.1016/j.jksuci.2019.04.013 Khatoon S. Rahman S.M.M. Alrubaian M. Alamri A. Privacy-Preserved, Provable Secure, Mutually Authenticated Key Agreement Protocol for Healthcare in a Smart City Environment IEEE Access 7 2019 47962 47971 10.1109/ACCESS.2019.2909556 Kim M.I. Johnson K.B. Personal health records: evaluation of functionality and utility J. Am. Med. Inform. Assoc. 9 2002 171 180 10.1197/jamia.m0978 11861632 Kirsal Ever Y. Secure-Anonymous User Authentication Scheme for e-Healthcare Application Using Wireless Medical Sensor Networks IEEE Syst. J. 13 2019 456 467 10.1109/JSYST.2018.2866067 Kumar V. Jangirala S. Ahmad M. An Efficient Mutual Authentication Framework for Healthcare System in Cloud Computing J. Med. Syst. 42 2018 142 10.1007/s10916-018-0987-5 29956007 Kumar V. Ahmad M. Kumari A. A secure elliptic curve cryptography based mutual authentication protocol for cloud-assisted TMIS Telemat. Informatics 38 2019 100 117 10.1016/j.tele.2018.09.001 Kumari A. Kumar V. Abbasi M.Y. Kumari S. Chaudhary P. Chen C.-M. CSEF: Cloud-Based Secure and Efficient Framework for Smart Medical System Using ECC IEEE Access 8 2020 107838 107852 10.1109/ACCESS.2020.3001152 Latha K. Sheela T. Block based data security and data distribution on multi cloud environment J. Ambient Intell. Humaniz. Comput. 2019 10.1007/s12652-019-01395-y Li C.-T. Shih D.-H. Wang C.-C. Cloud-assisted mutual authentication and privacy preservation protocol for telecare medical information systems Comput. Methods Programs Biomed. 157 2018 191 203 10.1016/j.cmpb.2018.02.002 29477428 Liu C.-H. Chung Y.-F. Secure user authentication scheme for wireless healthcare sensor networks Comput. Electr. Eng. 59 2017 250 261 10.1016/j.compeleceng.2016.01.002 Lyubashevsky V. Peikert C. Regev O. On ideal lattices and learning with errors over rings Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2010 Springer Berlin, Heidelberg 1 23 10.1007/978-3-642-13190-5_1/COVER Mahmood K. Naqvi H. Alzahrani B.A. Mehmood Z. Irshad A. Chaudhry S.A. An ameliorated two-factor anonymous key exchange authentication protocol for mobile client-server environment Int. J. Commun. Syst. 31 2018 Mo J. Shen W. Pan W. Guerrieri A. An Improved Anonymous Authentication Protocol for Wearable Health Monitoring Systems Wirel. Commun. Mob. Comput. 2020 2020 10.1155/2020/5686498 Mohit P. An efficient mutual authentication and privacy prevention scheme for e-healthcare monitoring J. Inf. Secur. Appl. 63 2021 102992 10.1016/j.jisa.2021.102992 Mustacoglu A.F. Catak F.O. Fox G.C. Password-based encryption approach for securing sensitive data Secur. Priv. 3 2020 e121 Nikou S. Agahari W. Keijzer-Broers W. de Reuver M. Digital healthcare technology adoption by elderly people: A capability approach model Telemat. Informatics 53 2020 101315 10.1016/j.tele.2019.101315 NIST, 2001. Announcing the ADVANCED ENCRYPTION STANDARD (AES). https://doi.org/10.6028/NIST.FIPS.197 Park Y. Yoon H.-J. Understanding Personal Health Record and Facilitating its Market Healthc. Inform. Res. 26 2020 248 250 10.4258/hir.2020.26.3.248 32819044 Qi M. Chen J. Anonymous biometrics-based authentication with key agreement scheme for multi-server environment using ECC Multimed. Tools Appl. 78 2019 27553 27568 10.1007/s11042-019-07812-w Qiu S. Xu G. Ahmad H. Wang L. A Robust Mutual Authentication Scheme Based on Elliptic Curve Cryptography for Telecare Medical Information Systems IEEE Access 6 2018 7452 7463 10.1109/ACCESS.2017.2780124 Regev O. On lattices, learning with errors, random linear codes, and cryptography J. ACM 56 2009 10.1145/1568318.1568324 Saheb T. Izadi L. Paradigm of IoT big data analytics in the healthcare industry: A review of scientific literature and mapping of research trends Telemat. Informatics 41 2019 70 85 10.1016/j.tele.2019.03.005 Sahoo S.S. Mohanty S. Majhi B. A secure three factor based authentication scheme for health care systems using IoT enabled devices J. Ambient Intell. Humaniz. Comput. 12 2021 1419 1434 10.1007/s12652-020-02213-6 Selvam L. Renjit J.A. On developing dynamic and efficient cryptosystem for safeguarding healthcare data in public clouds J. Ambient Intell. Humaniz. Comput. 12 2021 3353 3361 10.1007/s12652-020-02033-8 Shamshad S. Ayub M.F. Mahmood K. Kumari S. Chaudhry S.A. Chen C.M. An enhanced scheme for mutual authentication for healthcare services Digit. Commun. Networks 8 2022 150 161 10.1016/j.dcan.2021.07.002 Shen J. Gui Z. Ji S. Shen J. Tan H. Tang Y. Cloud-aided lightweight certificateless authentication protocol with anonymity for wireless body area networks J. Netw. Comput. Appl. 106 2018 117 123 10.1016/j.jnca.2018.01.003 Shen J. Liu D. Liu Q. Sun X. Zhang Y. Secure Authentication in Cloud Big Data with Hierarchical Attribute Authorization Structure IEEE Trans. Big Data 7 2021 668 677 10.1109/TBDATA.2017.2705048 Son S. Park Y. Park Y. A Secure, Lightweight, and Anonymous User Authentication Protocol for IoT Environments Sustainability 13 2021 10.3390/su13169241 Sowjanya K. Dasgupta M. Ray S. An elliptic curve cryptography based enhanced anonymous authentication protocol for wearable health monitoring systems Int. J. Inf. Secur. 19 2020 129 146 10.1007/s10207-019-00464-9 Tu H. Kumar N. Chilamkurti N. Rho S. An improved authentication protocol for session initiation protocol using smart card Peer-to-Peer Netw. Appl. 8 2015 903 910 10.1007/s12083-014-0248-4 Wang C. Wang D. Xu G. He D. Efficient privacy-preserving user authentication scheme with forward secrecy for industry 4.0 Sci. China Inf. Sci. 65 2021 112301 10.1007/s11432-020-2975-6 Wei J. Hu X. Liu W. An Improved Authentication Scheme for Telecare Medicine Information Systems J. Med. Syst. 36 2012 3597 3604 10.1007/s10916-012-9835-1 22374237
PMC010xxxxxx/PMC10287188.txt	==== Front Infect Dis Model Infect Dis Model Infectious Disease Modelling 2468-2152 2468-0427 The Authors. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. S2468-0427(23)00044-1 10.1016/j.idm.2023.05.009 Article The effect of non-pharmaceutical interventions on COVID-19 outcomes: A heterogeneous age-related generalisation of the SEIR model Mendes Jorge M. a∗ Coelho Pedro S. ab a NOVA Information Management School (NOVAIMS), Universidade Nova de Lisboa, Campus de Campolide, 1070-312, Lisbon, Portugal b NOVA Cairo at the Knowledge Hub Universities, New Admnistrative Capital, Cairo, Egypt ∗ Corresponding author. 22 6 2023 22 6 2023 21 9 2022 26 5 2023 29 5 2023 © 2023 The Authors 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Successive generalisations of the basic SEIR model have been proposed to accommodate the different needs of the organisations handling the SARS-CoV-2 epidemic and the assessment of the public health measures adopted and named under the common umbrella of Non-Pharmaceutical Interventions (NPIs). So far, these generalisations have not been able to assess the ability of these measures to avoid infection by the SARS-CoV-2 and thus their contribution to contain the spread of the disease. This work proposes a new generalisation of SEIR model and includes a heterogeneous and age-related generation of infections that depends both on a probability that a contact generates the transmission of the disease and a contact rate. The results show (1) thanks to the universal wearing of facial coverings, the probability that a contact provokes the transmission of the disease was reduced by at least 50% and (2) the impact of the other NPI is so significant that otherwise Portugal would have gone into a non-sustainable situation of having 80% of its population infected in the first 300 days of the pandemic. This situation would have led to a number of deaths almost twenty times higher than the number that was actually recorded by December 26th, 2020. Moreover, the results suggest that even if the requirement of universal wearing of facial coverings was adopted sooner jointly with closing workplaces and resorting to teleworking would have postponed the peak of the incidence, altought the epidemic path would have result in a number of infections hardly managed by the National Health System. Complementary, results confirm that (3) the health authorities adopted a conservative approach on the criteria to consider an infected individual not infective any longer; and (4) the most effective NPIs and stringency levels either impacting on self-protection against infection or reducing the contacts that would eventually result in infection are, in decreasing order of importance, the use of Facial coverings, Workplace closing and Stay at home requirements. Handling Editor: Dr Daihai He ==== Body pmc1 Introduction In December 2019, a new coronavirus named Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV-2), causing severe acute respiratory disease emerged in the region of Wuhan, China (Zhu et al. (2020) and Chan et al. (2020)). SARS-CoV-2 is an acute respiratory infectious disease that spreads through the respiratory tract by droplets, respiratory secretions, person-to-person contact and objects or materials which are likely to carry infection, such as clothes, utensils, and furniture (Chen (2020), Wang et al. (2020) and Rosa et al. (2020)). At the time of writing this introduction almost all European countries have been through at least five “waves”, but the effect induced by the high vaccination coverage started to appear (Gozzi, Bajardi, and Perra (2021)). Consequently, many countries began to ease some of the heavy measures in force and the mandatory use of face masks was even under judicious evaluation. Indeed, in the fight against this coronavirus and the absence of a vaccine or treatment, the international community was more interested in putting in force measures to control the spread of the virus and/or mitigate it. The extent (stringency and duration) which these measures were adopted depends either on the epidemiological situation, the population being targeted or the governments sensitivity to their economic impact. Furthermore, the actions taken potentially mitigate the spread of infection by the SARS-CoV-2 and contribute to preserving the health of citizens (Ferguson et al. (2020); Nunan and Brassey (2020)). It is possible to organise the spectra of all mitigation measures taken under a common umbrella named Non-Pharmaceutical Interventions (NPIs), that is, any public health measures that aim to prevent and/or control infection transmission in the community. During the period when there was no effective and safe vaccine to protect those at risk of severe COVID-19, NPIs were seen as the most effective public health instruments against COVID-19. The European Centre for Disease Prevention and Control (ECDC) guidelines detail available options for NPIs in various epidemiological scenarios, assess the evidence for their effectiveness and address implementation issues, including potential barriers and facilitators (European Centre for Disease Prevention, European Center for, and Control, 2020, Disease Prevention, European Centre for, and Control, 2020). According to the ECDC, NPIs can be classified into three levels of implementation: individual, environmental and population levels. In parallel to these guidelines, the ECDC keeps tracks of selected national public NPIs presented in the weekly COVID-19 country overviews report. The best available scientific evidence is required to design effective NPIs and disseminate the knowledge to help public officials assess the potential benefits and costs of NPIs to contain COVID-19 outbreaks, as it is to expect that different measures will also present different levels of cost-effectiveness. Therefore, it is essential to describe how different countries implemented NPIs, and at what point of the epidemic. It is also necessary to explore how those NPIs have impacted the number of cases, the mortality, and the capacity of health care facilities to deliver healthcare services. As early as December 2022, a search of published scientific material on “Covid-19” and NPIs (searched as non-pharmacologial interventions, non-pharmacological interventions or non-pharmaceutical) on ISI Web of Knowledge returned 452 entries. A refinement based on scientific areas, publishing outlets and topics addressed restricted the number of publications, yet far above 200. A read over their abstracts unveiled a broad spectrum of used methodologies, although their conclusions were coincident to some extent. It is not the authors’ purpose to perform a systematic literature review here, especially when others have been successfully doing so. Therefore, regarding the different methodologies used to perform the NPIs assessment, we refer to the outstanding systematic review by Banholzer et al. (2022). In this work, the authors systematically review the literature on assessing the effectiveness of non-pharmaceutical interventions between January 1, 2020 and January 12, 2021, using a total of 248 publications. Although despite the substantial variation in methodologies with respect to study setting, outcome, intervention, methodological approach, and effectiveness assessment, which prevents comparability among studies, the heterogeneity in the used methodologies may be desirable to assess the robustness of results, the study points out to shortcomings of existing studies and make recommendations for the design of future ones. Regarding the empirical studies and their results about the effectiveness of NPIs we refer to the excellent review carried out by Mendez-Brito et al. (2021). In this review the authors followed the a systematic methodology (PRISMA) (Moher et al. (2009)) to search for published literature and preprints, respectively, available in English from January 1, 2020 to the beginning of March 2021. The authors rank the NPIs according to their effectiveness in reducing reproduction number, infection growth rate, and other incidence-related measurements. Overall, school closing was the most effective measure in reducing the number of cases. As they point out, several authors report a mean reproduction number reduction after the closures of educational facilities. Indeed, after the closure of schools and universities Brauner et al. (2021) estimated a mean reproduction number reduction of 39%. In turn, Haug et al. (2020) estimated a reduction of 73% after the school closing. Workplace closing to at greater extent and business or venue closing and public event bans to a less one are considered the most effective in reducing cases. Other NPIs such as lockdowns, movement limitations through national or international travel restrictions, social gathering bans ranging from 10 people to mass gathering bans, social distancing, public information campaigns and mask-wearing requirements are reported as a group of intermediate effectiveness yet still important. Several authors seem to have found no evidence of the effectiveness of other NPIs like public transport closure, testing and contact tracing strategies and quarantining or isolation of individuals. Also, limited work and evidence indicate that stringency may play an important role when adopting a NPI (Liu et al. (2021)). Finally, the specific topic of face masks has also been addressed, suggesting that the broad adoption of even relatively ineffective face masks may meaningfully reduce community transmission of COVID-19. However, several of these results are contradictory, limited in outcomes analysed (e.g. limited to the reproduction number) and partial in the sense that they do not address the combined effect of different NPIs and, in particular, the role of face masks in the presence of other NPIs (Eikenberry et al. (2020)). Although almost all the conclusions of the empirical studies have pointed out to coincident effects of some NPIs regarding their ability to mitigate or stop disease spread, several factors have also been reported to be of critical importance for NPIs effective impact (e.g. Barbarossa and Fuhrmann (2021)). These factors affect the population's compliance to adopting NPIs, which can even be graded in concern, government performance's perception or agreement and compliance (e.g.} Santos et al. (2022)). Their nature ranges from demographic, social and psychological (e.g. Seale et al. (2020), Shittu et al. (2022), Hengartner et al. (2022), Downing et al. (2022) and Sopory et al. (2022)). As Kretzschmar et al. (2022) points out, mathematical modelling and statistical inference provide a framework to evaluate different non-pharmaceutical and pharmaceutical interventions to control epidemics. However, addressing the challenges in modelling the health, economic and political aspects of interventions need a wide variety of interdisciplinary expertise and cross-collaboration between scientists and policymakers, combining mathematical knowledge with biological and social insights, including health economics and communication skills. All things considered, there are still crucial knowledge gaps on the effectiveness of different interventions to adequately justify the preparation, implementation, or cancellation of various NPIs. In addition, governments across the World need evidence as to the combination and timings of each potential NPI, which remains lacking (e.g.@Patino2020). Despite this discussion, the ultimate goal of the different NPIs referred to above is to reduce the number of contacts between humans or lower the probability of transmission, especially between the most exposed to infection and those who exhibit particular vulnerabilities, and therefore mitigate and contain the spread of the infection. The reduction in human contacts is hard to quantify, but proxy variables exist that aid the investigation of NPIs effects, Google (https://google.com/covid19/mobility/) or Apple mobility (https://covid19.apple.com/mobility) data are two examples of those proxies. The motivation of this work is anchored on the current lack of knowledge about the effectiveness of different NPIs. Additionally, the authors aim to provide a comprehensive framework to assess the effect of different NPIs adopted during the pandemic. Indeed, using an extension of the widely known compartmental model, the S usceptible- E xposed- I nfected- R emoved (SEIR) model, the authors set up a methodology to assess the direct effects NPIs on population mobility and their indirect impacts on the spread of the infection. Additionally, simulations are performed to assess what would have happened if those NPIs or only a subset of them were adopted. Therefore, its purpose is to understand how NPIs have affected the mobility of citizens and the probability of disease transmission. These data are subsequently used to mimic how the contact rate evolved and how it impacts the daily generation of confirmed COVID-19 cases. A set of simulations also aims to compare the effects of different NPIs on the most relevant outcomes of the epidemic, namely the accumulated number of infections and deaths. All this is performed by proposing a generalisation of the SEIR model that accommodates heterogeneity between age-groups and a Bayesian estimation framework. A time-varying contact-mobility matrix is also proposed as an input for the estimation framework. This work is organised as follows. Section 1 introduces the research problem and the overall research objective. Section 2.1 describes the data used, namely the COVID-19 daily epidemic data (Section 2.1.1), the age-group specific contact matrices (Section 2.1.2) and the NPIs data (Section 2.1.3). Section 2.2 presents the SIR model, its extension to accommodate the necessary period of virus incubation, as known as SEIR model (Section 2.2.1), and its generalisation to account for COVID-19 infection heterogeneity across different age groups due to age-dependent contact rates and mobility levels temporal paths (section 2.2.2). These age-dependent contact rates and mobility levels temporal paths are incorporated into the model's time-varying contact-mobility matrix obtained as described in Section 2.2.3. These section finishes with a full description of the Bayesian framework conceptual model and it estimation details (Section 2.2.4). Section 3 presents the main results and, lastly, Section 4 discusses the main conclusions as well as their implications in epidemiology policy-making. 2 Materials and methods 2.1 Data 2.1.1 Data on COVID-19 pandemic Epidemic data were collected from the Data Science for Social Good (DSSG (2022)) relative to Portugal. The dataset used contains the data released by the Direção-Geral de Saúde (Directorate-General of Health, DGS), the Portuguese health authority. Since the beginning of the pandemic, DGS releases daily data on its course. The key released variables used for modelling purposes as explained below are age-stratified (in decennial groups, 0–9, 10–19, 20–29, 30–39, 40–49, 50–59, 60–69 and 70 and more years old) number of confirmed cases number of deaths, and number of recovered cases (not age-stratified). The period of analysis ranges from March 3rd, 2020 to December 26th, 2020. It corresponds to a period where several NPIs were implemented (ranging from simple recommendations to law enforcement) and matches the period when vaccines were not yet available. The course of the epidemic over this period is shown in Fig. 1 .Fig. 1 Course of the COVID-19 epidemic in Portugal. (a) Daily infected cases. (b) Deaths (cumulative). Fig. 1 Fig. 1 illustrates the occurrence of two distinct waves. The first occurred between mid-March and the beginning of May 2020 and the second commenced in late August until late December 2020. Yet out of the current time frame, a third wave occurred in January 2021, leading to the highest spikes Portugal had experienced ever. 2.1.2 Age-group contact matrix Quantifying what entails a contact that is sufficient to transmit a disease can be extraordinary difficult, especially with a disease like COVID-19 that can be spread with even causal incidental contact with a infected person. Several comprehensive works have been made throughout the last fifteen years to quantify contacts (e.g. Mossong and Joël (2008), Prem et al. (2017); Arregui et al. (2018)). Here, due to nonexistence of infection data by location, we shall use the contact matrix for all locations estimated by Prem et al. (2017) for Portugal. As Chikina and Pegden (2020) points out, the matrices provided by Prem et al. (2017) are asymmetric, because his empirical work information collected regards contacts experienced with other that might not be part of the target population and matrices correspond to absolute frequencies of interactions between age groups, rather than frequencies relative to the sizes of the age groups. The age-structured model used here already account for the susceptible population which belongs to each age group, thus we processed the contact matrices as Chikina and Pegden (2020) describes:1. To correct for the second source of asymmetry by dividing each column by the proportion of the population in the corresponding age group; 2. To correct the first source of asymmetry by averaging out the pairs of elements reflected across the diagonal. The age-group contact matrix C is a G × G matrix representing the average daily number of contacts between each pair of age-groups, where G here denotes the number of age-groups considered. We denote the matrix entry [C ij], i, j = 1, …, G as the average daily number of contacts the people in the age-group i (matrix row index) have with people in age-group j (matrix column index). Prem et al. (2017) reports matrices for 16 quinquennial age-groups ([0,4], [5,9],…, [70,74], [75,∞]). As the COVID-19 related data is only available at decennial age-groups, prior to the asymmetry correction described above, were obtained by aggregating quiquennial age-groups to decennial ones ([0,9] = [0,4]+[5,9], [10,19] = [10,14]+[15,19],…, [70,…] = [70,74]+[75,∞]), taking in consideration their relative demographic weight. The used matrix is represented in Figure A4(a), where the cell figures represent the average number of daily contacts of people in the age-group i (row index, i = 1, 2, …, 8) with people in the age-group j (column index, j = 1, 2, …, 8), and the 5-tone cyan scale denotes their intensity. Notice that there is a strong tendency for people to prefer contacts with their peers. Also, there is evidence of the child/parent and grandchild/grandparent interactions. 2.1.3 Non-pharmaceutical interventions data Information on NPIs was mainly obtained from the Oxford COVID-19 Government Response Tracker (OxCGRT) (https://www.bsg.ox.ac.uk/research/research-projects/covid-19-government-response-tracker) which collects systematic information on policy measures that governments have taken to tackle COVID-19. The different policy responses have been tracked since 1 January 2020, covering more than 180 countries, coded into 21 indicators, including a miscellaneous notes field organised into five groups, containment and closure policies, economic policies, health system policies, vaccination policies and miscellaneous policies. These policies are recorded on a scale to reflect the extent and stringency of government action. For more detailed information please refer to Table A1 in Appendix. The selection of NPI indicators used in this work, along with their time extent and stringency, is represented in Fig. 2 .Fig. 2 Daily infections and Non-Pharmaceutical Interventions data in Portugal. Fig. 2 The stringency scale of NPIs represented in Fig. 2 differs across the eight categories (cf. Table A1 in Appendix). However its possible to group them into three groups. A first group of NPIs whose stringency scale ranges from 0 to 4, which is the case of Resctriction on gatherings, Facial coverings and International travelers controls; a second group whose scale ranges from 0 to 3, which the case of Stay at home requirements (lockdown), Workplace closing and School closing, and, finally, a group composed of Close public transport and Cancel public events whose scale ranges from 0 to 2. In any case the “0” denotes absence of the measures and the scale maximum denotes the most higher stringency. The colour scale used in Fig. 2 uses no colour (white) for scale value “0” and “the darker cyan tone” for the most stringest case. The intermediate stringency levels are represented by lighter cyan tones. Be aware that in Portugal some intermediate stringency levels of some measure do not exist, therefore aggregation was done for modelling purposes (cf. Table A1 in Appendix). Fig. 2 shows the NPI and the curve of daily cases in the analysis timeline. During the SARS-CoV-2 epidemic in Portugal four distinct periods can be distinguished. The first one, between the beginning of the epidemic and the declaration of a state of emergency, the first hard lockdown control measures (on March 18th, 2020). The second one, between March 18th, 2020, and the declaration of the situation of calamity when the lockdown measures started easing (May 4th, 2020). A third period from May 4th onwards and until August 18th, corresponding to the traditional Portuguese holidays period and the beginning of the school year and finally, from August 18th onwards. For modelling purposes, it is observable that the discriminating levels (cf. Table A1 in Appendix) of some NPIs might raise collinearity issues for model estimation. Section 2.2.4 explains the aggregation of some stringency levels and justifies the exclusion of NPIs from the analysis. 2.2 Methods 2.2.1 The SIR and SEIR models Much work has been done so far on the course of the COVID-19 outbreak using the so-called compartmental models (Abou-Ismail (2020); Ndaïrou et al. (2020); Wang et al. (2020)). The origin of compartmental models dates back to the early 20th century with the seminal work of Kermack and McKendrick (in 1927) (Kermack and McKendrick (1927)). Compartmental models simplify the mathematical modelling of infectious diseases. The population is assigned to compartments and may progress between compartments. Compartmental order usually shows flow patterns between the compartments; for example, SIS means susceptible, infectious, and then susceptible again (Hethcote (2000)). The models are most often run with ordinary differential equations (which are deterministic). Still, they can also be used with a stochastic framework (Hethcote (2000)). Compartmental models allow predicting how a disease spreads through the total number infected, the duration of an epidemic, and infection reproductive number (R 0), among other important epidemic course parameters. Moreover, such models could potentially be used as a framework to show how different public health interventions (e.g. vaccination, limited social contacts, lockdown) may affect the outcome of the epidemic. The basic compartmental model is the SIR model (Kermack and McKendrick (1927)). Nonetheless, it still captures the main properties of an epidemic (Anderson (1991); Kaye (1993)), and it has been widely used in epidemic modeling studies. The SIR model comprises three compartments: S for the number of susceptible, I for the number of infectious, and R for the number of removed (recovered, deceased, or immune) individuals at a particular time (Fig. 3 (a)). This model assumes incidence grows exponentially, which is not in agreement with the observed epidemic course, as the measures adopted by public health authorities at different moments, as well as the change in human behaviours tend to flatten the parabolic incidence curve. Therefore, its predictive value for infectious diseases that are transmitted from humans to humans, without any change in the constant transmission rate is of limited usefulness.Fig. 3 The SIR and SEIR models. (a) SIR model. (b) Generic transitions between compartments. Fig. 3 Fig. 4 The SEIRD model. Fig. 4 To represent that the number of susceptible, infectious and removed individuals may vary over time (even if the total population size remains constant) a time index, t, might be added: S(t), I(t) and R(t). For infections with the characteristics of COVID-19, there is a significant incubation period during which individuals have been infected but are not yet infectious themselves (European Center for Disease Prevention and Control (2020), Tang et al. (2020); Lauer et al. (2020); Zhou et al. (2020)). During this period the individual is in compartment E(t) (for exposed), resulting in the so-called SEIR model. Moreover, many authors (Godio et al. (2020); Wang et al. (2020); Mendes and Coelho (2021)) have been proposing generalisations of this basic SEIR model by splitting the compartment, R into two compartments, one to account for the infected people that genuinely recover from the disease, R, and a second one to account for the mortality induced by the infection, D. For a specific disease within a particular population, these functions may be worked out to predict possible outbreaks and bring them under control. In Fig. 3, Fig. 4 the boxes represent the compartments (or states) and arrows represent transitions from one compartment to another. S(t) is the number of people susceptible at time t, E(t) is the number of people exposed to the infection at time t (people that become infected but not infectious), I(t) is the number of infective people infected at time t, R(t) is the number of people recovered at time t and D(t) is the number of deceased at time t. Transitions from one compartment to another usually follow the framework represented in Fig. 3(b). The rate describes how long the transition takes, population is the group of individuals that this transition applies to, and probability is the probability of the transition taking place for an individual. The full SEIRD model is depicted in Fig. 4. The transitions from E to I and from I to R or D occur with the probability being 1 (everyone infected will certainly become infectious), 1 − μ′ (a share of 1 − μ′ infectious people will recover, usually expressed in “%“) and μ′ (a share of μ′ infectious people will die, usually expressed in “%“) and the population is E and I, respectively. Here δ is the rate at which infected people become infectious (days−1). The recovery rate is γ′ (expressed in days−1), μ′ is the fatality share (usually expressed in “%“), and α is the rate at which people die (expressed in days−1). In general, it is not possible to identify parameters α, μ′ and γ′, therefore a simplification is assumed hereafter, that is, γ = γ′ × (1 − μ′) (expressed in days−1) and μ = α × μ′ (expressed in days−1), where γ′ and μ′ represent the latent, non-identifiable, crude recovery and fatality share, respectively. The following differential equations describe transitions of the SEIRD model:(1) ∂S(t)∂t=−βS(t)I(t)N∂E(t)∂t=βS(t)I(t)N−δE(t)∂I(t)∂t=δE(t)−γI(t)−μI(t)∂R(t)∂t=γI(t)∂D(t)∂t=μI(t)N=S(t)+E(t)+I(t)+R(t)+D(t) The parameter β denotes the transmission rate, that is, the expected number of people an infected person infects per day and is the result of the contact rate c - the number of people an average person enters into contact with each day - and the probability that a contact provokes the transmission of the disease, ψ). Usually no data is available allowing estimation of ψ and c separately; hence, β is directly estimated. However, either ψ might change over time as a result, for example, of new virus variants or c is affected by changes in individual behaviour and NPIs put in place by the governments. Here we propose to replace c with age-structured contact matrices describing the rate of contact between each pair of ages. More specifically, we consider an age-structured model as used by (Ram and Schaposnik (2021)) in which we compute the age distribution of each compartment in decennial age-groups (0–9,10-19,…,60-60, 70+), incorporating age-group-dependent contact-mobility matrix describing the rate of contact between each pair of age-groups. In doing so, we aim to capture the heterogeneous effect of COVID-19 across different ages as well as the impact of the NPI both across different age-groups and age-specific daily environments. 2.2.2 The age-structured SEIRD model In the following the rationale underlying the age-structured SIR model is presented. As the permanent knowledge of the deaths generated by COVID-19 is of crucial importance to monitoring the pandemic's effects we have split the previous Removed compartment into Recovered and Death compartments, allowing us to monitor the recovering rate and the age-specific fatality rates. The system of ordinary differential equations in (1) is replaced by a system of age-stratified ordinary differential equations here:(2) ∂Si(t)∂t=−ψ(t)Si(t)∑j=1GMij(t)Ij(t)Nj∂Ei(t)∂t=ψ(t)Si(t)∑j=1GMij(t)Ij(t)Nj−δEi(t)∂Ii(t)∂t=δEi(t)−γiIi(t)−μiIi(t)∂Ri(t)∂t=γiIi(t)∂Di(t)∂t=μiIi(t)Ni=Si(t)+Ei(t)+Ii(t)+Ri(t)+Di(t)N=S(t)+E(t)+I(t)+R(t)+D(t)E(t)=∑i=1GEi(t),I(t)=∑i=1GIi(t),R(t)=∑i=1GRi(t),D(t)=∑i=1GDi(t). where i and j, i, j = 1, …, G are the age groups, ψ(t) is the probability of transmission given a contact, δ is the rate at which an infectious individual becomes infective, γ i is the rate of recovery, μ i is the age-group specific fatality rate, and N i is the population size of age group g. The parameter δ is the inverse of the average incubation period and governs the lag between having undergone an infectious contact and becoming infective. It is generally considered fixed because it depends mainly on the infectious agent's characteristics. Here we use the value 1/5.1 (Linton et al. (2020); Li et al. (2020)). The parameter γ i is the recovery rate for infected people. It corresponds to the inverse of the average time required for an active case to recovers. It provides precious information about how fast the people may recover from the disease (in days, in general). We believe that γ i is dependent on the individual health status which subsequently highly dependent on the individual age, but relatively constant in time. However, in this work, due to the lack of reliable empirical information on the daily recovered patients, we consider γ to be fixed across all age-groups and equal to 1/14 (1/days) (Tang et al. (2020)). The practitioner that needs to estimate this parameter might set a gamma prior, Ga(1, 1/14), that corresponds to a mean recovery period of 14 days and accounts for some uncertainty and variability in the patients recovery time. As discussed by Mendes and Coelho (2021), other formulations are possible to account for temporal variation as well. The parameter μ i is the fatality rate and provides information simultaneously on the proportion of infected individuals who, unfortunately, die and the time undergone infective as before death. It depends, indeed, on the patients’ resilience, the severity of the disease and the health system capability to treat people over time (e.g., with the introduction of a new therapy). However, due to the short period under analysis we do not consider μ i to be time-dependent. The probability of transmission of the disease given a contact ψ(t) depends mainly on the virus spreading ability which might be affected by self-protection measures. Whilst the former is highly related to the relative importance of the virus variants in circulation in a community, the latter is associated with the general level of use of facial covering devices. Here, due to the relatively short period in analysis which was not particularly affected by diverse virus variants, but reliable information on the use of facial coverings is available, we model ψ(t) as a step function distinguishing three distinct periods i.e., before any facial covering has even been recommended (cf. stringency “0” according to the OxCGRT scale), the period when they were required in some specified shared/public spaces outside the home with other people present, or some situations when social distancing not possible (cf. stringency “2” according to the OxCGRT scale) and the period when they were required in all shared/public spaces outside the home with other people present or all situations when social distancing not possible (cf. stringency “3” accoding to the OxCGRT scale). In the Bayesian framework proposed below all the parameters associated with these binary variables are considered stochastic with associated prior distributions. The contact-mobility matrix entries, M(t)=[Mij(t)]i,j=1G, defined in (5) are considered stochastic as described in Section 2.2.4. The simulation of the course of the epidemic assuming no NPIs were adopted is done assuming the contact-mobility matrix is constant over time, as described in Section 2.2.3, in (6). For modelling performed hereafter, one uses a discrete-time approximation to the stochastic continuous-time SEIRD model defined in (2). Consider a time interval (t, t + h), where h represents the length between the time points at which measurements are taken, here h = 1 day. Let dE i(t) denote the daily number of susceptible individuals who become infected, dI i(t) the daily number of infected individuals who become infectious, dR i(t) the daily number of infectious cases who recover, and dD i(t) the daily number of infectious cases who decease in age group i, i = 1, …, G, t = 1, …, T. Given initial conditions S i(0) = s i(0), E i(0) = e i(0), I i(0) = i i(0), R i(0) = 0 and D i(0) = 0, and the population size N, the discretised stochastic SEIRD model is specified by:(3) St+1g=Si(t)−dEi(t),g=1,…,G,t=0,…,T−1Et+1g=Ei(t)+dEi(t)−dIi(t),g=1,…,G,t=0,…,T−1It+1g=Ii(t)+dIi(t)−dRi(t)−dDi(t),g=1,…,G,t=0,…,T−1Rt+1g=Ri(t)+dRi(t),g=1,…,G,t=0,…,T−1Dt+1g=Di(t)+dDi(t),g=1,…,G,t=0,…,T−1dEi(t)=ψ(t)Si(t−1)∑j=1GMij(t)Ij(t−1)Nj,i=1,…,G,t=1,…,TdIi(t)=δEi(t−1),i=1,…,G,t=1,…,TdRi(t)=γIi(t−1),i=1,…,G,t=1,…,TdDi(t)=μiIi(t−1),i=1,…,G,t=1,…,TNg=Si(t)+Ei(t)+Ii(t)+Ri(t)+Di(t),g=1,…,G,t=1,…,TN=S(t)+E(t)+I(t)+R(t)+D(t),t=1,…,TE(t)=∑i=1GEi(t),I(t)=∑i=1GIi(t),R(t)=∑i=1GRi(t),D(t)=∑i=1GDi(t),t=1,…,T Again, to simulate the course of the epidemic, assuming no NPIs were adopted to mitigate the spread of the infection and flatten the incidence and prevalence curves we assume(4) dEi(t)=ψ(t)Si(t−1)∑j=1GMij′Ij(t−1)Nj,i=1,…,G,t=1,…,T where Mij′, is the time-invariant average daily number of contacts between each pair of age-groups as specified in (6). This set of equations, jointly with the initial conditions, define the age-structured SEIRD model used in this work. 2.2.3 Contact-mobility matrix The daily time-varying contact-mobility matrix, M(t) result from scaling the contact matrix of Section 2.1.2 by a mobility index, I(t), as follows:(5) M(t)=C⋅Idx(t), The purpose of Idx(t) is to measure the relative change in baseline contacts due to the implementation of several NPIs, incorporating in the baseline contact matrix C the adjustments induced by the NPIs whose role is to limit the contacts between individuals. The time-varying age-group contact-mobility matrix entries, M ij(t) are conditionally independent random variables as described in Section 2.2.4. It is generally accepted that flattening the infection incidence curve has been successfully achieved in most European states (e.g. Banholzer et al. (2021); Snoeijer et al. (2021); Suryanarayanan et al. (2021)). However, we do not know what would have happened if the NPIs had not been adopted and, in many cases, enforced by law. Therefore, we decided to perform a simulation where one assumes the contact matrix remains constant, that is,(6) M′=C. Here, it is assumed the daily contact rate among people prior to pandemic holds for the whole analysis period, leading to the simulation of what would have been the infection course if the number of contacts among people would mimic the population mixing prior to the pandemic, given the probability of infection caused by a single contact. The Google Mobility Reports (GMR) data (https://www.google.com/covid19/mobility/), were used to derive the mobility indexes, Idx(t). GMR provides daily data on several categories of places are changing in each geographic region. The used categories are: Retail & recreation, Grocery & pharmacy, Transit stations, Workplaces and Residential. The category Parks was not accounted for as the authors consider it has no influence on infections. These five categories originated other five indexes expressed in percentage change of the baseline (prior pandemic) contacts (that is, indexes are greater than 1, if contacts grow; 1, if contacts remain constant; and less than 1, if contacts decrease; see Fig. 5 ). When the lockdown policy was implemented, mobility in workplaces and other locations decreased sharply and increased smoothly again after the measures had been ease. On the contrary, mobility in residential areas shows the opposite course.Fig. 5 Google mobility. Mobility reported by Google by category and Mobility index used to obtain time-varying contact-mobility matrix. Fig. 5 These five indexes were collected in the T × 5 matrix of mobility, W. A singular value decomposition of the matrix W, which is admitted to be full-rank, was performed to obtain a single summary index as follows:(7) W=PEQ′, where P is a T × 5 matrix, E is a 5 × 5 diagonal matrix containing the singular values (square root of eigenvalues) of W′W and Q′ is a 5 × T matrix of eigenvectors of W′W. The resulting principals components scores are contained in the columns of the T × 5 matrix Z:(8) Z=PE=WQ. The first principal component (corresponding of first column of the scores matrix Z) accounts for 96.6% of the total variation of the data. Therefore, we used the temporal relative change of these scores as our summary mobility index, Idx(t), which are represented in Fig. 5. The total number of contacts in Portugal between March, 3rd and December26th 2020 (a total of 299 days), the whole period under analysis) can be estimated by:∑t=1T∑i=18∑j≥i8Mij(t)Ni. Assuming the situation prior to pandemic had not changed as a result of the set of taken NPIs, the total number of contacts in Portugal in the same period can be estimated by:300×∑i=18∑j≥i8CijNi. Therefore, as a result of the changes in mixing and social contacts among citizens, the expected number of contacts decreased from 201015.5 millions to 164418.2 millions. This decay (18.2%) in the number of contacts in the middle of March 2020 was responsible for avoiding millions of infections and thousands of deaths. The results of a simulation comparing these two extreme situations is reported in Section 3. 2.2.4 Bayesian hierarchical model Using the discrete-time approximation to the stochastic continuous-time SEIRD model (3) defined above we set up a Bayesian hierarchical model where the incidence variables are assumed stochastic (the time unit h is one day). Bayesian hierarchical models attempt to decompose the observed data into a series of conditional models, all linked together formally through basic probability relationships. In essence the strategy is based on the formulation of three primary statistical models or stages:● Stage 1: Data model: [data\|Process, ν 1], ● Stage 2: Process model: [Process\|ν 2], ● Stage 3: Prior distributions on parameters: [ν 1, ν 2], where the square brackets notation denotes probability distribution and ν 1 and ν 2 generically represent parameters introduced in the modelling. The following paragraphs follow this framework describing the (1) likelihood and associated distributional assumptions, (2) the process parameters’ link functions and (3) the prior distributions of the model parameters. The section closes with modelling considerations and decisions that have turned model estimation possible. Table 1 summarises the notation used in the Bayesian hierarchical model likelihood and distributional assumptions below.Table 1 Bayesian hierarchical likelihood notation. The data on recoveries are not age-grouped stratified. The time series for compartments Si, Ei, Ii, R, and Di are fully determined by applying SEIRD model equation (3) from given initial conditions, Si(0), Ei(0), Ii(0), R(0) and Di(0), i = 1, …, 8. Table 1Notation Description dEi(t) Observed counts of susceptible individuals who become infected in age-group i, i = 1, 2, …, 8, at time t, t = 1, …, T dIi(t) Observed counts of infected individuals who become infectious in age-group i, i = 1, 2, …, 8, at time t, t = 1, …, T dR(t) Observed counts of infectious individuals who recover, at time t, t = 1, …, T dDi(t) Observed counts of infectious individuals who perish in age-group i, i = 1, 2, …, 8, at time t, t = 1, …, T dE {dEi, i = 1, …, 8}, Vector of observed counts of susceptible individuals who become infected dI {dIi, i = 1, …, 8}, Vector of observed counts of infected individuals who become infectious dR {dR(t), t = 1, …, T}, Vector of observed counts of infectious individuals who recover dD {dDi, i = 1, …, 8}, Vector of observed counts of infectious individuals who perish Si(t) Observed state of susceptible compartment, in age-group i, i = 1, …, 8, at time t, t = 1, …, T Ei(t) Observed state of exposed compartment, in age-group i, i = 1, …, 8 at time t, t = 1, …, T Ii(t) Observed state of infectious compartment, in age-group i, i = 1, …, 8 at time t, t = 1, …, T I(t) ∑i=18Ii(t), Observed state of infectious compartment, at time t, t = 1, …, T M {S, E, I, R, D}, states of the SEIRD model compartments, where R = {R(t), t = 1, …T}, S = {Si(t), i = 1, …, 8; t = 1, …T}, and E, I, and D are defined similarly M(t) {S(t), E(t), I(t), R(t), D(t)}, Vector of states of the SEIRD model compartments at time t, t = 1, …, T Mij(t) Average number of contacts observed daily between people of age-group i and age-group j, i, j = 1, …, 8 at time t, t = 1, …, T Mi(t) Vector of average number of contacts observed daily between people of age-group i, i = 1, …, 8 at time t, t = 1, …, T M {M(t), t = 1, …, T}, Vector of contact-mobility matrices Ni Size of population in age-group i, i = 1, …, 8 N ∑i=18Ni(t), Size of population X Matrix of covariates (dummies) concerning NPIs C1, C2, C3, C4, C5, C6, C7, and C8 H Matrix of covariates (dummies) concerning NPIs H6, T × 3, where the first column is filled with 1's ν Vector of model parameters 2.2.4.1 Likelihood and distributional assumptions Given the three processes defined in the previous section, the likelihood is given as follows:(9) L=p(dE,dI,dR,dD,M\|M,ν)=p(dE\|dI,dR,dD,M,M,ν)×p(dI\|dR,dD,M,M,ν)×p(dR\|dD,M,M,ν)×p(dD\|M,M,ν)×p(M\|M,ν) where p(u, v) represents the joint probability density of random vectors u, v and p(u\|v) represent the conditional density of u, given the vector v, M represents the SEIRD model and the states of its correspondent compartment (as defined in Table 1) and ν is vector of all model parameters. To formalise the likelihood we need first to make some assumptions, some of them for computational simplicity, others inspired by the data generation process. The transitions of individuals from one compartment to the next of in the SEIRD model are considered stochastic movements between the corresponding population compartments. In each period, an individual either stays in or moves on to the next compartment. In reliability analysis, life-time is usually considered to follow an exponential distribution. By analogy, here the time length an individual spends in a compartment is exponentially distributed with some compartment-specific rate λ(t). Therefore, the probability of extending the stay by a further period of length h is exp(−λ(t)h) and the probability of leaving is therefore 1 − exp(−λ(t)h). The summation over the individual Bernoulli trials assuming they are independent and identical for all compartment members, would result in binomial distributions (Diekmann et al. (1990)). Due to the scale we are working with, we found it useful to take advantage of the approximation of the binomial to the Poisson distribution.Assumption 1 Given (3), we assume that dE conditional on M(t−1), M i(t) and ν is independent in time and among age-groups: (10) p(dE\|dI,dR,dD,M,M,ν)=p(dE\|M(t−1),Mi(t),ν)=∏i=18∏t=1Tp(dEi(t)\|Si(t−1),I(t−1),Mi(t),ν). Assumption 2 Given (3), we assume that dI conditional on M(t−1) and ν is independent in time and among age-groups: (11) lllp(dI\|dR,dD,M,M,ν)=p(dI\|M(t−1),ν)=∏i=18∏t=1Tp(dIi(t)\|Ei(t−1),ν). Assumption 3 Given (3), we assume that dR conditional on M(t−1) and ν is independent in time: (12) p(dR\|dD,M,M,ν)=p(dR\|M(t−1),ν)=∏t=1Tp(dR(t)\|I(t−1),ν). Assumption 4 Given (3), we assume that dD conditional on M(t−1) and ν is independent in time and among age-groups: (13) p(dD\|M,M,ν)=p(dD\|M(t−1),ν)=∏i=18∏t=1Tp(dDi(t)\|Ii(t−1),ν). Assumption 5 The incidences dE i(t), dI i(t), dR i(t), and dD i(t) are considered conditionally independent Poisson random variables, that is: (14) p(dEi(t)\|Si(t−1),Mi(t),I(t−1),ν)∼Pois(Si(t−1)×πdEi(t)),i=1,…,8,t=1,…,Tp(dIi(t)\|Ei(t−1),ν)∼Pois(Ei(t−1)×πdIi(t)),i=1,…,8,t=1,…,Tp(dR(t)\|I(t−1),ν)∼Pois(I(t−1)×πdR),i=1,…,8,t=1,…,Tp(dDi(t)\|Ii(t−1),ν)∼Pois(Ii(t−1)×πdDi(t)),i=1,…,8,t=1,…,T, where transition probabilities are given by:(15) πdEi(t)=1−exp−ψ(t)∑j=1GMij(t)Ij(t−1)Nj,i,j=1,…,8,t=1,…,T,πdIi=1−exp−δ,i=1,…,8,t=1,…,TπdR=1−exp−γ,i=1,…,8,t=1,…,TπdDi(t)=1−exp−μi(t)i=1,…,8,t=1,…,T. The model further assumes the population size is N=∑i=18Ni, remains constant, and individuals mix homogeneously.Assumption 6 The time-varying age-group contact-mobility matrix entries, M ij(t), are considered stochastic, conditionally independent, given a set of covariates that impact, directly or indirectly, in the contacts as depicted by the stringency of the NPIs of Table 2 . The model matrix X is a matrix containing dummy variables for each level of stringency as described in Table 2. Due to collinearity issues, some of the stringency levels on Table 2 result from aggregation of the raw ones as explained below.Table 2 Non-pharmaceutical interventions and corresponding levels considered in this work. Table 2Name Coding (C1) School closing 0 - no measures 1 - Recommend closing or all schools open with alterations resulting in significant differences compared to non-Covid-19 operations or require closing (only some levels or categories, e.g. just high schools, or just public schools) or require closing all levels (C11) (C2) Workplace closing 0 - no measures 1 - require closing (or work from home) for some sectors or categories of workers (C21) 2 - require closing (or work from home) for all-but-essential workplaces (e.g. grocery stores, doctors) (C22) (C5) Close public transport 0 - no measures 1 - recommend closing (or significantly reducing volume/route/means of transport available) (C51) (C6) Stay at home requirements 0 - no measures 1 - recommend not leaving house (C61) 2 - require not leaving house with exceptions for daily exercise, grocery shopping, and ‘essential’ trips (C62) (C7) Restrictions on internal movement 0 - no measures 1 - recommend not to travel between regions/cities or internal movement restrictions in place (C71) (H6) Facial coverings 0 - No policy 1 - Required in some specified shared/public spaces outside the home with other people present or in some situations when social distancing is not possible (H61) 2 - Required in all shared/public spaces outside the home with other people present or all situations when social distancing is not possible (H62) The three levels of stringency of C1(t), t = 1, …, T, were collapsed into two levels as we think the distinction between recommendation and requirement is useless given that the public education system in Portugal (representing 98% of the sector) complied quickly with all the interventions and even anticipated them. The four levels of stringency of C4(t), t = 1, …, T, were collapsed in a single one as we think the distinction based on the size of the gatherings is useless for our purposes. Due to Spearman's perfect correlation between C3(t) and C4(t) and highly collinearity of these and C8(t) with C1(t), we removed variables C3(t), C4(t), and C8(t) from the analysis, letting C1(t) as being their proxy. This prevents the Markov chain chain simulation process from mixing poorly being unable to distinguish the effects of those variables on the contact-mobility levels. Therefore, the model matrix is defined as: X = (C1 1, C2 1, C2 2, C5 1, C6 1, C6 2, C7 1), where C1 1 = {C11(t), t = 1, …, T} is the vector of T observations of the dummy variable representing the stringency level 1 of C1(t) and the remaining variables are defined in a similar manner. Moreover, the time-varying age-group contact-mobility matrix entries are considered conditionally independent and follow a gamma model as follows:(16) p(Mij(t)\|X)∼Gaθi,θiξij(t),i,j=1,…,8,t=1,…,T where θ i is the shape parameter and θgξig(t) is the rate parameter, implying E(M ij(t)\|X) = ξ ij(t). The above assumptions simplify the likelihood as follows:(17) L=p(dE,dI,dR,dD,M)=p(dE\|dI,dR,dD,M)p(dI\|dR,dD,M)p(dR\|dD,M)p(dD\|M)p(M)=∏t=1TpdR(t)\|I(t−1)∏i=18pdEi(t)\|Si(t−1),I(t−1),Mi(t)×pdIi(t)\|Ei(t−1)pdDi(t)\|Ii(t−1)∏j=18pMij(t) 2.2.4.2 Link functions In order to account for the effects of some NPIs such as lockdown and social distancing, we assume that the transmission rate is affected by: (1) the time-varying age-group contact-mobility matrix M(t) or the constant age-group specific contact-mobility matrix M′ (obtained as explained in Section 2.2.3) and (2) the probability of virus transmission ψ(t), which is assumed to be a step function where the intervention moments correspond to the entry in force of an important self-protection measure not accounted for by mobility indexes, that is, the use of facial coverings. It is modelleded through two dummy variables H61(t) and H62(t):(18) H61(t)=1,H6(t)=10,otherwise,t=1,…,T, (19) H62(t)=1,H6(t)=20,otherwise,t=1,…,T, which correspond to the periods when facial coverings were required in some specified shared/public spaces outside the home with other people present, or some situations when social distancing was not possible (cf. Table 2) and when they were required in all shared/public spaces outside the home with other people present or all situations when social distancing was not possible (cf. Table 2). This implies ψ(t) to be formulated as:(20) logit(ψ(t))=H′(t)η=η0+η1H61(t)+η2H62(t)=η0t<τ1η0+η1,τ1≤t<τ2η0+η2,t≥τ2,t=1,…,T, where logit(•) is the log(p/(1 − p)) transformation, η′ = (η 0, η 1, η 2), where η 0 is the baseline probability that a contact provokes the transmission of the disease, η 1 and η 2 are the (expected) change in this probability due to the introduction of two different stringency levels of wearing a facial covering and h′(t) is the tth row of the T × 3 matrix H, composed of a column of 1's and the two T-length column vectors representing the dummy variables describing the stringency of wearing facial coverings (H6(t)), H = (1, H6 1, H6 2). This formulation allows for assessing the effect of the introduction of facial coverings on the risk of getting infected. The link function for ξ ij(t) in (16) is defined by:(21) logξij(t)=λij0+x′(t)λ,i,j=1,…,8;t=1,…,T, where λij0 models the expected log-level value of the baseline contact-mobility matrix M ij(1) (that is, C ij), and λ=(λC11,λC21,λC22,λC51,λC61,λC62,λC71) are the regression coefficients multiplying x′(t), t = 1, …, T, the rows of X (as defined before). The expected value of the gamma distribution above is defined to assess NPI effects on the average number of daily contacts responsible for infection spreading. The mean of the posterior distribution of exp(λij0), t = 1, is expected to fit the entries of the contact matrix C=[C]i,j=18, and the posterior mean of exp(X′λ) is meant to fit the mobility index Idx(t), t = 1, …, T defined in Section 2.2.3. A comment on the previous model matrix is necessary. The authors think the facial coverings interventions (H6) may be better accounted for in the model of probability of virus transmission, ψ, as it might might be seen as an reduction on the baseline probability of infection. As a matter of fact, several other authors have used this framework to assess the potential effect of using masks (e.g., Eikenberry et al. (2020) and Leech et al. (2022)). Evoking the aforementioned reasons (cf. Section 2.2.2), the parameters μ i(t), i = 1, …, 8, which govern the passage from the compartment I to the compartment D, are considered age-group dependent. For the first four age groups, [0,9], [10,19], [20,29] and [30,39] deaths are rare; therefore the fatality rate is considered time-invariant:(22) μi(t)=μi0,i=1,2,3,4,t=1,…,T, For the last four age-groups, [40,49], [50,59], [60,69] and 70+ the parameters μ i(t) are considered time-dependent following an exponentially decreasing trend, with an increasing rate constant across the four age-groups:(23) μi(t)=μi0exp(−ρμt),i=5,6,7,8,t=1,…,T, where the parameter μi0, i = 5, 6, 7, 8, represents the initial value of the fatality rate. As discussed about the SEIRD model represented in Fig. 4, the fatality rates, μ (⋅)(t) in (22), (23) represent the product of two distinct quantities: (1) the probability of moving from compartment I to compartment D and (2) the inverse of the period an infected person takes to perish. It is measured in days −1. The mortality rate is supposed to decrease over time at a rate of ρ μ (expressed in days). 2.2.4.3 Prior distributions The following independent prior distributions are assigned to each of the previous parameters:(24) η∼N3(0,Pη),λij0∼N(0,0.0001),i,j=1,…,8,λ∼N7(0,Pλ),θi∼Ga(0.0001,0.0001),i=1,…8μi∼Ga(0.0001,0.0001),i=1,…8,ρμ∼Ga(0.0001,0.0001). where Ga(a, b) refers to a gamma distribution with shape a and rate b, mean a/b, and variance a/b 2, N(0, 0.0001) to a normal distribution with a mean of zero and a precision of 0.0001, and N(•)(0, P (•)) to a (•)-variate multivariate normal distribution with a mean vector of zeros and precision matrix P (•). The precision matrix P • follows a Wishart distribution with a diagonal scale matrix with diagonal elements equal to 1000 and dim(P (•)) + 2 degrees of freedom to ensure a non-informative hyperprior. Given the hierarchical representation presented above and using the same notation as in (17), one can evaluate the posterior distribution of all of the processes and parameters, given the observed data:(25) Π(ν\|dE,dI,dR,dD,M,X,H)∝p(dE,dI,dR,dD,M\|ν,X,H)p(ν)=p(dE,dI,dR,dD,M\|ν,X,H)p(η)∏i=18∏j=18p(λij0)p(λ)p(θ)p(μ)p(ρμ)=∏t=1TpdR(t)\|I(t−1)∏i=18pdEi(t)\|Si(t−1),I(t−1),Mi(t)×pdIi(t)\|Ei(t−1)pdDi(t)\|Ii(t−1)∏j=18pMij(t)×p(η)∏i=18∏j=18p(λij0)p(λ)p(θ)p(μ)p(ρμ) 2.2.4.4 Other considerations The epidemic model specified in (3), (4), (14), (15), the fatality rate models (22) and (23) with the transmission probability model (20), and the time-varying age-group contact-mobility matrix model (16) and (21) has parameter vector ν=(η,λ1,10,λ1,20,…,λ8,80,λ,θ,μ,ρμ), which we would like to estimate from the knowledge of the initial conditions S i(0), E i(0), I i(0), R i(0) = 0, and D i(0) = 0, i = 1, …, 8, the population sizes, N i, i, 1, …, 8, and the data of {dE, dI, dR, dD, M, X, H}. The initial conditions I i(0), i = 1, …, 8 are unknown. Indeed, the number of confirmed cases in the very beginning of the epidemic is far from the true prevalence of COVID-19 infections. However, for simplicity, one assumes it to be known and equals the number of confirmed infected cases on March 3rd, 2020, in each age-group. To determine the initial conditions S i(0), E i(0), i = 1, …, 8, one needs to know at least one of them. One decided to estimate the initial condition E i(0), i = 1, …, 8 from the series of exposed, dE i(t), whose paths were based on the time between exposure and infectiousness, that is, 5.1 days, on average (Mendes and Coelho (2021)). A gamma distribution with a μ X/σ X parameter of 5.1/4.5, corresponding to a mean and median incubation period of 5.1 and 4.2 days was used to backcast the values of dI i(t) to the previous compartment and provide the initial values E i(0), i = 1, …, 8. The initial conditions S i(0), i = 1, …, 8 were then obtained by difference to the total population of the respective age-group, N i, i = 1, …, 8. This procedure was only used to estimate the initial conditions S i(0) and E i(0). Indeed, the observed processes dE i(t), i = 1, …, 8, t = 1, …T are not observed and thus considered missing, requiring numeric integration over the support of the probability distribution of dE i(t) in (25). One cannot evaluate the posterior distribution (25) analytically and must resort to numeric simulation methods. We use the special case of MCMC known as Gibbs sampling (Gilks et al. (1996)) and implemented the algorithm using the R package JAGS (all code used in this paper can be obtained from the authors upon request). Additionally to the estimation of model parameters ν, we decided to perform four additional simulations to assess the effect of the absence of all/some NPIs on infections. The first considers the case where no NPIs are considered, except facial coverings (denoted in Section 3 as Facial coverings(+), other NPIs(−)). The second accounts for all NPIs effects are considered but facial coverings (denoted in Section 3 as Facial coverings(−), other NPIs(+)). The third accounts for the case where no NPIs effects are considered (denoted in Section 3 as Facial coverings(−), other NPIs(−)). This simulation in based on a constant transmission probability ψ(t) = η 0 and a constant contact-mobility matrix, that is M′ = C. Finally, the fourth simulation accounts only for the effects of Facial coverings and Workplace closing (denoted in Section 3 as Facial coverings(+), Workplace closing(+)). 3 Results After an adaptation phase of 5000 iterations and a burn-in period of 2000 iterations by which we consider convergence has been achieved. A sample of size of 300, resulting from running four independent chains with a thin step of 50 iterations (to avoid serial correlation) was used to obtain marginal posterior distributions for all model parameters. For the vast majority of the parameters the Gelman-Rubin statistic (Brooks and Gelman (1998)) achieved values below 1.05 which reinforces the belief convergence was achieved (cf. Table A2 in Appendix). Analysing the model's fit against real data is relevant as a tool of external validation. Therefore, the model fit is assessed by plotting the available data against the estimated expected values of the posterior distribution. The model's fit is evident from Figures A1 and A2, in the Appendix, regarding the age-group epidemic data, and Fig. 6, Fig. 7, Fig. 8 for accumulated and daily number of infected cases, accumulated deaths and recoveries, regarding all ages, for the period corresponding to the used data. Indeed, the observed number of accumulated infected cases, ∑ t dI(t), is 391.7 thousands and the fitted value, ∑ tE[dI(t)], is 384.1 thousands, representing an underestimation of about 1.9%; on the other hand, the observed number of accumulated deaths for the period under analysis, D(t) on 26t h December, 2020, is 5907 and the fitted value, ∑ tE[dD(t)], is 5911, representing an overestimation of only 0.07%.Fig. 6 Model fitting results on infections. (a) Observed daily accumulated cases of infection (red) and mean of predictive distribution of ∑tdI(t), ∑tE[dI(t)] (blue). (b) Observed daily confirmed cases of infection (red) and mean of predictive distribution of dI(t), E[dI(t)] (blue). Fig. 6 Fig. 7 Model fitting results on deaths. (a) Observed daily accumulated deaths (red) and mean of predictive distribution of ∑tdD(t), ∑tE[dD(t)] (blue). (b) Observed daily daily deaths (red) and mean of predictive distribution of dD(t), E[dD(t)] (blue). Fig. 7 Fig. 8 Model fitting results on recoveries. (a) Observed daily accumulated recoveries (red) and mean of predictive distribution of ∑tdR(t), ∑tE[dR(t)] (blue). (b) Observed daily daily recoveries (red) and mean predictive distribution of dR(t), E[dR(t)] (blue). Fig. 8 The contact-mobility matrix, M(t), and its baseline source C and mobility index Idx(t) (see Section 2.2.3), are targets of the proposed model. Therefore, Figures A3 and A4 summarise model's fitting results regarding the mobility index Idx(t), and the contact matrix C. Figure A3 shows the mobility index, Idx(t), t = 1, …, T and the mean of posterior distribution of exp(Xλ) (see equation (21)). It fairly follows the mobility index temporal path. Finally, Figure A4 shows, side by side, the target matrix C and the mean of the marginal posterior distribution of exp(λij0), i, j = 1, …, 8. Some small differences are noticed but the 5-tone colour pattern is similar in both matrices. Moreover, the root mean square deviation between the entries of C and their fitted values is equal to 0.06 daily contacts. These results suggest that jointly considering a time-varying probability of infection, ψ(t), and a time-varying contact-mobility matrix is able to capture the age-group and the overall course of the epidemic. It also confirms something else: mobility data, represented here through the Google Mobility data, mimics the reduction in daily contacts among individuals which inevitably led to mitigating the natural spread of the disease. Fig. 8(a) represents the overall number of recovery people paths (as mentioned previously age-group recovery data are unavailable). As described in Section 2.2.4, we considered the recovery rate, γ, as a fixed parameter. The purpose was to mimic the recovery path using a generally literature-accepted recovery rate (corresponding to 14 days of infectiousness) and compare it with the non-reliable figures released by national heath authorities. As Fig. 8 shows, the estimated path of recoveries shows some departure from the observed one, confirming the expert perception of the delay of releasing figures about infected people off the isolation period followed a conservative criterion. Besides, the spikes in the observed path denote the moments when DGS released data on recoveries accumulated for some weeks. Table 3 presents the posterior means and posterior standard deviations of model parameters. The results regarding the infection parameters equation (20) are all statistically significant at 5%. The posterior mean of η 1 is −0.513 (se = 0.010). Therefore, the risk of a contact provokes the transmission of the disease suffers a reduction of 40.2% (1 − exp(η 1) × 100), when facial coverings become required in some specified shared/public spaces outside the home with other people present, or some situations when social distancing not possible. Moreover, when the stringency of this NPI increased, that is, when facial covering become required in all shared/public spaces outside the home with other people present or all situations when social distancing not possible the risk fall down even sharply by 59.0%, as the posterior mean of η 2 is −0.892.Table 3 Posterior means, standard deviations and p-values of the estimated Bayesian hierarchical SEIRD model. Table 3Parameter Posterior mean (std) p-value η0 −6.004 (0.009) 0.00 η1 −0.513 (0.01) 0.00 η2 −0.892 (0.009) 0.00 μ1 5 × 10−6 (5 × 10−6) 0.18 μ2 3 × 10−6 (3 × 10−6) 0.16 μ3 4 × 10−6 (2 × 10−6) 0.04 μ4 9 × 10−6 (4 × 10−6) 0.00 μ5 5.1 × 10−5 (1.2 × 10−5) 0.00 μ6 1.36 × 10−4 (2.1 × 10−5) 0.00 μ7 5.49 × 10−4 (4.9 × 10−5) 0.00 μ8 0.015754 (5.57 × 10−4) 0.00 ρμ 0.00289 (1.55 × 10−4) 0.00 θ1 425.428 (9.461) 0.00 θ2 426.769 (9.302) 0.00 θ3 425.855 (11.818) 0.00 θ4 425.122 (12.647) 0.00 θ5 425.438 (14.446) 0.00 θ6 426.489 (17.431) 0.00 θ7 424.998 (24.956) 0.00 θ8 25.7 (59.4) 0.33 The results regarding the age-group fatality rates, μ i, i = 1, …, 8 are also reported. These parameters are expressed in days −1 units. They are not significantly different from zero (at 5% significance level) for the first two age-groups, [0,9] and [10,19] years of age. For the age-groups [20,29] and [30,39] fatality rates are statistically significant but still not different from zero in practical terms. For older ages, the results, being statistically significant, are not practically different from zero. In the period under analysis Portugal undergone a pandemic landscape dominated by the variant alpha of SARS-CoV-2 which was not one of deadliest among all SARS-CoV-2 variants identified so far. Moreover, the fatality rate μ here represents the product of the probability of an infectious individual die (α) by the rate (in days) it takes to perish (μ′) (see Section 2.2.1). The estimated value of μ 8, given by the mean of its marginal posterior distribution is μ^8= 0.0157538. If we assume that only a tiny fraction of the infectious individuals go through a serious respiratory condition leading eventually to death, usually reported between 1% and 4% (the value of α), that it would correspond to a average of between 0.5 and 2.5 days to perish after becoming infectious, that is, hosting a sufficient viral load capable of spreading out to other individuals. Table 4 shows the expected impact (in per cent) of the combination NPI/level considered in this work on the average contact-mobility matrix entries. The values were computed transforming the posterior means of the log link function} (21), (exp(λ (•)) − 1) × 100. The statistical significance of the parameter (at 5% significance level) is indicated in the table by the superscript “a”.Table 4 Results on the link functions of the contact-mobility matrix gamma model. Statistically significant parameters at 5% level are indicated by superscript “a”. Results are indicated in ±% variation on mobility from baseline ((exp(λ(•)) − 1) × 100%) where “•” represents the combination NPI/level. Table 4NPI Level % School closing (C1) Recommend closing or all schools open with alterations resulting in significant differences compared to non-Covid-19 operations or require closing (only some levels or categories, e.g. just high school, or just public schools) or require closing all levels (C11) −0.59 Workplace closing (C2) Require closing (or work from home) for some sectors or categories of workers (C21) −14.19a Require closing (or work from home) for all-but-essential workplaces (e.g. grocery, stores, doctors) (C22) −35.57a Close public transport (C5) Recommend closing (or significantly reduce volume/route/means of transport available) (C51) 2.16a Stay at home requirements (C6) Recommend not leaving house (C61) −8.08a Require not leaving house with exceptions for daily exercise, grocery shopping, and ‘essential’ trips (C62) −1.68a Restrictions on internal movement (C7) Recommend not to travel between regions/cities or internal movement restrictions in place (C71) −2.45a The results show that, holding the other NPIs fixed, on average, closing (or working from home) for some sectors or categories of workers (C21, starting on March 12th, 2020) had a statistically significant effect on reducing prior pandemic contact-mobility levels by 14.19%. However, when this NPI became more strict (closing (or working from home) for all-but-essential workplaces (e.g. grocery stores, doctors) (C22, starting on March 19th, 2020), the prior pandemic contact-mobility levels dropped, on average by 35.57%. The result on recommend not leaving house (C61, stating on March 15th, 2020) shows it had more impact in reducing the prior pandemic contact-mobility levels (−8.08%), than requiring not leaving the house with exceptions for daily exercise, grocery shopping, and ‘essential’ trips (C62), which started four days later, on March 19th, 2020 (−1.68%). A behaviour of self protection may explain this result, as at the beginning of the pandemic, a generalised feeling of fear spread out in the Portuguese population. Much debate is going on about the effect of School closing (C11, starting on March 9th, 2020). Results presented in Table 4 on school closing do not indicate a high potential for reduction of prior pandemic contact-mobility levels (−0.59%). This NPI mainly aimed at preventing students to infect their families after infectious contacts at school environment and the other way around. Its lack of effect is likely due to its nature that did not have direct effects on the generalised contact-mobility levels. Table 4 also reports the estimated impact of two other combinations NPI/level. First, The result regarding recommend closing public transport (or significantly reduce volume/route/means of transport available) (C51, starting on March 19th, 2020) shows a statistically significant impact on raising the prior pandemic contact-mobility levels (2.16%). This result is not surprising as we know that limitations on the offer of public transportation may, on one hand prevent close contacts among passengers but, on the other hand, may imply the citizens to move around using other transportation means and thus raising the contact-mobility levels as they are measured in this work. Second, the result regarding recommend not to travel between regions/cities or internal movement restrictions in place (C71, starting on March 19th, 2020) exhibits a marginal, but still statistically significant, impact on reducing the prior pandemic contact-mobility levels by only 2.45%. Three other NPIs that were not considered for model estimation Cancel public events, (C3), Restrictions on gatherings (C4), and International travel controls (C8) as explained in Section 2.2.4. In Portugal, these measures were in place for a extended period of time (cf. Fig. 2) and were almost coincident (collinear) with school closing. Results about their proxy, C1, suggest they had a small or even null impact on changing the prior pandemic contact-mobility levels. The results of the four simulations performed are presented in Table 5 and Figures A5, A6 and A7. The results suggest a very much different situation on December 26th, 2020 regarding the accumulated number of infections, the daily infections and deaths under different NPI scenarios.Table 5 Results of simulations on December26th 2020, considering the absence of all or some NPIs. Table 5Measure Accumulated infections Accumulated deaths Daily maximum infections Observed 391.7 thous. 5907 thous. 9.2 thous. Simulation 1 (Facial coverings(+), other NPIs(−)) 8.2 mill. 88.5 thous. 107.8 thous. Simulation 2 (Facial coverings(−), other NPIs(−)) 9.7 mill. 147.3 thous. 225.0 thous. Simulation 3 (Facial coverings(−), other NPIs(+)) 9.4 mill. 122.7 thous. 181.8 thous. Simulation 4 (Facial coverings(+), Workplace closing(+)) 7.1 mill. 59.8 thous. 75.7 thous. In simulation 1 we consider a situation where measures are absent, except for facial coverings (one denotes it as Facial coverings(+), other NPIs(−)). This simulation shows the number of accumulated infections by December 26th would reach 8.2 million, instead of the 391.7 thousand actually verified. Similarly, this simulation shows the accumulated number of deaths would reach 88.5 thousand instead of the 5907 observed indeed. On the other hand, the maximum of daily number of infection cases would reach 107.8 thousands, a much larger number than the recorded daily maximum on December9th 2020 (9.2 thousand). Simulation 2 represents an even more extreme scenario (one denote it as Facial coverings(−), other NPIs(−)). Indeed, in the absence of any NPI, including neither the mandatory use of facial coverings in some or all places, the number of accumulated infections by December 26th would reach 9.7 million, the accumulated deaths would reach 147.3 thousand, and the maximum of daily infection would reach 225.0 thousand. Simulation 3 represents an intermediate situation somewhere between simulations 1 and 2 (one denotes it as Facial coverings(−), other NPIs(+)). Indeed, in the absence of facial covering, but assuming all other NPIs were taken, the number of accumulated infections by December 26th would reach 9.4 million, the accumulated deaths would reach 122.7 thousand, and the maximum of daily infections would reach 181.8 thousand. Simulation 3 shows that the joint adoption of a set of NPI whose intended role was to prevent people from closely contacting each other, has a more extensive impact than the universal use of facial coverings. Yet, the use of facial masks together with the additional NPIs has an additional and not negligible effect, which is easily concluded from the difference between the results of simulation 3 and the actual observed situation. These results jointly with the results of the impact of the NPIs above suggest the joint effect of Facial coverings and Workplace closing is major among all the NPI combinations. Simulation 4 (one denotes it as Facial coverings(−), Workplace closing(+)) considers the situation where only the effects of facial covers on the probability of transmission and the two Workplace closing stringency levels on the contact-mobility levels are accounted for. It is clear from the results reported in Table 5 and illustrated in Figures A5, A6 and A7 that even adopting only these two NPIs together would have resulted in posteponing the peak of the incidence (see Figure A6), and in a number of infections hardly managed by the National Health System. Indeed, by December 26th the number of accumulated infections, the accumulated deaths, and and the maximum of daily infections would have reached 7.1 million, 59.8 thousand, and 75.7 thousand, respectively. It is even possible to support the statement that the sooner enforcement of wearing face mask (the use of facial masks became required only on October 28th, 2020) would have been positive both for incidence of infections and deaths, but not sufficient to avoid an unmanageable burden in the health system. 4 Discussion and future work In this work, the main classes of public health measures that aim to prevent and/or control infection transmission in the community were characterised and presented under a common umbrella named Non-Pharmaceutical Interventions (NPIs). In particular, those put into force in Portugal were considered. Ultimately, the scientific community and government officials need proof of their effectiveness in reducing the number of contacts between individuals, specially between the most exposed to infection and those who present particular vulnerabilities, and therefore mitigate and contain the spread of the infection. This proof is even more critical when there was no effective and safe vaccine to protect those at risk of severe COVID-19, NPIs were seen as the most effective measures against COVID-19. As the social and economic costs may vary significantly between NPI measures, it is of utmost importance to offer some insight into the ones that may be more effective in preventing the spread of the disease. The reduction of individual contacts is hard to quantify, but proxy variables exist that aid in the investigation of NPI effects, having the mobility information widely available on the Internet as an example of a potential proxy. A SEIRD model was fitted to the COVID-19 epidemic data of Portugal to achieve this primary objective. The model was estimated through a Bayesian hierarchical framework and a set of singular model features used to assess the role of mobility data as a potential proxy of the reduction of individual contacts. Furthermore, once the critical role of mobility data was established, several other issues were investigated, namely the quantification of the impact of those NPIs on the course of the epidemic curves. The methodology presented in this work might be even extended for other purposes such as simulation of the effect of specific interventions. The age-structured SEIRD model framework considered the generation of daily infections to be a function of the virus-inherent probability of transmission and the contact rate - the number of people an average person entered into contact with on a daily basis. We considered the contact rates to be heterogeneous across age-groups. A matrix of average daily prior pandemic contacts per age-group is used. It is supposed to evolve according to a corresponding mobility index computed using Google mobility data. A full contact-mobility matrix resulted and is used to mimic the temporal path of mixing pattern. The contact-mobility matrix entries are considered stochastic and assumed to vary according to the impact of the NPIs. Assuming a constant virus-inherent probability of transmission, which is not a hard assumption given the short period of analysis (the period prior to the vaccination phase which corresponds to the first 300 days of the pandemic in Portugal), the model fit is considered fair. Indeed, the visual fit is from Figures A1, A2, in the Appendix, regarding the age-group data, and Fig. 6, Fig. 7, Fig. 8 for all ages, depicting the observed and fitted number of accumulated and daily number of infected cases, accumulated deaths and recoveries, for the period corresponding to the used data. Indeed, the fitted number of accumulated infected cases underestimates the observed value by 1.9% and the fitted number of accumulated deaths for the same period overestimates the observed value by 0.07%. Given the unreliable nature of the data on daily recoveries, the fitted model considers the curve of recoveries a function of a constant and widely accepted recovery rate. Under this circumstance one would have necessarily expected a departure between the observed and estimated curves of recoveries. It was the case, which suggests a conservative approach by the public health authorities in the criteria to consider an infected individual not infectious any more. Regarding the contact-mobility matrix, another target of the proposed model, the fit fair as well. Indeed, on one hand the baseline prior pandemic contact matrix is accurately reproduced (Figure A4), and on the other, the main festures of its temporal path, represented by the summary mobility index, are fairly fitted (Figures A3). The age-group fatality rates, μ i,i = 1, …, 8, are not significantly different from zero (at 5% significance level) for the first two age-groups, [0,9] and [10,19] years of age. For the remaining age-groups, fatality rates are statistically significant, but still not different from zero in practical terms. The last age group is where fatality concentrate the most. The baseline probability that a contact provokes the transmission of the disease is, under the model, highly low, around 0.25%. However, the risk of a contact provoking the transmission of the disease suffers a reduction of 40.2%, when facial coverings became required in some specified shared/public spaces outside the home with other people present, or some situations when social distancing was not possible. Moreover, when the stringency of this NPI increased, that is, when facial covering became required in all shared/public spaces outside the home with other people present or all situations when social distancing was not possible the risk dropped even sharper by 59.0%. Although, these two effects are statistically significant, they had a moderate impact in the course of the pandemic. In fact, the comparison between simulations shows the impact of using facial covering is much lower than the effect of all other NPIs in lowering contacts between individuals. Among the other NPIs, the crucial contributions to the reduction of the prior pandemic contact levels and therefore to the spread of infection must be emphasised, were by the following order of importance Workplace closing, Stay at home requirements (lockdown) and Restrictions on internal movement. These results are, in general, in line with what has been published in the literature (e.g. Mendez-Brito et al. (2021)). The lockdown intervention impact on the number of infections cannot be measured directly. However, results confirm the growth of the contact-mobility mixing in residential areas, certainly reducing the mixing in places where the infection would have spread out easily. Works exist reporting the importance of School closing as a mean of reducing the spread of the infection (cf. Section 1). These results are fully understandable, given the mixing of students within schools and their families. However, as we do not have data on infections by source of infection it turns hard to assess this effect. Moreover, the results reported here do not support this thesis. In short, the results obtained here are fourfold. Firstly, thanks to the universal wearing of facial coverings, the probability that a contact provoked the transmission of the disease was reduced by more than 50%. Secondly, the impact of the other NPI (from which Workplace closing stands clearly out) is so significant that otherwise Portugal would have gone into a non-sustainable situation of having 80% of its population infected in the first 300 days of the pandemic. This situation would have led to a number of deaths more than twenty times higher than the number that was actually recorded by December 26th, 2020. Thirdly, the requirement of wearing face masks and working-from-home (Workplace closing) whenever possible had a impact on reducing the probability of transmission and preventing physical contacts among the population that would not have been sufficient for Portugal to avoid an unmanageable number of infection. Therefore, the it seems clear the the economic and social effects of NPIs adopted could not have been avoided. And lastly, the observed curve of recovery cases in Portugal suggests health authorities followed a conservative approach on the criteria to consider an infected individual not infectious any longer. This work is clearly limited by the unavailability of data on infections by source of infection and mobility data stratified by age-group. This lack of data prevented the authors to go deeper in the consideration of different contact matrices according to the place where infections occurred and forced them to use of a mobility index uniform across all age-groups. Consideration of these data would have turned possible modelling accurately some sources of heterogeneity. Moreover, the modelling approach does not consider the impact in a more extended period as the vaccination effect was not considered. Including the effect of the vaccination in the burden caused to the health system should be a priority in the near future. These implies extending the analysis to the year of 2021, where the hardest wave of infections hit the country, requiring (1) the consideration of other model compartments, as suggested by Mendes and Coelho (2021), and (2) the incorporation of effects of different SARS-Cov-2 variants on the probability of transmission and morbidity. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. A Appendix Table A1 Non-pharmaceutical interventions considered in this work. More information can be found here https://www.bsg.ox.ac.uk/research/research-projects/covid-19-government-response-tracker and here https://www.bsg.ox.ac.uk/research/publications/variation-government-responses-covid-19 Table A1Name Coding (C1) School closing 0 - no measures 1 - recommend closing or all schools open with alterations resulting in significant differences compared to non-Covid-19 operations 2 - require closing (only some levels or categories, e.g. just high school, or just public schools) 3 - require closing all levels Levels 1, 2 and 3 aggregated together (C2) Workplace closing 0 - no measures 1 - recommend closing (or recommend work from home) or all businesses open with alterations resulting in significant differences compared to non-Covid-19 operation 2 - require closing (or work from home) for some sectors or categories of workers 3 - require closing (or work from home) for all-but-essential workplaces (e.g. grocery stores, doctors) Level 1 does not apply to Portugal (C3) Cancel public events 0 - no measures 1 - recommend cancelling 2 - require cancelling Level 1 does not apply to Portugal (C4) Restrictions on gatherings 0 - no restrictions 1 - restrictions on very large gatherings (the limit is above 1000 people) 2 - restrictions on gatherings between 101-1000 people 3 - restrictions on gatherings between 11-100 people 4 - restrictions on gatherings of 10 people or less Level 1 does not apply to Portugal. Levels 2, 3 and 4 aggregated together. (C5) Close public transport 0 - no measures 1 - recommend closing (or significantly reducing volume/route/means of transport available) 2 - require closing (or prohibit most citizens from using it) Level 2 does not apply to Portugal. (C6) Stay at home requirements 0 - no measures 1 - recommend not leaving house 2 - require not leaving house with exceptions for daily exercise, grocery shopping, and ‘essential’ trips 3 - require not leaving house with minimal exceptions (e.g. allowed to leave once a week or only one person can leave at a time, etc) Level 3 does not apply to Portugal. (C7) Restrictions on internal movement 0 - no measures 1 - recommend not to travel between regions/cities 2 - internal movement restrictions in place Levels 1 and 2 aggregated together. (C8) International travel controls 0 - no restrictions 1 - screening arrivals 2 - quarantine arrivals from some or all regions 3 - ban arrivals from some regions 4 - ban on all regions or total border closure Levels 1, 2 and 4 do not apply to Portugal (H6) Facial coverings 0 - No policy 1 - Recommended 2 - Required in some specified shared/public spaces outside the home with other people present or some situations when social distancing is not possible 3 - Required in all shared/public spaces outside the home with other people present or all situations when social distancing is not possible 4 - Required outside the home at all times regardless of location or presence of other people Levels 1 and 4 do not apply to Portugal Table A2 Gelman and Rubin's statistic Table A2Parameter R Parameter R μ1 1.03 λ2,60 1.05 μ2 1.00 λ2,70 1.05 μ3 1.02 λ2,80 1.04 μ4 1.00 λ3,30 1.01 μ5 1.00 λ3,40 1.06 μ6 1.00 λ3,50 1.06 μ7 1.00 λ3,60 1.05 μ8 1.00 λ3,70 1.06 ρμ 1.00 λ3,80 1.04 η0 1.02 λ4,40 1.00 η1 1.01 λ4,50 1.03 η2 1.02 λ4,60 1.07 θ1 0.99 λ4,70 1.03 θ2 1.00 λ4,80 1.05 θ3 1.01 λ5,50 1.01 θ4 1.01 λ5,60 1.06 θ5 1.00 λ5,70 1.06 θ6 1.00 λ5,80 1.04 θ7 1.01 λ6,60 1.00 θ8 1.54 λ6,70 1.04 λ1,10 1.00 λ6,80 1.04 λ1,20 1.06 λ7,70 1.00 λ1,30 1.03 λ7,80 1.05 λ1,40 1.03 λ8,80 1.06 λ1,50 1.05 λC11 1.06 λ1,60 1.06 λC21 1.06 λ1,70 1.02 λC22 1.08 λ1,80 1.03 λC41 1.08 λ2,20 1.01 λC51 1.00 λ2,30 1.03 λC61 1.00 λ2,40 1.05 λC71 1.00 λ2,50 1.09 Fig. A1 Model fitting - observed and fitted values. Observed daily confirmed cases of infection and means of posterior distributions of dIi(t), i = 1, …, 8. Due to intra week variation observed values were smoothed using a gaussian kernel smoother with bandwidth 14 days. Fig. A1 Fig. A2 Model fitting - observed and fitted values. Observed daily accumulated deaths and means of posterior distribution of Di(t), i = 1, …, 8. Fig. A2 Fig. A3 Model fitting results on the mobility matrix index, Idx(t). Lines represent the mean of the posterior distribution of exp(λX), in blue, (where X is the full matrix as in equation (21)), the mean of the posterior distribution of exp(λC21C21(t)+λC22C22(t)), in green (corresponding to simulation 4 where only workplace closing and facial coverings are considered) and Idx(t) as defined in Section 2.2.3, in red. Fig. A3 Fig. A4 (a) Age-group baseline prior pandemic contact matrix C (Section 2.1.2) used in this work. Numbers represent the daily average number of contacts of people in the row age-group with people in the column age-group, and the 5-tone blue scale denotes their intensity. All ages are in years. (b) Results on the fitted contact matrix. The cell values correspond to mean of posterior distribution of exp(λij0), (see equation (21)). Fig. A4 Fig. A5 Results of simulations on observed cumulative incidence and simulated values (mean of the predictive distribution). Fig. A5 Fig. A6 Results of simulations on observed daily incidence and simulated values (mean of the predictive distribution). Fig. A6 Fig. A7 Results of simulations on observed accumulated deaths and simulated values (mean of the predictive distribution). Fig. A7 Peer review under responsibility of KeAi Communications Co., Ltd. ==== Refs References Abou-Ismail A. Compartmental models of the COVID-19 pandemic for physicians and physician-scientists SN Comprehensive Clinical Medicine 2 2020 10.1007/s42399-020-00330-z Anderson R.M. Discussion: The Kermack-McKendrick epidemic threshold theorem Bulletin of Mathematical Biology 53 1991 10.1007/BF02464422 Arregui S. Alberto A. Sanz J. Moreno Y. Projecting social contact matrices to different demographic structures PLoS Computational Biology 14 2018 10.1371/journal.pcbi.1006638 Banholzer N. Lison A. Ozcelik D. Stadler T. Feuerriegel S. Werner V. The methodologies to assess the effectiveness of non-pharmaceutical interventions during COVID-19: A systematic review European Journal of Epidemiology 37 10 2022 1003 1024 10.1007/s10654-022-00908-y 36152133 Banholzer N. Van Weenen E. Lison A. Alberto C. Seeliger A. Kratzwald B. Tschernutter D. Puig Salles J. Bottrighi P. Lehtinen S. Feuerriegel S. Werner V. Estimating the effects of non-pharmaceutical interventions on the number of new infections with COVID-19 during the first epidemic wave PLoS One 16 2021 10.1371/journal.pone.0252827 Barbarossa M.V. Fuhrmann J. Compliance with NPIs and possible deleterious effects on mitigation of an epidemic outbreak Infectious Disease Modelling 6 2021 859 874 10.1016/j.idm.2021.06.001 34308001 Brauner J.M. Mindermann S. Sharma M. Johnston D. Salvatier J. Gavenčiak T. Stephenson A.B. Leech G. Altman G. Mikulik V. John Norman A. Monrad J.T. Besiroglu T. Ge H. Hartwick M.A. Teh Y.W. Chindelevitch L. Gal Y. Kulveit J. Inferring the effectiveness of government interventions against COVID-19 Science 371 2021 10.1126/science.abd9338 Brooks S.P. Gelman A. General methods for monitoring convergence of iterative simulations)? Journal of Computational & Graphical Statistics 7 1998 10.1080/10618600.1998.10474787 Chan J.F.W. Yuan S. Kok K.H. Wang To K.K. Chu H. Jin Y. Xing F. Liu J. Yip C.C.Y. Poon R.W.S. Tsoi H.W. Lo S.K.F. Chan K.H. Poon V.K.M. Chan W.M. Ip J.D. Cai J.P. Cheng V.C.C. Chen H. Ming Hui C.K. Kwok Y.Y. A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: A study of a family cluster The Lancet 395 2020 10.1016/S0140-6736(20)30154-9 Chen J. Pathogenicity and transmissibility of 2019-nCoV—a quick overview and comparison with other emerging viruses Microbes and Infection 22 2020 10.1016/j.micinf.2020.01.004 Chikina M. Pegden W. Modeling strict age-targeted mitigation strategies for COVID-19 15 Braunstein L.A. Plos one 2020 e0236237 10.1371/journal.pone.0236237 Diekmann O. Heesterbeek J.A.P. Metz J.A.J. On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations Journal of Mathematical Biology 28 1990 10.1007/BF00178324 Disease Prevention, European Center for, and Control Outbreak of novel coronavirus disease 2019 (COVID-19): increased transmission globally – fifth update Rapid Risk Assessment 2019 2020 Disease Prevention, European Centre for, and Control Guidelines for non-pharmaceutical interventions to reduce the impact of COVID-19 in the EU/EEA and the UK 2020 Downing S.T. Mccarty R.J. Guastello A.D. Cooke D.L. Mcnamara J.P.H. Assessing the predictors of adaptive and maladaptive covid-19 preventive behaviours: An application of protection motivation theory Psychology Health & Medicine 2022 10.1080/13548506.2022.2093925 DSSG Covid19pt-Data 2022 https://Raw.githubusercontent.com/Dssg-Pt/Covid19pt-Data/Master/Data.csv Eikenberry S.E. Marina M. Enahoro I. Tin P. Keenan E. Yang K. Eric K. Abba B.G. To mask or not to mask: Modeling the potential for face mask use by the general public to curtail the COVID-19 pandemic Infectious Disease Modelling 5 2020 10.1016/j.idm.2020.04.001 Ferguson N.M. Daniel L. Gemma N.-G. Natsuko I. Ainslie K. Marc B. Sangeeta B. Adhiratha B. Zulma C. Gina C.-D. Amy D. Ilaria D. Han F. Katy G. Will G. Arran H. Wes H. Okell L.C. van Elsland Sabine …Azra C G. Impact of non-pharmaceutical interventions (NPIs) to reduce COVID-19 mortality and healthcare demand 2020 10.25561/77482 Imperial College COVID-19 Response Team Gilks R. Richardson S. Spiegelhalter D.J. Markov chain monte Carlo in practice: Interdisciplinary statistics 1996 Godio A. Pace F. Vergnano A. Seir modeling of the Italian epidemic of sars-cov-2 using computational swarm intelligence International Journal of Environmental Research and Public Health 17 2020 10.3390/ijerph17103535 Gozzi N. Bajardi P. Nicola P. The importance of non-pharmaceutical interventions during the COVID-19 vaccine rollout PLoS Computational Biology 17 2021 10.1371/journal.pcbi.1009346 Haug N. Geyrhofer L. Londei A. Dervic E. Desvars-Larrive A. Loreto V. Pinior B. Thurner S. Klimek P. Ranking the effectiveness of worldwide COVID-19 government interventions Nature Human Behaviour 4 2020 10.1038/s41562-020-01009-0 Hengartner M.P. Waller G. von Wyl A. Factors related to non-compliance with non-pharmaceutical interventions to mitigate the spread of SARS-CoV-2: Results from a survey in the Swiss general adult population Frontiers in Public Health 10 2022 10.3389/fpubh.2022.828584 Hethcote H.W. Mathematics of infectious diseases SIAM Review 42 2000 10.1137/S0036144500371907 Kaye P.M. Infectious diseases of humans: Dynamics and control. Roy m. Anderson and robert m. May 1993 Oxford University Press 10.1016/0167-5699(93)90204-X 1992. £22.50 (Viii + 757 Pages) ISBN 0 19 854040 x.” Immunology Today 14 Kermack W.O. McKendrick A.G. A contribution to the mathematical theory of epidemics Series A, containing papers of a mathematical and physical character Vol. 115 1927 Proceedings of the Royal Society of London 10.1098/rspa.1927.0118 Kretzschmar M.E. Ben Ashby E.F. Overton C.E. Panovska-Griffiths J. Pellis L. Quaife M. Rozhnova G. Scarabel F. Stage H.B. Swallow B. Thompson R.N. Tildesley M.J. Villela D. Challenges for modelling interventions for future pandemics Epidemics 38 2022 10.1016/j.epidem.2022.100546 Lauer S.A. Grantz K.H. Bi Q. Jones F.K. Zheng Q. Meredith H.R. Azman A.S. Reich N.G. Lessler J. The incubation period of coronavirus disease 2019 (CoVID-19) from publicly reported confirmed cases: Estimation and application Annals of Internal Medicine 172 2020 10.7326/M20-0504 Leech G. Rogers-Smith C. Teperowski Monrad J. Sandbrink J.B. Snodin B. Zinkov R. Rader B. Brownstein J.S. Gal Y. Bhatt S. Sharma M. Mindermann S. Brauner J.M. Aitchison L. Mask wearing in community settings reduces SARS-CoV-2 transmission Proceedings of the National Academy of Sciences 119 23 2022 e2119266119 10.1073/pnas.2119266119 Li Q. Guan X. Wu P. Wang X. Zhou L. Tong Y. Ren R. Kathy S. Leung M. Eric H. Lau Y. Wong J.Y. Xing X. Xiang N. Wu Y. Li C. Chen Q. Li D. Tian L. …Feng Z. Early transmission dynamics in wuhan, China, of novel coronavirus–infected pneumonia New England Journal of Medicine 382 2020 10.1056/nejmoa2001316 Linton N.M. Kobayashi T. Yang Y. Hayashi K. Akhmetzhanov A.R. Jung S.-mok Yuan B. Kinoshita R. Nishiura H. Epidemiological characteristics of novel coronavirus infection: A statistical analysis of publicly available case data 2020 10.1101/2020.01.26.20018754 medRxiv Mendes J.M. Coelho P.S. Addressing hospitalisations with non-error-free data by generalised SEIR modelling of COVID-19 pandemic Scientific Reports 11 2021 10.1038/s41598-021-98975-w Mendez-Brito A. El Bcheraoui C. Pozo-Martin F. Systematic review of empirical studies comparing the effectiveness of non-pharmaceutical interventions against COVID-19 Journal of Infection 83 2021 Moher D. Liberati A. Tetzlaff J. Altman D.G. Altman D. Antes G. Atkins D. Barbour V. Barrowman N. Berlin J.A. Clark J. Clarke M. Cook D. D'Amico R. Deeks J.J. Devereaux P.J. Dickersin K. Egger M. Ernst E. …Tugwell P. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement PLoS Medicine 6 2009 10.1371/journal.pmed.1000097 Mossong N.A.N.D.J. Joël A.N.D.H. Social contacts and mixing patterns relevant to the spread of infectious diseases PLoS Medicine 5 3 2008 1 10.1371/journal.pmed.0050074 Ndaïrou F. Area I. Bader G. Nieto J.J. Torres D.F.M. Corrigendum to ‘mathematical modeling of COVID-19 transmission dynamics with a case study of wuhan’ [chaos solitons 2020 Fractals 135 2020 109846 Chaos, Solitons & Fractals (2020) 135(109846), (10.1016/j.chaos.2020.109846)).” Chaos, Solitons and Fractals 141 Nunan D. Brassey J. What is the evidence for mass gatherings during global pandemics? The Centre for Evidence Based Medicine 2020 Prem K. Cook A.R. Jit M. Projecting social contact matrices in 152 countries using contact surveys and demographic data PLoS Computational Biology 13 2017 10.1371/journal.pcbi.1005697 Ram V. Schaposnik L.P. A modified age-structured SIR model for COVID-19 type viruses Scientific Reports 11 2021 10.1038/s41598-021-94609-3 Rosa Giuseppina La Bonadonna L. Lucentini L. Kenmoe S. Suffredini E. Coronavirus in water environments: Occurrence, persistence and concentration methods - a scoping review Water Research 179 2020 10.1016/j.watres.2020.115899 Santos J.V. Joana Gomes da Costa Costa E. Almeida S. Joana Cima Pita-Barros P. Factors associated with non-pharmaceutical interventions compliance during COVID-19 pandemic: A Portuguese cross-sectional survey Journal of Public Health 2022 10.1093/pubmed/fdac001 Seale H. Dyer C.E.F. Abdi I. Rahman K.M. Sun Y. Qureshi M.O. Dowell-Day A. Sward J. Saiful Islam M. Improving the impact of non-pharmaceutical interventions during COVID-19: Examining the factors that influence engagement and the impact on individuals BMC Infectious Diseases 20 1 2020 10.1186/s12879-020-05340-9 Shittu E. Adewumi F. Ene N. Keluo-Udeke S.C. Wonodi C. Examining psychosocial factors and community mitigation practices to limit the spread of COVID-19: Evidence from Nigeria Healthcare 10 3 2022 10.3390/healthcare10030585 Snoeijer B.T. Burger M. Sun S. Dobson R.J.B. Folarin A.A. Measuring the effect of non-pharmaceutical interventions (NPIs) on mobility during the COVID-19 pandemic using global mobility data Npj Digital Medicine 4 2021 10.1038/s41746-021-00451-2 Sopory P. Novak J.M. Noyes J.P. Quarantine acceptance and adherence: Qualitative evidence synthesis and conceptual framework Journal Of Public Health-Heidelberg 30 9 2022 2091 2101 10.1007/s10389-021-01544-8 Suryanarayanan P. Tsou C.H. Poddar A. Mahajan D. Dandala B. Madan P. Agrawal A. Wachira C. Samuel O.M. Bar-Shira O. Kipchirchir C. Okwako S. Ogallo W. Otieno F. Nyota T. Matu F. Barros V.R. Shats D. Kagan O. …Rosen-Zvi M. AI-assisted tracking of worldwide non-pharmaceutical interventions for COVID-19 Scientific Data 8 2021 10.1038/s41597-021-00878-y Tang B. Bragazzi N.L. Qian L. Tang S. Xiao Y. Wu J. An updated estimation of the risk of transmission of the novel coronavirus (2019-nCov) Infectious Disease Modelling 5 2020 10.1016/j.idm.2020.02.001 Wang K. Lu Z. Wang X. Li H. Li H. Lin D. Cai Y. Xing F. Song Y. Feng Z. Ji W. Wang X. Yin Y. Wang L. Peng Z. Current trends and future prediction of novel coronavirus disease (COVID-19) epidemic in China: A dynamical modeling analysis Mathematical Biosciences and Engineering 17 2020 10.3934/MBE.2020173 Yang L. Morgenstern C. Kelly J. Lowe R. Munday J. Villabona-Arenas C.J. Gibbs H. Pearson C.A.B. Prem K. Leclerc Q.J. Meakin S.R. Edmunds W.J. Jarvis C.I. Gimma A. Funk S. Quaife M. Russell T.W. Emory J.C. Abbott S. …Jit M. The impact of non-pharmaceutical interventions on SARS-CoV-2 transmission across 130 countries and territories BMC Medicine 19 2021 10.1186/s12916-020-01872-8 Zhou F. Ting Y. Du R. Fan G. Liu Y. Liu Z. Xiang J. Wang Y. Song B. Gu X. Guan L. Yuan W. Li H. Wu X. Xu J. Tu S. Zhang Y. Chen H. Cao B. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in wuhan, China: A retrospective cohort study The Lancet 395 2020 10.1016/S0140-6736(20)30566-3 Zhu N. Zhang D. Wang W. Li X. Yang B. Song J. Zhao X. Huang B. Shi W. Lu R. Niu P. Zhan F. Ma X. Wang D. Xu W. Wu G. Gao G.F. Tan W. A novel coronavirus from patients with pneumonia in China, 2019 New England Journal of Medicine 382 2020 10.1056/nejmoa2001017
PMC010xxxxxx/PMC10287546.txt	==== Front Lancet Reg Health Eur Lancet Reg Health Eur The Lancet Regional Health - Europe 2666-7762 The Author(s). Published by Elsevier Ltd. S2666-7762(23)00090-X 10.1016/j.lanepe.2023.100671 100671 Articles Post COVID-19 condition, work ability and occupational changes in a population-based cohort Kerksieck Philipp ac Ballouz Tala ac Haile Sarah R. a Schumacher Celine a Lacy Joanne a Domenghino Anja ab Fehr Jan S. a Bauer Georg F. a Dressel Holger a Puhan Milo A. a∗ Menges Dominik a a Epidemiology, Biostatistics and Prevention Institute (EBPI), University of Zurich (UZH), Hirschengraben 84, Zurich 8001, Switzerland b Department of Visceral and Transplantation Surgery, University Hospital Zurich (USZ), Rämistrasse 100, Zurich 8091, Switzerland ∗ Corresponding author. Epidemiology, Biostatistics and Prevention Institute (EBPI), University of Zurich (UZH), Hirschengraben 84, Zurich CH-8001, Switzerland. c These authors share first authorship based on equal contribution. 23 6 2023 23 6 2023 10067127 4 2023 5 6 2023 7 6 2023 © 2023 The Author(s) 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Evidence on the impact of post COVID-19 condition (PCC) on work ability is limited but critical due to its high prevalence among working-age individuals. This study aimed to evaluate the association between PCC, work ability, and occupational changes in a population-based cohort. Methods We used data from working-age adults included in a prospective, longitudinal cohort of a random sample of all individuals infected with SARS-CoV-2 between August 2020 and January 2021 in the Canton of Zurich, Switzerland. We evaluated current work ability, work ability related to physical and mental demands, and estimated future work ability in 2 years (assessed using Work Ability Index), and PCC-related occupational changes one year after infection. Findings Of 672 individuals included in this study, 120 (17.9%) were categorised as having PCC (defined as presence of self-reported COVID-19 related symptoms) at 12 months. There was very strong evidence that current work ability scores were mean 0.62 (95% CI 0.30–0.95) points lower among those with PCC compared to those without in adjusted regression analyses. Similarly, there was very strong evidence for lower odds of reporting higher work ability with respect to physical (adjusted odds ratio (aOR) 0.30, 95% CI 0.20–0.46) and mental (aOR 0.40, 0.27–0.62) demands in individuals with PCC. Higher age and history of psychiatric diagnosis were associated with more substantial reductions in current work ability. 5.8% of those with PCC reported direct effects of PCC on their occupational situation, with 1.6% of those with PCC completely dropping out of the workforce. Interpretation These findings highlight the need for providing support and interdisciplinary interventions to individuals affected by PCC to help them maintain or regain their work ability and productivity. Funding Federal Office of Public Health, 10.13039/501100006447 Department of Health of the Canton of Zurich , University of Zurich Foundation, Switzerland; 10.13039/100018693 Horizon Europe . Keywords COVID-19 SARS-CoV-2 Post COVID-19 condition Long covid Work Work ability Occupation Cohort Observational study ==== Body pmc Research in context Evidence before this study We searched PubMed for all articles evaluating the association of post COVID-19 condition (PCC) and work ability and occupational changes, indexed up to 19 April 2023 with no language or time restrictions. The search string ‘(“post covid-19” or “long covid” or “pasc” or “post-acute sequelae”) and (“work ability” or “occupation∗” or “return to work”)’ was used. We screened 311 articles, of which 16 were eligible. Six additional articles were identified through bibliographic searches of relevant articles. Evidence on the association of PCC and work-related outcomes primarily stems from studies that focused on return to work and occupational changes in highly selective populations, such as health care workers, individuals hospitalised for COVID-19, or patients with PCC recruited through post COVID-19 clinics or social media. There was high variability in reported estimates, relating to differences in underlying populations and timepoints of follow-up. Return to work outcomes are highly dependent on systemic and organisational factors, limiting the generalisability of the existing literature. Since perceived work ability is an important determinant of return to work and occupational performance, the Work Ability Index (WAI) provides a validated measure of work-related functioning that is more independent of certain systemic or organisational factors. However, it was only assessed in three studies reporting evidence of reduced work ability in patients with PCC. None assessed risk factors associated with reduced work ability. Added value of this study To our knowledge, this is the first study that examined work ability (based on the WAI) and occupational changes and their association with PCC in a population-based cohort of working-age individuals infected with SARS-CoV-2. We found that PCC was strongly associated with a reduction in work ability at one year after infection with a more substantial reduction among older individuals and those with a history of psychiatric diagnosis. We also found that approximately one in fifteen individuals with self-reported COVID-19 related symptoms had occupational changes attributed to PCC within one year, with one to two in 100 completely dropping out of the workforce. This proportion is relevantly lower than reported by others, likely due to non-population-based sampling in other studies, but still constitutes a significant burden on economic and healthcare systems on a global scale. Implications of all the available evidence Together with existing evidence, our study suggests that PCC has significant consequences on the workforce. This may have severe implications for affected individuals, employers and the economy. These findings highlight the need to provide timely support and interdisciplinary public health interventions that can support affected individuals in regaining and retaining their work ability. Introduction Post COVID-19 condition (PCC) affects 10–20% of individuals infected with SARS-CoV-2.1, 2, 3, 4, 5, 6, 7, 8, 9, 10 Symptoms associated with PCC are varied and can be physical (such as fatigue, pain, and dyspnoea) or mental (such as memory and concentration difficulties), with a fluctuating course and frequently reported post-exertional exacerbation.1 , 2 , 11 , 12 Many of these symptoms adversely impact individuals’ everyday functioning, including impairments to their ability to engage in physical activities and participate in social life and work.1 , 2 , 13 The prevalence of PCC is highest among those of working age10 , 11 and the resulting socioeconomic implications are likely considerable.14 , 15 While it is important to develop effective management strategies and interventions to reduce the health burden of PCC, it is thus critical to also consider its impact on the workforce and establish sensible pathways to restore occupational participation in those severely affected. Several studies have evaluated the association of PCC with work-related functioning or subsequent occupational changes.16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37 Existing studies were primarily conducted in highly selective populations and mostly focused on describing work absenteeism, showing that 11% up to about half of workers with PCC do not return to work several months after COVID-19.2 , 13 Various individual, organisational, and systemic aspects (e.g., supportive return-to-work policies) contribute to successful return to work after an illness, including having sufficient actual work ability.38, 39, 40 Work ability is a multifactorial measure frequently used in clinical practice and research to assess the degree to which an individual is physically and mentally able to cope with demands at work.41, 42, 43 In addition to short- and long-term sickness absence,44, 45, 46, 47, 48 poor work ability is also associated with early retirement49 , 50 and disability at work,50 , 51 all of which carry large repercussions for the labour market and economy. Rehabilitation programs targeted at the working-age population generally aim to improve or preserve work ability. Given the substantial prevalence of PCC and its potential for long-term work-related consequences, understanding the association of PCC with work ability is crucial for the development of policies and multidisciplinary strategies aimed at supporting affected individuals in their recovery. In this study, we aimed to comprehensively evaluate the association between PCC, work ability, and occupational changes in a working-age population within a prospective population-based cohort of SARS-CoV-2 infected individuals. Methods Study design and participants We used data from a prospective, population-based, observational cohort of individuals with diagnosed SARS-CoV-2 infection from the Canton of Zurich, Switzerland (Zurich SARS-CoV-2 Cohort; ISRCTN14990068).5 , 52 Based on mandatory reporting of all SARS-CoV-2 infections to the Department of Health of the Canton of Zurich, we prospectively invited an age-stratified (18–39 years, 40–64 years, ≥65 years), daily, random sample of eligible individuals diagnosed between 06 August 2020 and 19 January 2021 for study participation. Eligibility criteria were being 18 years or older, able to follow study procedures, residing in the Canton of Zurich, and having sufficient knowledge of the German language. All participants were enrolled upon or shortly after diagnosis, infected with wildtype SARS-CoV-2, and unvaccinated at time of infection (recruitment took place prior to vaccine rollout in Switzerland). In this study, we included individuals of working age (18–64 years old; retirement age is 65 years in Switzerland) who did not report being retired at enrolment. To ensure that evaluated outcomes were not related to reinfection with SARS-CoV-2 over the course of follow-up, we excluded individuals reporting a reinfection event. The study was approved by the ethics committee of the Canton of Zurich (BASEC-Nr. 2020-01739) and we obtained written or electronic consent from all participants. Data sources We collected data using electronic questionnaires. At baseline immediately after enrolment, we collected data on the acute primary infection (i.e., symptoms, severity), pre-existing comorbidities (any of hypertension, diabetes, cardiovascular disease, chronic respiratory disease, chronic kidney disease, malignancy, or immune suppression), pre-infection health status, and socio-demographic characteristics. In this ongoing cohort, we collect follow-up data on participants' health trajectories in regular intervals after infection.5 , 52 At the intermediary follow-up time point of 12 months, we additionally elicited measures of work ability and asked participants to report any occupational changes over the first 12 months post-infection. Simultaneously, we asked participants to report any pre-existing psychiatric diagnoses before infection and any new or worsened psychiatric diagnoses during follow-up, as this emerged as an important aspect with respect to PCC over the course of the study.53 Participants were also asked to provide further details in free text fields. One researcher (DM) additionally conducted personal phone interviews with participants for whom questionnaire information was not unequivocal (n = 4). Outcome measurement We assessed self-perceived work ability using selected measures from the Work Ability Index, a validated and frequently used instrument for assessing work ability.41 , 42 , 44 The primary outcome was the current work ability scale (score from 0 to 10, 10 being best ability and 0 no ability to work). In sensitivity analyses, we categorised current work ability into poor (scores ≤6), moderate (scores 7–8), and excellent (scores ≥9).54 Secondary outcomes included items evaluating work ability related to physical and mental demands (5-point Likert scale) and estimated future work ability in 2 years (3-point Likert scale), as well as occupational changes attributed to PCC by participants during follow-up. Occupational changes attributed to PCC were determined based on a pre-defined list of such potential changes and further information obtained from participants through comments in free text fields or phone interviews (i.e., where no further information on the specific occupational change was provided by the participant or where the relation to SARS-CoV-2 infection was unclear). Further information on evaluated outcomes related to work ability and occupational changes is presented in Supplementary Table S1. We defined PCC using two different measures to allow better comparability with the heterogenous reporting in other studies and since previous research has shown that the use of multiple definitions allows better evaluation of its impact on affected individuals.5 , 55 First, we defined the presence of PCC as participants reporting any COVID-19 related symptom at 12 months of follow-up (self-reported COVID-19 related symptoms). This self-reported measure combined the information of two questions, eliciting whether participants experienced any out of a list of 23 symptoms commonly reported to be related to PCC and whether participants deemed these symptoms to be related to COVID-19 (i.e., not to other causes such as pre-existing, chronic, or incident conditions). This definition was as closely aligned with the World Health Organization definition56 as it was possible in our study. Second, we used a combined measure of whether participants had fully recovered and how they assessed their current health status at 12 months ((non-)recovery and health impairment). This self-reported measure combined a question on how participants felt at the time of follow-up compared to before the SARS-CoV-2 infection (fully recovered and symptom-free vs. other responses combined into non-recovered) and the EuroQol visual analogue scale (EQ-VAS); non-recovered participants were categorised into mild (EQ-VAS >70), moderate (EQ-VAS 51–70) and severe health impairment (EQ-VAS ≤50) based on population-normative values from previous research.5 , 57, 58, 59 Further measures indicating potential presence of PCC were individual self-reported COVID-19 related symptoms, commonly reported PCC-related symptom clusters (fatigue/physical exertion, cardiorespiratory (defined as dyspnoea, palpitation, or chest pain), or neurocognitive (defined as concentration, memory, or sleeping problems)), EuroQol 5-dimension 5-level scale (EQ-5D-5L), Fatigue Assessment Scale (FAS), 21-item Depression, Anxiety and Stress Scale (DASS-21), and modified Medical Research Council (mMRC) dyspnoea scale. Additional details on question wording and categorisation of PCC-related outcomes are provided in Supplementary Table S1. Sample size The sample size for this study was determined by the overall sample of individuals enrolled in the cohort. We based the sample size of the overall cohort on the epidemiological situation and expected case numbers at the time of study inception (May 2020). The cohort study had several aims and sample size calculations were adapted to meet the objectives of several research questions of relevance for the public health response to the pandemic. For this prospective cohort, we determined that a sample size of 1200 would be sufficient to comprehensively evaluate health outcomes over time, allow for pertinent group comparisons, and detect relevant specific health sequelae after infection.5 , 52 Statistical analysis We descriptively compared the reported work ability outcomes in individuals with self-reported COVID-19 related symptoms or reporting non-recovery with associated level of health impairment 12 months after diagnosis. We further descriptively analysed differences between individuals reporting individual symptoms, symptom clusters, or problems in any of the standardised health assessments (EQ-5D-5L overall and subdomains, FAS, DASS-21, mMRC dyspnoea scale) and those without. We descriptively analysed differences in work ability between participant subgroups based on sex (male vs. female), age (40–64 years vs. 18–39 years), comorbidity count (0–1 comorbidity vs. ≥2 comorbidities), and history of psychiatric diagnosis (present vs. absent), and based on the occurrence of new or worsened psychiatric diagnoses. Furthermore, we described the occupational changes experienced by participants overall and specifically attributed to PCC. We used univariable and multivariable regression models to evaluate the association of PCC-related outcomes with work ability outcomes. Model selection included age, sex, baseline health status, hospitalisation during acute infection as a priori covariates, with education level, comorbidity count, and history of psychiatric diagnosis added based on improved model fit using the Bayesian Information criterion (BIC; 2-point change considered relevant). We carefully examined all respective model assumptions, which were reasonably met. For current work ability (scores 0–10), we used linear regression in primary analyses to allow interpretation in terms of mean score differences and ordinal logistic regression in sensitivity analyses. We used ordinal logistic regression for Likert scale-based work ability outcomes. Correspondingly, we report adjusted linear model estimates (score differences) and adjusted odds ratios (aORs) with corresponding 95% confidence intervals (CIs). Unadjusted (crude) model results are provided in the Supplementary Material. We evaluated differences in the strength of associations (i.e., effect modification) between participant subgroups (as defined above) by using interaction models. We assumed missing data arising from non-response or loss to follow-up to be missing at random and excluded corresponding observations from the analyses. We based our reporting on the framework of the strength of statistical evidence (using levels ranging from weak to very strong evidence) instead of applying a single threshold to determine significance.60 , 61 We did not adjust p-values for multiple testing to facilitate the interpretation of our results based on this framework. We performed all statistical analyses using R (v4.2.2). Role of the funding source The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Results In the Zurich SARS-CoV-2 Cohort, 3185 individuals were randomly sampled and invited in the study by the Department of Health of the Canton of Zurich, 1294 agreed to being contacted by the study team, and 1106 consented for participation (participation rate 34.7%; Supplementary Fig. S1). 306 of 1106 cohort participants were not part of the working-age population, 15 were excluded due to reinfection, and 113 did not provide data at 12 months. Of 672 participants included in this study, 364 (54.2%) were female, 390 (58.0%) were aged 40–64 years, 79 (11.8%) were asymptomatic, and 9 (1.3%) were hospitalised at initial infection (Table 1 ). 19 participants (2.8%) reported being unemployed and 4 (0.6%) reported receiving disability insurance benefits at baseline. With respect to PCC, 120 of 672 (17.9%) participants reported having COVID-19 related symptoms, and 93 of 655 (14.2%; data on EQ-VAS missing for 7 individuals) reported not having recovered at 12 months, with mild (N = 72, 11.0%), moderate (N = 13, 2.0%), and severe health impairment (N = 8, 1.2%), respectively. There were differences in age, sex, severity of acute infection, comorbidities, and history of psychiatric diagnoses between those categorised as having PCC and those without.Table 1 Detailed study population characteristics, stratified by the presence of self-reported COVID-19 related symptoms and (non-)recovery and health impairment at 12 months after diagnosis of primary infection. Self-reported COVID-19 related symptoms (Non-) recovery and health impairment Overall No symptoms Symptoms Recovered Mild Moderate Severe (N = 552) (N = 120) (N = 562) (N = 72) (N = 13) (N = 8) (N = 672) Age (years) Mean (SD) 41.1 (12.2) 46.6 (10.9) 41.2 (12.2) 46.0 (11.4) 49.2 (8.8) 49.5 (12.8) 42.1 (12.2) Median (IQR) 41.0 (30.8–51.2) 50.0 (38.0–55.0) 41.5 (30.0–51.8) 48.0 (35.8–55.2) 51.0 (45.0–56.0) 53.0 (45.0–58.0) 43.0 (31.0–53.0) Range 18–63 21–62 18–63 25–62 32–59 24–62 18–63 Age group 18–39 years 249 (45.1%) 33 (27.5%) 250 (44.5%) 22 (30.6%) 3 (23.1%) 2 (25.0%) 282 (42.0%) 40–64 years 303 (54.9%) 87 (72.5%) 312 (55.5%) 50 (69.4%) 10 (76.9%) 6 (75.0%) 390 (58.0%) Sex Female 285 (51.6%) 79 (65.8%) 292 (52.0%) 47 (65.3%) 11 (84.6%) 6 (75.0%) 364 (54.2%) Male 267 (48.4%) 41 (34.2%) 270 (48.0%) 25 (34.7%) 2 (15.4%) 2 (25.0%) 308 (45.8%) Symptom count at infection Asymptomatic 67 (12.1%) 12 (10.0%) 70 (12.5%) 5 (6.9%) 1 (7.7%) 1 (12.5%) 79 (11.8%) 1–5 symptoms 233 (42.2%) 33 (27.5%) 230 (40.9%) 25 (34.7%) 1 (7.7%) 3 (37.5%) 266 (39.6%) ≥6 symptoms 252 (45.7%) 75 (62.5%) 262 (46.6%) 42 (58.3%) 11 (84.6%) 4 (50.0%) 327 (48.7%) Hospitalisation at infection Non-hospitalised 547 (99.1%) 115 (95.8%) 559 (99.5%) 69 (95.8%) 11 (84.6%) 7 (87.5%) 662 (98.5%) Hospitalised 5 (0.9%) 5 (4.2%) 3 (0.5%) 3 (4.2%) 2 (15.4%) 1 (12.5%) 10 (1.5%) with ICU stay 0 (0.0%) 1 (0.8%) 0 (0.0%) 0 (0.0%) 1 (7.7%) 0 (0.0%) 1 (0.1%) Smoking status Non-smoker 343 (62.4%) 70 (58.3%) 347 (62.0%) 42 (58.3%) 9 (69.2%) 6 (75.0%) 413 (61.6%) Ex-smoker 123 (22.4%) 33 (27.5%) 127 (22.7%) 20 (27.8%) 2 (15.4%) 1 (12.5%) 156 (23.3%) Smoker 84 (15.3%) 17 (14.2%) 86 (15.4%) 10 (13.9%) 2 (15.4%) 1 (12.5%) 101 (15.1%) Missing 2 (0.4%) 0 (0%) 2 (0.4%) 0 (0%) 0 (0%) 0 (0%) 2 (0.3%) BMI (kg/sqm) Mean (SD) 24.2 (4.3) 25.6 (5.1) 24.3 (4.3) 24.9 (4.7) 29.9 (7.5) 22.4 (1.9) 24.4 (4.5) Median (IQR) 23.6 (21.5–25.9) 24.8 (22.1–28.6) 23.6 (21.5–26.0) 24.5 (21.7–27.0) 30.4 (26.0–31.1) 22.3 (20.7–23.7) 23.7 (21.5–26.2) Range 13–63 17–45 17–63 18–40 20–45 20–25 13–63 Missing 5 (0.9%) 1 (0.8%) 6 (1.1%) 0 (0%) 0 (0%) 0 (0%) 6 (0.9%) Comorbiditya None 453 (82.1%) 79 (65.8%) 460 (81.9%) 49 (68.1%) 6 (46.2%) 4 (50.0%) 532 (79.2%) 1 comorbidity 80 (14.5%) 33 (27.5%) 85 (15.1%) 18 (25.0%) 5 (38.5%) 3 (37.5%) 113 (16.8%) ≥2 comorbidities 19 (3.4%) 8 (6.7%) 17 (3.0%) 5 (6.9%) 2 (15.4%) 1 (12.5%) 27 (4.0%) Comorbidity counta Median (IQR) 0 (0–0) 0 (0–1) 0 (0–0) 0 (0–1) 1 (0–1) 0.5 (0–1) 0 (0–0) Range 0–3 0–2 0–3 0–2 0–2 0–2 0–3 History of psychiatric diagnosis None 472 (88.6%) 93 (78.8%) 486 (87.7%) 63 (87.5%) 6 (50.0%) 5 (62.5%) 565 (86.8%) Any 61 (11.4%) 25 (21.2%) 68 (12.3%) 9 (12.5%) 6 (50.0%) 3 (37.5%) 86 (13.2%) Depression 27 (4.9%) 18 (15.0%) 32 (5.7%) 5 (6.9%) 5 (38.5%) 3 (37.5%) 45 (6.7%) Anxiety 12 (2.2%) 2 (1.7%) 13 (2.3%) 0 (0.0%) 0 (0.0%) 1 (12.5%) 14 (2.1%) Burnout 6 (1.1%) 3 (2.5%) 6 (1.1%) 1 (1.4%) 2 (15.4%) 0 (0.0%) 9 (1.3%) Bipolar disorder 2 (0.4%) 0 (0.0%) 2 (0.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.3%) ADHD 2 (0.4%) 1 (0.8%) 2 (0.4%) 1 (1.4%) 0 (0.0%) 0 (0.0%) 3 (0.4%) PTSD 4 (0.7%) 1 (0.8%) 4 (0.7%) 0 (0.0%) 0 (0.0%) 1 (12.5%) 5 (0.7%) Eating disorder 1 (0.2%) 1 (0.8%) 2 (0.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.3%) Sleep disorder 2 (0.4%) 0 (0.0%) 2 (0.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.3%) Other 2 (0.4%) 0 (0.0%) 2 (0.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.3%) Missing 19 (3.4%) 2 (1.7%) 8 (1.4%) 0 (0%) 1 (7.7%) 0 (0%) 21 (3.1%) Education level None or mandatory school 17 (3.1%) 5 (4.2%) 16 (2.9%) 4 (5.6%) 0 (0.0%) 0 (0.0%) 22 (3.3%) Vocational training or specialised baccalaureate 194 (35.1%) 55 (46.6%) 199 (35.5%) 33 (46.5%) 7 (53.8%) 3 (37.5%) 249 (37.2%) Higher technical school or college 165 (29.9%) 29 (24.6%) 173 (30.8%) 16 (22.5%) 2 (15.4%) 2 (25.0%) 194 (29.0%) University 176 (31.9%) 29 (24.6%) 173 (30.8%) 18 (25.4%) 4 (30.8%) 3 (37.5%) 205 (30.6%) Missing 0 (0%) 2 (1.7%) 1 (0.2%) 1 (1.4%) 0 (0%) 0 (0%) 2 (0.3%) Employment at infection Employed or self-employed 488 (88.4%) 99 (82.5%) 494 (87.9%) 61 (84.7%) 11 (84.6%) 4 (50.0%) 587 (87.4%) Student 42 (7.6%) 4 (3.3%) 45 (8.0%) 1 (1.4%) 0 (0.0%) 0 (0.0%) 46 (6.8%) Housewife/family manager 9 (1.6%) 1 (0.8%) 9 (1.6%) 1 (1.4%) 0 (0.0%) 0 (0.0%) 10 (1.5%) Unemployed 9 (1.6%) 10 (8.3%) 10 (1.8%) 7 (9.7%) 1 (7.7%) 1 (12.5%) 19 (2.8%) Disability insurance benefits 0 (0.0%) 4 (3.3%) 0 (0.0%) 0 (0.0%) 1 (7.7%) 3 (37.5%) 4 (0.6%) Other 4 (0.7%) 2 (1.7%) 4 (0.7%) 2 (2.8%) 0 (0.0%) 0 (0.0%) 6 (0.9%) Income <6′000 CHF 149 (28.0%) 40 (34.8%) 156 (28.7%) 23 (32.9%) 3 (25.0%) 4 (50.0%) 189 (29.2%) 6′000–12′000 CHF 231 (43.3%) 52 (45.2%) 234 (43.1%) 33 (47.1%) 7 (58.3%) 1 (12.5%) 283 (43.7%) >12′000 CHF 153 (28.7%) 23 (20.0%) 153 (28.2%) 14 (20.0%) 2 (16.7%) 3 (37.5%) 176 (27.2%) Missing 19 (3.4%) 5 (4.2%) 19 (3.4%) 2 (2.8%) 1 (7.7%) 0 (0%) 24 (3.6%) Nationality Swiss 465 (84.2%) 97 (80.8%) 479 (85.2%) 56 (77.8%) 9 (69.2%) 7 (87.5%) 562 (83.6%) Non-Swiss 87 (15.8%) 23 (19.2%) 83 (14.8%) 16 (22.2%) 4 (30.8%) 1 (12.5%) 110 (16.4%) ADHD, attention deficit hyperactivity disorder; BMI, body mass index; CHF, Swiss Francs; ICU, intensive care unit; IQR, interquartile range; PTSD, post-traumatic stress disorder; SD, standard deviation. a Comorbidities were assessed as any of the following: hypertension, diabetes, cardiovascular disease, chronic respiratory disease, chronic kidney disease, past or present malignancy, or immune suppression. In descriptive analyses of current work ability, ability related to physical and mental demands at work, and estimated future work ability in 2 years, there was a relevant reduction in work ability across all four outcomes among those reporting COVID-19 related symptoms compared to those without and among those reporting non-recovery compared to those that had recovered at 12 months (Fig. 1 and Supplementary Table S2). Work ability among those reporting non-recovery was more strongly reduced in those with moderate and severe health impairment compared to those with mild health impairment.Fig. 1 Current work ability, work ability related to physical and mental demands, and estimated future work ability in 2 years by presence of self-reported COVID-19 related symptoms and (non-)recovery and health impairment at 12 months after diagnosis of primary infection. Panels a–d demonstrate the level of current work ability (a), work ability related to physical (b) and mental (c) demands, and estimated work ability in 2 years (d) between individuals with self-reported COVID-19 related symptoms at 12 months compared to those without symptoms. Panels e–h show the level of current work ability (e), work ability related to physical (f) and mental (g) demands, and estimated work ability in 2 years (h) between individuals reporting non-recovery with mild, moderate, or severe health impairment at 12 months compared to those reporting full recovery at 12 months. In adjusted regression analyses, there was very strong evidence that current work ability scores were mean 0.62 (95% CI 0.30–0.95; p = 0.0002) points lower among those reporting COVID-19 related symptoms compared to those without (Fig. 2 and Supplementary Tables S3–S6). Current work ability scores were mean 0.55 (0.21–0.88, p = 0.0016), 3.37 (2.58–4.16, p < 0.0001), and 5.10 (4.16–6.04, p < 0.0001) points lower among those with non-recovery and mild, moderate, and severe health impairment, respectively, compared to those reporting full recovery (very strong evidence). Similarly, there was very strong evidence for a lower odds of having higher work ability with respect to physical (aOR 0.30, 95% CI 0.20–0.46, p < 0.0001) and mental (aOR 0.40, 0.27–0.62, p < 0.0001) demands among those reporting COVID-19 related symptoms compared to those without. Results were similar when evaluating non-recovered individuals compared to those reporting recovery, while reductions in work ability were more pronounced with higher levels of health impairment. There was no evidence for lower odds of having higher estimated future work ability in 2 years (aOR 0.52, 0.26–1.06, p = 0.074) among those reporting COVID-19 related symptoms compared to those without and among those with non-recovery and mild health impairment compared to those reporting recovery, but strong evidence for a reduction in those with moderate or severe health impairment compared to recovered participants. Sensitivity analyses treating current work ability as an ordinal outcome showed similar results (Supplementary Fig. S2 and Table S7).Fig. 2 Results from multivariable regression analyses of the association between presence of post COVID-19 condition (defined as presence of self-reported COVID-19 related symptoms and (non-)recovery and health impairment) and current work ability, work ability related to physical and mental demands, and estimated future work ability in 2 years at 12 months after diagnosis of primary infection. Each panel demonstrates results from multivariable linear regression (current work ability) or ordinal logistic regression (work ability related to physical and mental demands, estimated work ability in future) adjusted for sex, age, education level, baseline EuroQol visual analogue scale (EQ-VAS), comorbidity count, history of psychiatric diagnosis, and hospitalisation at acute infection. Separate models were estimated for the two definitions of post COVID-19 condition based on self-reported COVID-19 related symptoms (symptoms vs. no symptoms) and (non-)recovery and health impairment (severe, moderate or mild health impairment vs. recovery). Legend: CI, confidence interval; OR, odds ratio; PCC, post COVID-19 condition; Ref., reference. Further analyses demonstrate the association between the presence of specific symptom clusters (i.e., fatigue/physical exertion, cardiorespiratory, and neurocognitive symptoms), individual self-reported COVID-19 related symptoms, and presence of health problems in scale-based outcomes (EQ-5D-5L, FAS, DASS-21, and mMRC dyspnoea scale) and work ability outcomes at 12 months (Supplementary Fig. S3–S5 and Tables S8–S18). There was very strong evidence for an association between PCC-related symptom clusters and current work ability, as well as work ability related to physical and mental demands. Meanwhile, there was evidence for an association between some, but not all individual self-reported COVID-19 related symptoms and work ability outcomes. There was very strong evidence for an association between problems on any of the scale-based outcomes and current work ability, as well as work ability related to physical and mental demands. In subgroup analyses, there was strong evidence for a difference in the association (i.e., effect modification) of the presence of self-reported COVID-19 related symptoms with current work ability and work ability related to physical demands between participants aged 40–64 years and those aged 18–39 years, with a higher reduction in work ability in the older group (Table 2 , Supplementary Tables S19–S23). Meanwhile, there was no evidence for a difference in the association of self-reported COVID-19 related symptoms with any work ability outcome between male and female participants, or between participants with 0–1 comorbidity and participants with ≥2 comorbidities. Last, there was a stronger association of self-reported COVID-19 related symptoms with current work ability and work ability related to mental demands in participants with history of psychiatric diagnosis compared to those without. Further descriptive analyses demonstrated relevant differences between participants with different mental health trajectories, indicating a stronger reduction in work ability among participants with history of psychiatric diagnosis and those with a new or worsened psychiatric diagnosis compared to those without history or new or worsened diagnosis, respectively (Supplementary Table S24).Table 2 Results from multivariable regression analyses for the association of the presence of self-reported COVID-19 related symptoms with work ability outcomes at 12 months after diagnosis of primary infection (symptoms vs. no symptoms) within subgroups based on sex, age group, comorbidity count, or history of psychiatric diagnosis. Interaction Current work ability Physical demands Mental demands Future (2 years) N Adj. Estimate (95% CI) p-value N aOR (95% CI) p-value N aOR (95% CI) p-value N aOR (95% CI) p-value Male vs. female 634 633 631 633 Female −0.82 (−1.23 to −0.42) <0.0001 0.37 (0.22–0.62) 0.0002 0.40 (0.23–0.67) 0.0006 0.54 (0.23–1.27) 0.16 Male −0.30 (−0.81 to 0.21) 0.25 0.22 (0.11–0.43) <0.0001 0.42 (0.22–0.81) 0.0092 0.48 (0.14–1.64) 0.24 Differencea 0.53 (−0.12 to 1.17) 0.11 0.60 (0.26–1.38) 0.23 1.06 (0.46–2.42) 0.89 0.89 (0.21–3.88) 0.88 40–64 years vs. 18–39 years 634 633 631 633 18–39 years −0.27 (−0.86 to 0.32) 0.37 0.74 (0.33–1.65) 0.46 0.69 (0.33–1.44) 0.32 0.93 (0.19–4.45) 0.92 40–64 years −0.78 (−1.16 to −0.40) <0.0001 0.20 (0.12–0.34) <0.0001 0.33 (0.20–0.55) <0.0001 0.43 (0.19–0.96) 0.040 Differencea −0.51 (−1.20 to 0.19) 0.15 0.27 (0.11–0.71) 0.0064 0.48 (0.20–1.17) 0.11 0.46 (0.08–2.67) 0.37 ≥2 comorbidities vs. 0–1 comorbidityb 634 633 631 633 0–1 comorbidity −0.64 (−0.97 to −0.31) 0.0001 0.31 (0.20–0.47) <0.0001 0.39 (0.26–0.60) <0.0001 0.46 (0.22–0.96) 0.040 ≥2 comorbidities −0.92 (−2.34 to 0.50) 0.20 0.10 (0.01–0.79) 0.028 0.52 (0.08–3.34) 0.49 1.19 (0.10–14.80) 0.89 Differencea −0.28 (−1.74 to 1.18) 0.70 0.34 (0.04–2.67) 0.31 1.32 (0.20–8.90) 0.78 2.59 (0.19–36.17) 0.46 History of psychiatric diagnosis vs. no history of psychiatric diagnosis 634 633 631 633 No history of psychiatric diagnosis −0.39 (−0.74 to −0.04) 0.031 0.28 (0.17–0.44) <0.0001 0.35 (0.22–0.55) <0.0001 0.49 (0.22–1.08) 0.077 History of psychiatric diagnosis −1.73 (−2.47 to −0.99) <0.0001 0.49 (0.18–1.31) 0.15 0.88 (0.33–2.35) 0.80 0.65 (0.15–2.76) 0.56 Differencea −1.34 (−2.15 to −0.54) 0.0010 1.76 (0.60–5.17) 0.30 2.55 (0.88–7.38) 0.086 1.33 (0.27–6.67) 0.72 Adj., adjusted; CI, confidence interval; aOR, adjusted odds ratio. a Differences are interpreted as the difference in adjusted mean score differences (current work ability) or adjusted odds ratios (work ability related to physical and mental demands, estimated future work ability in 2 years) for the comparison between individuals with self-reported COVID-19 related symptoms compared with those without symptoms, quantifying the extent of effect modification for the respective stratification variable. P-values for differences were calculated using likelihood ratio tests for models with and without interaction term for the respective stratification variable. We used multivariable linear regression models (current work ability) and multivariable ordinal logistic regression models (work ability related to physical and mental demands, estimated future work ability in 2 years) including an interaction term for the respective stratification variable and adjusted for age (or age group for the corresponding analysis), sex, education status, baseline EuroQol visual analogue scale (EQ-VAS), comorbidity count (as a continuous variable, or as a dichotomous categorical variable for the corresponding analysis), history of psychiatric diagnosis, and hospitalisation due to COVID-19. b Comorbidities were assessed as any of the following: hypertension, diabetes, cardiovascular disease, chronic respiratory disease, chronic kidney disease, past or present malignancy, or immune suppression. When evaluating occupational changes up to 12 months, overall 119 (18.1%) participants reported to have had such a change during follow-up (Table 3 ), with a slightly higher proportion among participants reporting COVID-19 related symptoms (31/120, 25.8%) compared to those without such symptoms (88/552, 16.3%). 7 participants (1.1% of all participants, 5.8% of those with self-reported COVID-19 related symptoms) reported to have faced direct effects by PCC on their occupational situation. Work ability at 12 months was relevantly reduced among those 7 participants with occupational changes attributed to PCC compared to those without occupational changes and those with PCC-unrelated occupational changes (Supplementary Table S25).Table 3 Occupational changes related to post COVID-19 condition and overall, stratified by presence of self-reported COVID-19 related symptoms and (non-)recovery and health impairment at 12 months after diagnosis of primary infection. Self-reported COVID-19 related symptoms (Non-)recovery and health impairment Overall No symptoms Symptoms Recovered Mild Moderate Severe (N = 552) (N = 120) (N = 562) (N = 72) (N = 13) (N = 8) (N = 672) Occupational change No occupational change 451 (83.7%) 89 (74.2%) 468 (83.9%) 58 (80.6%) 6 (46.2%) 3 (37.5%) 540 (81.9%) Occupational change unrelated to PCC 88 (16.3%) 24 (20.0%) 90 (16.1%) 13 (18.1%) 4 (30.8%) 3 (37.5%) 112 (17.0%) Occupational change attributed to PCC 0 (0.0%) 7 (5.8%) 0 (0.0%) 1 (1.4%) 3 (23.1%) 2 (25.0%) 7 (1.1%) Missing 13 (2.4%) 0 (0%) 4 (0.7%) 0 (0%) 0 (0%) 0 (0%) 13 (1.9%) Reason for occupational changea Retired 4 (4.6%) 2 (6.5%) 5 (5.6%) 0 (0.0%) 0 (0.0%) 1 (20.0%) 6 (5.1%) On permanent sick leave 3 (3.4%) 1 (3.2%) 2 (2.2%) 0 (0.0%) 1 (14.3%) 0 (0.0%) 4 (3.4%) Receiving disability benefits 0 (0.0%) 1 (3.2%) 0 (0.0%) 0 (0.0%) 1 (14.3%) 0 (0.0%) 1 (0.8%) Work leave for different reason 3 (3.4%) 0 (0.0%) 3 (3.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (2.5%) Newly self-employed 2 (2.3%) 0 (0.0%) 2 (2.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (1.7%) Changed workplace 42 (48.3%) 14 (45.2%) 46 (51.7%) 7 (50.0%) 3 (42.9%) 0 (0.0%) 56 (47.5%) Changed position within same workplace 13 (14.9%) 4 (12.9%) 11 (12.4%) 4 (28.6%) 0 (0.0%) 1 (20.0%) 17 (14.4%) Started training or university studies 5 (5.7%) 0 (0.0%) 4 (4.5%) 1 (7.1%) 0 (0.0%) 0 (0.0%) 5 (4.2%) Reduced working hours 2 (2.3%) 0 (0.0%) 2 (2.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (1.7%) Lost employment 4 (4.6%) 2 (6.5%) 5 (5.6%) 0 (0.0%) 1 (14.3%) 0 (0.0%) 6 (5.1%) Other 9 (10.3%) 7 (22.6%) 9 (10.1%) 2 (14.3%) 1 (14.3%) 3 (60.0%) 16 (13.6%) Missing 1 (1.1%) 0 (0%) 1 (1.1%) 0 (0%) 0 (0%) 0 (0%) 1 (0.8%) Financial difficulties due to occupational changea No 57 (64.8%) 17 (54.8%) 56 (62.2%) 10 (71.4%) 4 (57.1%) 3 (60.0%) 74 (62.2%) Rather not 12 (13.6%) 6 (19.4%) 15 (16.7%) 3 (21.4%) 0 (0.0%) 0 (0.0%) 18 (15.1%) Yes, a little 14 (15.9%) 7 (22.6%) 15 (16.7%) 1 (7.1%) 3 (42.9%) 1 (20.0%) 21 (17.6%) Yes, very much 5 (5.7%) 0 (0.0%) 4 (4.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 5 (4.2%) Unclear/no answer 0 (0.0%) 1 (3.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (20.0%) 1 (0.8%) PCC, post COVID-19 condition. a Percentages calculated within total of individuals with any occupational change (N = 119). The 7 participants with occupational changes attributed to PCC reported various individual stories in how PCC affected their work life. One participant lost their work due to PCC. Another reported to be on permanent sick leave at 12 months and being severely affected in daily life. One participant reported that they were unemployed at baseline and could not take on a new position due to PCC, and one was in a job re-integration program but was unable to re-enter the job market due to PCC. Another participant was so severely impacted cognitively that they could no longer use their professional skills (university level) and had to switch to doing simple administrative tasks. One health care worker reported that they had to take a different position that did not require working night shifts. And one participant reported that they had to discontinue their self-employed work as an instructor and seek another part-time job to cope financially because of PCC. Overall, 3 participants (43%) with PCC-attributed occupational changes reported to have some financial difficulties as a result of their condition and their resulting occupational situation. Discussion In this prospective population-based cohort of working-age individuals previously infected with SARS-CoV-2, we found that the presence of self-reported COVID-19 related symptoms was strongly associated with a reduction in work ability at 12 months after diagnosis. Among non-recovered, higher levels of health impairment were also associated with substantially lower current work ability and work ability related to physical and mental demands. We found strong evidence that higher age and a history of psychiatric diagnosis was associated with a stronger reduction in current work ability. About 1 in 15 of those with self-reported COVID-19 related symptoms reported having had occupational changes attributed to PCC within one year, with 1.6% completely dropping out of the workforce. Evidence on the impact of PCC on the occupational situation and work-related impairments is limited and heterogenous.16, 17, 18, 19, 20, 21 , 23, 24, 25, 26, 27, 28, 29, 30 , 32 , 34, 35, 36, 37 Prior studies have primarily evaluated specific populations, such as PCC-affected individuals recruited through specialised post COVID-19 clinics18 , 23 , 27 , 29 , 32 , 35 , 36 or social media,19 , 20 , 25 , 28 , 34 , 37 hospitalised COVID-19 patients,16 , 21 , 26 , 29 or healthcare workers,24 , 31 resulting in limited generalisability. They have reported that between 11% and 52% of affected individuals do not return to work16 , 18, 19, 20 , 25, 26, 27, 28, 29, 30 , 34 , 35 , 37 and that 10%–72% do not fully regain their work capacity 6–12 months after infection.19 , 21 , 23 , 25 , 27 , 31 , 32 , 34 , 36 , 37 We estimated this proportion to be 5.8% after one year, which is relevantly lower than reported in other studies. This is likely explained by differences in the evaluated populations and assessment time points (only few studies had a follow-up of six months or longer). Differences between countries in terms of sickness and disability benefits systems, as well as cultural and organisational factors, may also explain the wide range of estimates in the literature. Nevertheless, the impact of PCC on the working-age population appears to be substantial and will likely lead to long-term burdens on economic and healthcare systems. An important factor that determines sustainable return to work is the perceived work ability, which is also more independent of the specific context than return to work and occupational changes. To date, few studies have evaluated work ability in the context of PCC, which were also conducted within selective populations.22 , 31 , 33 Evidence from these studies and our study demonstrated lower work ability scores among those with PCC, with a higher reduction among those with occupational changes. However, it is important to note that although most of the participants with PCC did not have occupational changes and remained at work, decreased work ability in this group may still indicate reduced productivity and efficiency. Sickness presenteeism (i.e., continuing to work while sick) may have negative effects on both the individuals and their employers.62 Sick employees usually need extra efforts to cope with job demands which may lead to additional worsening of their health, and the costs of having a sick employee are estimated to be the same as or even higher than their actual absence.62 Strategies that improve work-related capacity in individuals affected by PCC and promote return to work are urgently needed. In addition, since reduced work ability also is a predictor of early retirement,49 , 50 it will be vital in the coming years to continuously monitor whether there are increases in the number of people retiring early due to PCC. We found a more substantial decrease in current work ability among individuals aged 40–64 years compared to younger individuals. This is concerning since the middle-aged population is typically viewed as the foundation of most economies, as they account for a significant proportion of the workforce, tax revenue, and gross domestic product. We also found that individuals with a history of psychiatric diagnosis had a greater reduction in work ability than those without. The relationship between work and mental health is well-established in the literature.40 , 63 Similarly, the association of pre-existing psychiatric disorders with PCC has also been demonstrated in several studies, with evidence of a higher risk of PCC among those with anxiety or depression prior to infection.53 , 64 Effectively, such conditions may simultaneously be a risk factor for the development of PCC (if pre-existing),53 , 64 part of the broader symptom complex of PCC,1 , 2 or a consequence of other (non-psychiatric) PCC-related symptoms,65 which are difficult to separate. Targeted strategies and support measures from occupational and rehabilitation medicine, possibly leveraging pre-existing programs for individuals with chronic illnesses, should be put in place to support individuals affected by PCC. In addition, both employees and employers need to be made aware of the mental health aspects of PCC and the impact of mental health on work, as health-promoting working conditions and, for example, supportive leadership may be relevant to the re-integration of relevant subgroups of employees.66 Fallout from reduced work capacity results not only in financial and health challenges for individuals affected by PCC, but can also have substantial consequences for public health and the economy and society in the longer term. Altogether, our findings underline the necessity for interdisciplinary interventions aimed at individuals affected by PCC, including those with moderate or even mild health impairment. Given that early intervention is a core principle of occupational rehabilitation, further research is warranted to determine whether earlier rehabilitation could improve work outcomes in people with persistent symptoms after COVID-19 but who are not yet diagnosed with PCC. Identifying specific COVID-19 symptoms that predict impairment in work ability will help to develop and provide such early interventions. From this perspective, it will also be crucial to determine what size of reduction in the work ability of affected individuals can be considered relevant (i.e., minimal important difference) for the context of PCC, which will aid in the design and interpretation of trials evaluating rehabilitation measures. Strengths of the study include its population-based approach, the recruitment of participants at or shortly after diagnosis during a time period when PCC was not yet a concern, the large sample size, and the high retention rate at one year (90%) limiting emigrative selection bias arising from loss to follow-up. In addition, the granularity of the data and the use of a validated, internationally used, and context-independent measure of work ability strengthens our evaluation. However, some limitations need to be considered. First, the participation rate was relatively low (35%). Immigrative selection may have occurred if individuals who were more health literate were more likely to participate or if individuals who had PCC and were more severely impacted were also more likely to be retained in the study. This may have led to an overestimation of the association between PCC and work ability. We previously evaluated differences between cohort participants and individuals not participating in our study and found that those in our study were less likely to be hospitalised and younger on average.5 This may have biased our findings towards lower estimates. Hence, the direction of any potential bias is unclear. Second, the relatively low proportion of hospitalised participants limits the generalisability of our results to those with the most severe acute disease, who may also suffer from more severe medical complications and sequelae of the hospital stay (e.g., post intensive care syndrome). Additionally, the generalisability of our findings to individuals infected with emerging SARS-CoV-2 variants of concern or who were vaccinated prior to infection is limited, since our participants were all infected with wildtype SARS-CoV-2 and unvaccinated at infection. The risk of PCC and severe health impairment is substantially reduced with vaccination and infection with newer variants, but still present.67, 68, 69, 70 As the impact of PCC on work ability is likely comparable in these contexts, this may have significant socioeconomic implications given that more than 45% of the global population is estimated to have been infected with the Omicron variant.71 Further research is needed to evaluate whether similar reduced work ability and occupational changes are observed in vaccinated populations and in the context of emerging variants of concern. Nonetheless, the population from the early stages of the pandemic included in this study remains highly relevant since these are the individuals experiencing long-term health consequences at present, posing a challenge to public health. Third, we assessed PCC using self-reported measures. Since we could not conduct a clinical validation of PCC (i.e., assess whether self-reported COVID-19 related symptoms or reported health impairment were indeed attributable to SARS-CoV-2 infection), we cannot fully exclude that reported symptoms and health impairment were related to the presence or worsening of pre-existing or incident conditions or other infections. Yet, we consider self-reported measures key in capturing the lived experience of those affected. Our study was not designed to fully capture all possible fluctuations or relapses of symptoms during follow-up. However, the comparable results across two different definitions of PCC and our previous findings of a higher prevalence of several symptoms among Zurich SARS-CoV-2 Cohort participants compared to an uninfected sample from the general population further strengthen the credibility of our findings.5 Fourth, multiple hypothesis tests were conducted in association analyses within this study, which may have resulted in spurious (false positive) findings. Fifth, we did not have data on participants’ work ability prior to SARS-CoV-2 infection or on work-related outcomes in a comparable non-infected group. Thus, we could not evaluate changes in work ability scores from before infection and whether these met a minimal important difference threshold. We also cannot be fully certain that the reduced work ability or occupational changes are entirely due to SARS-CoV-2 infection and not other causes, such as pre-existing or incident conditions or other effects of the pandemic. While we at least partially accounted for the lack of work-related information prior to infection by adjusting for baseline health status in our models, this may not have fully resolved this issue. However, work ability scores among those categorised as not having PCC in our study were broadly comparable with estimates from the Swedish general population in 2017, while scores among those affected by PCC were lower (Supplementary Table S26).48 In addition, the detailed evaluation of the participants' individual stories supports our finding of a reduced work ability related to PCC. In contrast, since we relied on participants’ self-reporting of occupational changes related to SARS-CoV-2 infection, it is possible that we did not capture all changes attributable to PCC if they were not perceived or reported as such by participants. This may have resulted in an underestimation of the proportion with PCC-attributable occupational changes. In conclusion, this population-based study found that the presence of PCC was associated with a reduction in work ability one year after SARS-CoV-2 infection and led to an inability to work in some of the infected individuals. Such reductions in productivity and work capacity can have severe implications for individuals, families, and society as a whole. It is critical that policymakers, healthcare professionals, and employers recognise the impact of PCC on the workforce and develop effective strategies and interventions that can support and enable affected individuals in regaining and retaining their work ability. Contributors TB, DM, JSF, and MAP conceived and planned the Zurich SARS-CoV-2 Cohort study. TB, DM, and MAP coordinated the Zurich SARS-CoV-2 Cohort study. PK, TB, MAP, and DM conceived and planned this analysis. TB, DM, AD, and MAP contributed to participant recruitment and data collection. MAP supervised the project. JSF and MAP obtained funding. TB and DM accessed and verified the data and prepared the analytic datasets. All authors had full access to the data. DM performed the statistical analysis and SH provided input on the statistical analysis. All authors contributed to the interpretation of the findings. PK, TB, and DM wrote the draft manuscript. All authors critically revised and provided feedback on the draft manuscript. All authors accept full responsibility for the content of the paper and have seen and approved the final manuscript. PK and TB contributed equally. Data sharing statement Deidentified individual participant data underlying the findings of this study will be available for researchers submitting a methodologically sound proposal to achieve the aims of the proposal. Proposals should be directed at the corresponding author (Prof. Dr. Milo A. Puhan, miloalan.puhan@uzh.ch). Declaration of interests JF reports receiving grants from Gilead Sciences Switzerland, ViiV healthcare, and Merck, unrelated to this work. The other authors declare no conflicts of interest. Appendix ASupplementary data Supplementary Materials Acknowledgements This study is part of the Corona Immunitas research network, coordinated by the Swiss School of Public Health (SSPH+), and funded by fundraising of SSPH+ including funds of the Swiss Federal Office of Public Health and private funders (ethical guidelines for funding stated by SSPH+ were respected), by funds of the Cantons of Switzerland (Vaud, Zurich, and Basel) and by institutional funds of the Universities. Additional funding specific to this study was received from the Department of Health of the Canton of Zurich, the 10.13039/501100002329 Swiss Federal Office of Public Health , the University of Zurich (UZH) Foundation, and the European Union's Horizon Europe Research and Innovation Programme under grant no 101046041 through the CoVICIS project. PK received funding from the 10.13039/501100001711 Swiss National Science Foundation , grant no 100019M_201113. TB received funding from the European Union’s Horizon 2020 Research and Innovation Programme under the Marie Skłodowska-Curie grant no 801076, through the SSPH + Global PhD Fellowship Program in 10.13039/100015425 Public Health Sciences (GlobalP3HS) of the SSPH+. JL received funding from Swiss Learning Health Systems (SHLS) through a PhD scholarship. DM received funding by the 10.13039/501100006447 University of Zurich Postdoc Grant, grant no FK-22-053. The authors thank the study administration team and the study participants for their continued and highly valuable support. Appendix A Supplementary data related to this article can be found at https://doi.org/10.1016/j.lanepe.2023.100671. ==== Refs References 1 NIHR Evidence Living with Covid 19 – second review 2021 10.3310/themedreview_45225 published online March 16 2 Nittas V. Gao M. West E.A. Long COVID through a public health lens: an umbrella review Publ Health Rev 43 2022 1604501 3 Global Burden of Disease Long COVID Collaborators Estimated global proportions of individuals with persistent fatigue, cognitive, and respiratory symptom clusters following symptomatic COVID-19 in 2020 and 2021 JAMA 328 16 2022 1604 1615 10.1001/jama.2022.18931 published online Oct 10 36215063 4 Ballering A.V. van Zon S.K.R. Olde Hartman T.C. Rosmalen J.G.M. Lifelines Corona Research Initiative Persistence of somatic symptoms after COVID-19 in the Netherlands: an observational cohort study Lancet 400 2022 452 461 35934007 5 Ballouz T. Menges D. Anagnostopoulos A. Recovery and symptom trajectories up to two years after SARS-CoV-2 infection: population based, longitudinal cohort study BMJ 381 2023 e074425 6 Bull-Otterson L. Post–COVID conditions among adult COVID-19 survivors aged 18–64 and ≥65 Years — United States, March 2020–November 2021 MMWR Morb Mortal Wkly Rep 71 2022 713 717 7 Hastie C.E. Lowe D.J. McAuley A. Outcomes among confirmed cases and a matched comparison group in the Long-COVID in Scotland study Nat Commun 13 2022 5663 36224173 8 Perlis R.H. Santillana M. Ognyanova K. Prevalence and correlates of long COVID symptoms among US adults JAMA Netw Open 5 2022 e2238804 9 Peter R.S. Nieters A. Kräusslich H.-G. Post-acute sequelae of covid-19 six to 12 months after infection: population based study BMJ 379 2022 e071050 10 Thompson E.J. Williams D.M. Walker A.J. Long COVID burden and risk factors in 10 UK longitudinal studies and electronic health records Nat Commun 13 2022 3528 35764621 11 Crook H. Raza S. Nowell J. Young M. Edison P. Long covid—mechanisms, risk factors, and management BMJ 374 2021 n1648 34312178 12 Davis H.E. McCorkell L. Vogel J.M. Topol E.J. Long COVID: major findings, mechanisms and recommendations Nat Rev Microbiol 21 2023 133 146 36639608 13 Gualano M.R. Rossi M.F. Borrelli I. Returning to work and the impact of post COVID-19 condition: a systematic review Work 73 2022 405 413 35938280 14 Cutler D.M. The costs of long COVID JAMA Health Forum 3 2022 e221809 15 Office for National Statistics Self-reported long COVID and labour market outcomes, UK: 2022 2023 https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/selfreportedlongcovidandlabourmarketoutcomesuk2022/selfreportedlongcovidandlabourmarketoutcomesuk2022 16 Amorim C.E.N. Gomes V.L.T. Cristelli M.P. High prevalence of long-COVID among kidney transplant recipients: a longitudinal cohort study Transplantation 106 2022 2408 36228200 17 Buonsenso D. Gualano M.R. Rossi M.F. Post-acute COVID-19 sequelae in a working population at one year follow-up: a wide range of impacts from an Italian sample Int J Environ Res Publ Health 19 2022 11093 18 Danesh V. Arroliga A.C. Bourgeois J.A. Widmer A.J. McNeal M.J. McNeal T.M. Post-acute sequelae of COVID-19 in adults referred to COVID recovery clinic services in an integrated health system in Texas Proc (Bayl Univ Med Cent) 34 2021 645 648 34732978 19 Davis H.E. Assaf G.S. McCorkell L. Characterizing long COVID in an international cohort: 7 months of symptoms and their impact EClinicalMedicine 38 2021 101019 20 Diem L. Schwarzwald A. Friedli C. Multidimensional phenotyping of the post-COVID-19 syndrome: a Swiss survey study CNS Neurosci Ther 28 2022 1953 1963 35975339 21 Dryden M. Mudara C. Vika C. Post-COVID-19 condition 3 months after hospitalisation with SARS-CoV-2 in South Africa: a prospective cohort study Lancet Global Health 10 2022 e1247 e1256 35961348 22 Ekstrand E. Brogårdh C. Axen I. Fänge A.M. Stigmar K. Hansson E.E. Perceived consequences of post-COVID-19 and factors associated with low life satisfaction Int J Environ Res Public Health 19 2022 15309 23 Heightman M. Prashar J. Hillman T.E. Post-COVID-19 assessment in a specialist clinical service: a 12-month, single-centre, prospective study in 1325 individuals BMJ Open Respir Res 8 2021 e001041 24 Kisiel M.A. Janols H. Nordqvist T. Predictors of post-COVID-19 and the impact of persistent symptoms in non-hospitalized patients 12 months after COVID-19, with a focus on work ability Ups J Med Sci 127 2022 10.48101/ujms.v127.8794 25 Lunt J. Hemming S. Burton K. Elander J. Baraniak A. What workers can tell us about post-COVID workability Occup Med (Lond) 2022 kqac086 26 Mazer B. Ehrmann Feldman D. Functional limitations in individuals with long COVID Arch Phys Med Rehabil 2023 10.1016/j.apmr.2023.03.004 published online March 24 27 Miskowiak K.W. Pedersen J.K. Gunnarsson D.V. Cognitive impairments among patients in a long-COVID clinic: prevalence, pattern and relation to illness severity, work function and quality of life J Affect Disord 324 2023 162 169 36586593 28 O’Mahony L. Buwalda T. Blair M. Impact of Long COVID on health and quality of life HRB Open Res 5 2022 31 36101871 29 O’Brien K. Townsend L. Dowds J. 1-year quality of life and health-outcomes in patients hospitalised with COVID-19: a longitudinal cohort study Respir Res 23 2022 115 35509060 30 Perlis R.H. Lunz Trujillo K. Safarpour A. Association of post–COVID-19 condition symptoms and employment status JAMA Netw Open 6 2023 e2256152 31 Peters C. Dulon M. Westermann C. Kozak A. Nienhaus A. Long-term effects of COVID-19 on workers in health and social services in Germany Int J Environ Res Publ Health 19 2022 6983 32 Rutsch M. Frommhold J. Buhr-Schinner H. Gross T. Schüller P.O. Deck R. Pneumologische Rehabilitation bei Long Covid – gesundheitliche Veränderungen am Ende der stationären Rehabilitationsmaßnahme Rehabilitation (Stuttg) 2023 10.1055/a-1964-7401 published online Jan 17 33 Sansone D. Tassinari A. Valentinotti R. Persistence of symptoms 15 Months since COVID-19 diagnosis: prevalence, risk factors and residual work ability Life 13 2023 97 34 Vaes A.W. Goërtz Y.M.J. Herck M.V. Recovery from COVID-19: a sprint or marathon? 6-month follow-up data from online long COVID-19 support group members ERJ Open Research 7 2021 e00141-2021 10.1183/23120541.00141-2021 35 Van Wambeke E. Bezler C. Kasprowicz A.-M. Charles A.-L. Andres E. Geny B. Two-years follow-up of symptoms and return to work in complex post-COVID-19 patients J Clin Med 12 2023 741 36769389 36 Yelin D. Margalit I. Nehme M. Patterns of long COVID symptoms: a multi-center cross sectional study J Clin Med 11 2022 898 35207171 37 Ziauddeen N. Gurdasani D. O’Hara M.E. Characteristics and impact of long covid: findings from an online survey PLoS One 17 2022 e0264331 38 Cancelliere C. Donovan J. Stochkendahl M.J. Factors affecting return to work after injury or illness: best evidence synthesis of systematic reviews Chiropr Man Therap 24 2016 32 39 Figueredo J.-M. García-Ael C. Gragnano A. Topa G. Well-being at work after return to work (RTW): a systematic review Int J Environ Res Publ Health 17 2020 7490 40 Gragnano A. Negrini A. Miglioretti M. Corbière M. Common psychosocial factors predicting return to work after common mental disorders, cardiovascular diseases, and cancers: a review of reviews supporting a cross-disease approach J Occup Rehabil 28 2018 215 231 28589524 41 Ilmarinen J. Work ability—a comprehensive concept for occupational health research and prevention Scand J Work Environ Health 35 2009 1 5 19277432 42 van den Berg T.I.J. Elders L.A.M. de Zwart B.C.H. Burdorf A. The effects of work-related and individual factors on the Work Ability Index: a systematic review Occup Environ Med 66 2009 211 220 19017690 43 Tuomi K. Ilmarinen J. Jahkola A. Katajarinne L. Tulkki A. Work ability index 19 1997 Finnish institute of occupational health 44 Ahlstrom L. Grimby-Ekman A. Hagberg M. Dellve L. The work ability index and single-item question: associations with sick leave, symptoms, and health – a prospective study of women on long-term sick leave Scand J Work Environ Health 36 2010 404 412 20372766 45 Kujala V. Tammelin T. Remes J. Vammavaara E. Ek E. Laitinen J. Work ability index of young employees and their sickness absence during the following year Scand J Work Environ Health 32 2006 75 84 16539175 46 Kinnunen U. Nätti J. Work ability score and future work ability as predictors of register-based disability pension and long-term sickness absence: a three-year follow-up study Scand J Publ Health 46 2018 321 330 47 Gustafsson K. Marklund S. Consequences of sickness presence and sickness absence on health and work ability: a Swedish prospective cohort study Int J Occup Med Environ Health 24 2011 153 165 21526385 48 Lundin A. Leijon O. Vaez M. Hallgren M. Torgén M. Predictive validity of the Work Ability Index and its individual items in the general population Scand J Publ Health 45 2017 350 356 49 van den Berg T.I.J. Elders L.A.M. Burdorf A. Influence of health and work on early retirement J Occup Environ Med 52 2010 576 583 20523241 50 Sell L. Predicting long-term sickness absence and early retirement pension from self-reported work ability Int Arch Occup Environ Health 82 2009 1133 1138 19365638 51 Alavinia S.M. de Boer A.G.E.M. van Duivenbooden J.C. Frings-Dresen M.H.W. Burdorf A. Determinants of work ability and its predictive value for disability Occup Med 59 2009 32 37 52 ISRCTN14990068: Zurich Coronavirus Cohort: an observational study to determine long-term clinical outcomes and immune responses after coronavirus infection (COVID-19), assess the influence of virus genetics, and examine the spread of the coronavirus in the population of the Canton of Zurich 2020 ISRCTN Registry Switzerland 10.1186/ISRCTN14990068 53 Tsampasian V. Elghazaly H. Chattopadhyay R. Risk factors associated with Post−COVID-19 condition: a systematic review and meta-analysis JAMA Intern Med 183 6 2023 566 580 10.1001/jamainternmed.2023.0750 published online March 23 36951832 54 von Bonsdorff M.B. Seitsamo J. Ilmarinen J. Nygård C.-H. von Bonsdorff M.E. Rantanen T. Work ability in midlife as a predictor of mortality and disability in later life: a 28-year prospective follow-up study CMAJ 183 2011 E235 E242 21282315 55 Munblit D. Nicholson T. Akrami A. A core outcome set for post-COVID-19 condition in adults for use in clinical practice and research: an international Delphi consensus study Lancet Respir Med 10 2022 715 724 35714658 56 Soriano J.B. Murthy S. Marshall J.C. Relan P. Diaz J.V. A clinical case definition of post-COVID-19 condition by a Delphi consensus Lancet Infect Dis 22 2022 e102 e107 34951953 57 Perneger T.V. Combescure C. Courvoisier D.S. General population reference values for the French version of the EuroQol EQ-5D health utility instrument Value Health 13 2010 631 635 20412541 58 Zanini A. Aiello M. Adamo D. Estimation of minimal clinically important difference in EQ-5D visual analog scale score after pulmonary rehabilitation in subjects with COPD Respir Care 60 2015 88 95 25336531 59 Wacker M.E. Jörres R.A. Karch A. Assessing health-related quality of life in COPD: comparing generic and disease-specific instruments with focus on comorbidities BMC Pulm Med 16 2016 70 27160582 60 Bland M. An introduction to medical statistics 2015 Oxford University Press 117 61 Muff S. Nilsen E.B. O’Hara R.B. Nater C.R. Rewriting results sections in the language of evidence Trends Ecol Evol 37 2022 203 210 34799145 62 Kinman G. Sickness presenteeism at work: prevalence, costs and management Br Med Bull 129 1 2019 69 78 10.1093/bmb/ldy043 published online Jan 11 30649219 63 van den Berg S. Burdorf A. Robroek S.J.W. Associations between common diseases and work ability and sick leave among health care workers Int Arch Occup Environ Health 90 2017 685 693 28550420 64 Wang S. Quan L. Chavarro J.E. Associations of depression, anxiety, worry, perceived stress, and loneliness prior to infection with risk of post–COVID-19 conditions JAMA Psychiatr 79 2022 1081 1091 65 Goodman M.L. Molldrem S. Elliott A. Robertson D. Keiser P. Long COVID and mental health correlates: a new chronic condition fits existing patterns Health Psychol Behav Med 11 2023 2164498 66 Bavel J.J.V. Baicker K. Boggio P.S. Using social and behavioural science to support COVID-19 pandemic response Nat Hum Behav 4 2020 460 471 32355299 67 Ballouz T. Menges D. Kaufmann M. Post COVID-19 condition after Wildtype, Delta, and Omicron SARS-CoV-2 infection and prior vaccination: pooled analysis of two population-based cohorts PLoS One 18 2023 e0281429 68 Azzolini E. Levi R. Sarti R. Association between BNT162b2 vaccination and long COVID after infections not requiring hospitalization in health care workers JAMA 328 2022 676 678 35796131 69 Notarte K.I. Catahay J.A. Velasco J.V. Impact of COVID-19 vaccination on the risk of developing long-COVID and on existing long-COVID symptoms: a systematic review eClinicalMedicine 53 2022 101624 70 Antonelli M. Pujol J.C. Spector T.D. Ourselin S. Steves C.J. Risk of long COVID associated with delta versus omicron variants of SARS-CoV-2 Lancet 399 2022 2263 2264 35717982 71 COVID-19 Forecasting Team Forecasting the trajectory of the COVID-19 pandemic into 2023 under plausible variant and intervention scenarios: a global modelling study 2023 Institute for Health Metrics and Evaluation published online March 9 https://www.healthdata.org/research-article/forecasting-trajectory-covid-19-pandemic-2023-under-plausible-variant-and
PMC010xxxxxx/PMC10288173.txt	==== Front JAAD Case Rep JAAD Case Rep JAAD Case Reports 2352-5126 by the American Academy of Dermatology, Inc. Published by Elsevier, Inc. S2352-5126(23)00180-7 10.1016/j.jdcr.2023.05.028 Case Report Transient dermatomyositis-like reaction following COVID-19 messenger RNA vaccination Gutierrez Rodrigo A. BS a Connolly Kari MD b Gross Andrew MD c Haemel Anna MD b∗ a School of Medicine, University of California, San Francisco b Department of Dermatology, University of California, San Francisco c Department of Rheumatology, University of California, San Francisco ∗ Correspondence to: Anna Haemel, MD, Department of Dermatology, University of California, 1701 Divisadero St, San Francisco, CA 94115. 1 6 2023 7 2023 1 6 2023 37 128130 © 2023 by the American Academy of Dermatology, Inc. Published by Elsevier, Inc. 2023 American Academy of Dermatology, Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Key words autoimmune disease dermatomyositis drug side effects mRNA vaccine type 1 interferons vaccine reaction Abbreviations used DM dermatomyositis mRNA messenger RNA ==== Body pmcIntroduction The spectrum of cutaneous immune responses previously observed with type I interferon therapy for melanoma and hepatitis C as well as the remarkable effectiveness of Janus Kinase inhibitors in dermatology highlight the central role of type I interferons in inflammatory skin disease.1 , 2 Early in the COVID-19 pandemic, reports of “pandemic chilblains” provided dermatologic indicators as to the importance of type I interferon signaling in the COVID-19 immune response; a patient with COVID-19 myopathy was subsequently found to have features of a type I interferonopathy, with pathogenesis akin to what is seen in idiopathic dermatomyositis (DM).3 , 4 Multiple case reports are now emerging of new or flaring DM associated with COVID-19 infection and/or messenger RNA (mRNA) vaccination.5 Although causality is difficult to establish, the key role of type I interferons in both COVID-19 and DM provides biologic plausibility for such an association. Notably, COVID-19-induced autoimmunity appears to be temporary in at least some patients; a transient inflammatory response may thus represent one clue toward a COVID-19 viral/vaccine-related trigger.6 Here, we report a classic, skin biopsy-supported DM-like reaction that occurred 2 weeks after COVID-19 mRNA vaccination and lasted 3 to 4 months before fully and spontaneously resolved. Case report A 50-year-old woman with no known history of COVID-19 infection presented with myalgias, muscle weakness, and skin eruption beginning 2 weeks after her Pfizer/BioNTech BNT162b2 booster. She had previously received a single dose of the Johnson & Johnson vaccine without side effects. COVID-19 polymerase chain reaction (PCR) was negative at symptom onset. There was no associated weight loss, cough, or dysphagia. Medical history included a single pneumonia 6 months prior as well as mild ulcerative colitis managed with dietary measures; colonoscopy and all other routine screenings were up to date. The patient was on no medications at symptom onset. Examination revealed erythematous lichenoid papules over the digits and elbows and a bilateral violaceous eruption over the lateral aspect of the thighs, consistent with Gottron’s papules, Gottron’s sign, and Holster sign, respectively. Examination also revealed violaceous psoriasiform erythema on scalp, but heliotrope eruptions, cuticular changes, and gross muscle weakness were absent (Figs 1 and 2 ). Workup revealed a creatine kinase (CK) >2700, antinuclear antibody (ANA) 1:160 with homogenous staining, and a negative limited myositis antibody panel, including Mi-2. Shave biopsy from the scalp demonstrated interface dermatitis with dermal mucinosis; shave biopsy from the right side of the second metacarpophalangeal joint similarly demonstrated interface dermatitis (Fig 3 ). Direct immunofluorescence for both sites revealed granular junctional deposition of C3 with immunoglobulin M also present for the hand biopsy; IgA/IgG were negative for both the sites. Given the clinicopathologic features concerning for DM, the patient was treated with a prednisone taper from 40 mg per day to off over several weeks with brisk and full normalization of muscle enzymes. The skin eruptions, particularly over the thighs, were slower to resolve, lasting several months. At 4 months from initial presentation, the patient reported resolution of all symptoms, including her skin lesions off treatment with follow-up laboratory testings demonstrating normal CK and aldolase and negative myositis antibody panel, including Mi-2, MDA-5, TIF-1 gamma, and NXP-2 (Fig 4 ).Fig 1 Image of scalp showing violaceous, psoriasiform erythema. Fig 2 Characteristic Gottron’s papules over the dorsal aspects of the metacarpophalangeal and proximal interphalangeal joints. Fig 3 Hematoxylin-eosin stain demonstrating interface dermatitis with dermal mucin. Direct immunofluorescence (not shown) demonstrated granular deposits of immunoglobulin M along DEJ (100×). Photo courtesy of Dr. Rony A. Francois Fig 4 Resolution of Gottron’s papules at 4-month follow-up. Discussion Multiple infectious triggers have been implicated in the onset of DM, such as parvovirus B19, hepatitis, and Epstein–Barr virus as well as toxoplasmosis; COVID-19/COVID-19 mRNA vaccinations, and other viral triggers, may produce a surge in type I interferons.7, 8, 9 Type I interferons, in turn, have the potential to promote or unmask autoimmunity in susceptible patients and/or in conditions where type I interferon signaling is central to disease pathogenesis, including DM.1 A real-world study in an autoimmune dermatology clinic further demonstrated that patients with existing DM were more likely to flare post COVID-19 vaccination than those with lupus.5 It is important to note that while such reactions to vaccines appear to be rare (at the level of case reports), earlier in the pandemic COVID-19-associated inflammatory myopathies, which share features with DM, may have affected up to one-third of COVID-19 patients.4 , 10 Regarding the relationship between COVID-19 and COVID-19 myopathy, a broad spectrum ranging from an acute worsening of existing myositis to chronic, postinfectious myalgic syndromes have been reported in infected patients, with anti-MDA5 antibody titers trending with severity.4 , 11 In previously reported cases of myositis induced by COVID-19 and/or COVID-19 vaccine, interface dermatitis has been a prominent feature.12 Prior cases of COVID-19 vaccine-associated DM involved an onset 5 to 6 days after vaccination, considerably earlier than peak antibody response induced by the vaccine, and a longer recovery period with more aggressive treatment compared with this case.13 , 14 The timing and self-resolving nature of the DM-like reaction in the current patient may suggest transient COVID-19/COVID-19 vaccine-related autoimmunity. First, the patient’s symptoms occurred roughly 2 weeks after administration of her booster, which aligns temporally with the peak antibody response of mRNA vaccines in healthy patients.15 Second, the patient’s symptoms resolved within a short time course and with minimal treatment. Typically, to be considered a drug side-effect: (1) symptoms must be temporally related, (2) withdrawal should lead to clearance, and (3) rechallenge should induce recurrence; this case meets conditions 1 and 2 above, and any association could likely be deemed possible or probable but not certain. In summary, we present a transient DM-like reaction following COVID-19 mRNA vaccination, which may have been triggered by a type I interferon surge and then progressively resolved over several months with minimal treatment. In retrospect, the presence of immunoglobulin M deposits alone on the direct immunofluorescence skin biopsy may have provided some indication of the acuity of this process. However, the passage of time has been the most reliable indicator that this reaction has (and remains) resolved. Conflicts of interest Dr Haemel reports serving as a consultant to CSL Behring and Guidepoint. Authors Gutierrez, Connolly, and Gross have no conflicts of interest to declare. We would like to acknowledge Michelle Mintz, MD, PhD, for her careful scientific reading and thoughtful suggestions to improve this manuscript. Funding sources: None. IRB approval status: Not applicable. ==== Refs References 1 Ioannou Y. Isenberg D.A. Current evidence for the induction of autoimmune rheumatic manifestations by cytokine therapy Arthritis Rheum 43 7 2000 1431 1442 10.1002/1529-0131(200007)43:7<1431::AID-ANR3>3.0.CO;2-E 10902743 2 Turnier J.L. Kahlenberg J.M. The role of cutaneous Type I IFNs in autoimmune and autoinflammatory diseases J Immunol 205 11 2020 2941 2950 10.4049/jimmunol.2000596 33229366 3 Arkin L.M. Moon J.J. Tran J.M. From your nose to your toes: a review of severe acute respiratory syndrome coronavirus 2 pandemic‒associated pernio J Invest Dermatol 141 12 2021 2791 2796 10.1016/j.jid.2021.05.024 34561087 4 Aschman T. Stenzel W. COVID-19 associated myopathy Curr Opin Neurol 35 5 2022 622 628 10.1097/WCO.0000000000001101 35950722 5 Sprow G. Afarideh M. Dan J. Autoimmune skin disease exacerbations following COVID-19 vaccination Front Immunol 13 2022 899526 10.3389/fimmu.2022.899526 6 Woodruff M.C. Ramonell R.P. Haddad N.S. Dysregulated naive B cells and de novo autoreactivity in severe COVID-19 Nature 611 7934 2022 139 147 10.1038/s41586-022-05273-0 36044993 7 Bax C.E. Maddukuri S. Ravishankar A. Pappas-Taffer L. Werth V.P. Environmental triggers of dermatomyositis: a narrative review Ann Transl Med 9 5 2021 434-434 10.21037/atm-20-3719 8 Saberin A. Lutgen C. Humbel R.L. Hentges F. Dermatomyositis-like syndrome following acute toxoplasmosis Bull Soc Sci Med Grand Duche Luxemb 2 2 2004 109 119 9 Harland C.C. Marsden J.R. Vernon S.A. Allen B.R. Dermatomyositis responding to treatment of associated toxoplasmosis Br J Dermatol 125 1 1991 76 78 10.1111/j.1365-2133.1991.tb06046.x 1873210 10 Saud A. Naveen R. Aggarwal R. Gupta L. COVID-19 and myositis: what we know so far Curr Rheumatol Rep 23 8 2021 63 10.1007/s11926-021-01023-9 34216297 11 Wang G. Wang Q. Wang Y. Presence of anti-MDA5 antibody and its value for the clinical assessment in patients with COVID-19: a retrospective cohort study Front Immunol 12 2021 791348 10.3389/fimmu.2021.791348 12 Gonzalez D. Gupta L. Murthy V. Anti-MDA5 dermatomyositis after COVID-19 vaccination: a case-based review Rheumatol Int 42 9 2022 1629 1641 10.1007/s00296-022-05149-6 35661906 13 Wu M. Karim M. Ashinoff R. COVID-19 vaccine-associated dermatomyositis JAAD Case Rep 23 2022 58 60 10.1016/j.jdcr.2022.02.023 35252516 14 Sugimoto T. Yorishima A. Oka N. Appearance of anti-MDA5 antibody-positive dermatomyositis after COVID-19 vaccination Mod Rheumatol Case Rep 7 1 2023 108 112 10.1093/mrcr/rxac064 35950798 15 Wheeler S.E. Shurin G.V. Yost M. Differential antibody response to mRNA COVID-19 vaccines in healthy subjects Microbiol Spectr 9 1 2021 e0034121 10.1128/Spectrum.00341-21
PMC010xxxxxx/PMC10288174.txt	==== Front Intensive Crit Care Nurs Intensive Crit Care Nurs Intensive & Critical Care Nursing 0964-3397 1532-4036 Elsevier Ltd. S0964-3397(23)00083-6 10.1016/j.iccn.2023.103466 103466 Response from the Author Preventing healthcare-acquired infections in cancer patients with febrile neutropenia in intensive care units: The role of granulocyte-colony stimulating factor prophylaxis Blot Stijn ⁎ Dept. of Internal Medicine and Pediatrics, Ghent University, Ghent, Belgium Timsit Jean-Francois INSERM, IAME UMR 1137, University of Paris, France Medical and Infectious Diseases ICU, Bichat-Claude Bernard Hospital, APHP, Paris, France Zahar Jean-Ralph NSERM, IAME UMR 1137, University of Paris, France Microbiology, Infection Control Unit, GH Paris Seine Saint-Denis, APHP, Bobigny, France ⁎ Corresponding author. 23 6 2023 10 2023 23 6 2023 78 103466103466 © 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcDear Editor, We thank the authors for their interest in our review article on healthcare-associated infection in ICU patients (Blot et al. 2022) and appreciate their critical reflections (Chou et al., 2023). However, we do not fully concur with their suggestions. Firstly, we are not convinced that – in the context of intensive care – the MASCC score adds value to clinical practice (Uys et al., 2004). Patients admitted to ICUs are per definition critically ill, thereby fulfilling the criteria for iv antimicrobial therapy, close monitoring, and, if needed, appropriate organ support. The second and third point raised by the authors involve G-CSF therapy in order to prevent febrile neutropenia, and as such, infection. To our feeling, this approach is better positioned in the hematology/oncology ward, rather than the ICU. As such, this practice may have its value to prevent chemotherapy-associated febrile neutropenia or even in patients who develop febrile neutropenia. However, once the patient is admitted to the ICU, either for febrile neutropenia or another life-threatening condition, the value of initiating G-CSF remains unproven (Chousterman and Arnaud, 2018). In conclusion, we believe that the suggestions made by the authors may be of value in non-ICU settings, while their benefit in critically ill patients is mostly unclear. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References Blot S. Ruppé E. Harbarth S. Asehnoune K. Poulakou G. Luyt C.-E. Rello J. Klompas M. Depuydt P. Eckmann C. Martin-Loeches I. Povoa P. Bouadma L. Timsit J.-F. Zahar J.-R. Healthcare-associated infections in adult intensive care unit patients: Changes in epidemiology, diagnosis, prevention and contributions of new technologies Intensive Crit. Care Nurs. 70 2022 103227 Chousterman B.G. Arnaud M. Is there a role for hematopoietic growth factors during sepsis? Front. Immunol. 9 2018 1015 29977234 Chou H.-C. Tsai Y.-S. Wang Y.-T. Shueng P.-W. Hsu C.-X. Preventing healthcare-acquired infections in cancer patients with febrile neutropenia in intensive care units: The role of granulocyte-colony stimulating factor prophylaxis Intensive Crit. Care Nurs. 2023 10.1016/j.iccn.2023.103465 Uys A. Rapoport B.L. Anderson R. Febrile neutropenia: a prospective study to validate the Multinational Association of Supportive Care of Cancer (MASCC) risk-index score Support. Care Cancer 12 8 2004 555 560 15197637
PMC010xxxxxx/PMC10288175.txt	==== Front Intensive Crit Care Nurs Intensive Crit Care Nurs Intensive & Critical Care Nursing 0964-3397 1532-4036 Elsevier Ltd. S0964-3397(23)00082-4 10.1016/j.iccn.2023.103465 103465 Letter to the Editor Preventing healthcare-acquired infections in cancer patients with febrile neutropenia in intensive care units: The role of granulocyte-colony stimulating factor prophylaxis Chou Hui-Chen ab1 Tsai Yin-Shin ab Wang Yu-Ting ab Shueng Pei-Wei ac Hsu Chen-Xiong a⁎1 a Division of Radiation Oncology, Department of Radiology, Far Eastern Memorial Hospital, New Taipei City, Taiwan b Department of Nursing, Far Eastern Memorial Hospital, New Taipei City, Taiwan c School of Medicine, College of Medicine, National Yang Ming Chiao Tung University, Taipei, Taiwan ⁎ Corresponding author at: Division of Radiation Oncology, Department of Radiology, Far Eastern Memorial Hospital, No. 21, Section 2, Nanya S. Road, Banqiao District, New Taipei City 220, Taiwan. 1 Hui-Chen Chou and Chen-Xiong Hsu contributed equally to this work. 23 6 2023 10 2023 23 6 2023 78 103465103465 3 4 2023 27 5 2023 © 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcDear Editor, We read with interest the review article by Blot et al. (2022), particularly as the authors emphasize the vital role played by antimicrobial prophylaxis for adult patients experiencing immunosuppression caused by cancer, particularly during the COVID-19 pandemic. Nevertheless, there remain several issues regarding the use of granulocyte-colony stimulating factor (G-CSF) prophylaxis for patients with neutropenia in intensive care units (ICUs) that warrant further exploration. Firstly, febrile neutropenia (FN) is a critical hematological condition that requires immediate attention as the mortality rate among FN patients admitted to the ICU remains alarmingly high (Cetintepe et al., 2021). To predict mortality and determine the risk group of patients with FN, several scoring systems have been developed. Among these, the multinational association for supportive care in cancer (MASCC) scoring system is particularly useful in deciding the appropriate treatment plan for FN patients based on their age, medical history, symptoms, outpatient/inpatient status, and comorbid conditions (Uys et al., 2004). Patients who score less than 21 on the MASCC assessment are considered to be at high-risk and require hospitalization for intravenous empirical treatment, while low-risk patients may be treated with oral antibiotics following a brief hospitalization or even an outpatient visit (Cetintepe et al., 2021). Secondly, a large cohort study of 9018 patients with cancer also highlights the essential role of G-CSF prophylaxis in reducing the risk of chemotherapy-induced FN (Aagaard et al., 2020). The authors observed that patients who developed FN during their initial treatment experienced increased rates of all-cause and infectious mortality, as well as an elevated risk of ICU admissions. Additionally, the study revealed a number of factors that were linked to heightened death and ICU admission rates in patients with FN, including a positive blood culture and low lymphocyte counts, particularly during the first month following FN. These findings highlight the need for effective interventions that can reduce the incidence of FN, such as the prophylactic use of G-CSF and antibiotics, which have the potential to substantially enhance patient outcomes (Aagaard et al., 2020). Thirdly, and most importantly, updating recommendations for G-CSF prophylaxis is imperative for cancer patients with FN, as the vulnerable patient population is at significant risk of infection due to the COVID-19 pandemic (Boccia et al., 2022). In response to this pressing need, experts from cancer societies have recently revised existing guidelines and recommendations for the use of G-CSF prophylaxis in patients at intermediate risk (10–20%) of FN (Boccia et al., 2022). Such revisions are expected to reduce the incidence of FN and prevent unnecessary hospitalizations or ICU admissions during the COVID-19 pandemic. In view of these issues, recent studies have the potential to illuminate the intricate interplay between various prevention measures and risk factors associated with cancer patients who are at risk for healthcare-associated infections and ICU admission. By gaining a deeper understanding of this complex relationship, we can develop more tailored and effective prophylactic strategies for these patients, ultimately transforming the landscape of nursing practices and communication between intensive care and ambulatory care. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References: Aagaard T. Reekie J. Jørgensen M. Roen A. Daugaard G. Specht L. Sengeløv H. Mocroft A. Lundgren J. Helleberg M. Mortality and admission to intensive care units after febrile neutropenia in patients with cancer Cancer Med. 9 9 2020 3033 3042 32144897 Blot S. Ruppé E. Harbarth S. Asehnoune K. Poulakou G. Luyt C.-E. Rello J. Klompas M. Depuydt P. Eckmann C. Martin-Loeches I. Povoa P. Bouadma L. Timsit J.-F. Zahar J.-R. Healthcare-associated infections in adult intensive care unit patients: Changes in epidemiology, diagnosis, prevention and contributions of new technologies Intensive Crit. Care. Nurs. 70 2022 103227 Boccia R. Glaspy J. Crawford J. Aapro M. Chemotherapy-induced neutropenia and febrile neutropenia in the US: A beast of burden that needs to be tamed? Oncologist 27 8 2022 625 636 35552754 Cetintepe T. Cetintepe L. Solmaz S. Calık S. Ugur M.C. Gediz F. Bilgir O. Determination of the relationship between mortality and SOFA, qSOFA, MASCC scores in febrile neutropenic patients monitored in the intensive care unit Support. Care Cancer. 29 7 2021 4089 4094 33404806 Uys A. Rapoport B.L. Anderson R. Febrile neutropenia: a prospective study to validate the Multinational Association of Supportive Care of Cancer (MASCC) risk-index score Support. Care Cancer. 12 8 2004 555 560 15197637
PMC010xxxxxx/PMC10288203.txt	==== Front Lancet Reg Health Am Lancet Reg Health Am Lancet Regional Health - Americas 2667-193X Published by Elsevier Ltd. S2667-193X(23)00117-5 10.1016/j.lana.2023.100543 100543 Articles Cost-effectiveness analysis of extended thromboprophylaxis with rivaroxaban versus no prophylaxis in high-risk patients after hospitalisation for COVID-19: an economic modelling study Carvalho de Oliveira Caroline Cândida abc Agati Leandro Barile a Ribeiro Camilla Moreira a Resende Aguiar Valéria Cristina ab Caffaro Roberto Augusto c da Silva Santos Marisa d Alves Fernandes Ricardo Ribeiro d Alberto da Silva Magliano Carlos d Tafur Alfonso e Spyropoulos Alex C. f Lopes Renato Delascio g Fareed Jawed h Ramacciotti Eduardo abh∗ a Science Valley Research Institute, Santo André, São Paulo, Brazil b Hospital e Maternidade Christóvão da Gama, Grupo DASA, Santo André, São Paulo, Brazil c Irmandade da Santa Casa de Misericórdia de São Paulo, São Paulo, Brazil d Instituto Nacional de Cardiologia, Rio de Janeiro, Brazil e Northshore University Health System, Evanston, IL, USA f Zucker School of Medicine at Hofstra/Northwell and the Feinstein Institutes for Medical Research, Manhasset, NY, USA g Duke Clinical Research Institute, Duke University School of Medicine, Durham, NC, USA h Hemostasis & Thrombosis Research Laboratories at Loyola University Medical Center, Maywood, IL, USA ∗ Corresponding author. Science Valley Research Institute, Santo André, São Paulo, Brazil. 23 6 2023 8 2023 23 6 2023 24 100543100543 10 1 2023 8 6 2023 12 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background In patients at high risk of thromboembolism who were discharged after hospitalisation due to COVID-19, thromboprophylaxis with rivaroxaban 10 mg/day for 35 days significantly improved clinical outcomes, reducing thrombotic events compared with no post-discharge anticoagulation. The present study aimed to estimate the cost-effectiveness of this anticoagulation strategy. Methods Using the database of the MICHELLE trial, we developed a decision tree to estimate the cost-effectiveness of thromboprophylaxis with rivaroxaban 10 mg/day for 35 days versus no thromboprophylaxis in high-risk post-discharge patients for COVID-19 through an incremental cost-effectiveness analysis. Findings 318 patients in 14 centres in Brazil were enrolled in the primary MICHELLE trial. The mean age was 57.1 years (SD 15.2), 127 (40%) were women, 191 (60%) were men, and the mean body-mass index was 29.7 kg/m2 (SD 5.6). Rivaroxaban 10 mg per day orally for 35 days after discharge decreased the risk of events defined by the primary efficacy outcome by 67% (relative risk 0.33, 95% CI 0.12–0.90; p = 0.03). The mean cost for thromboprophylaxis with rivaroxaban was $53.37/patient, and no prophylaxis was $34.22/patient, with an incremental cost difference of $19.15. The effectiveness means obtained in the intervention group was 0.1457, while in the control group was 0.1421, determining an incremental QALY difference of 0.0036. The estimated incremental cost-effectiveness ratio (ICER) was $5385.52/QALY. Interpretation Extended treatment with Rivaroxaban as thromboprophylaxis after hospital discharge for high-risk patients with COVID-19 is a cost-effective treatment option. Funding Modest funding was provided by Science Valley Research Institute, São Paulo, Brazil. Keywords Thromboprophylaxis COVID Anticoagulation Cost-effectiveness analysis Direct oral anticoagulants Thrombosis ==== Body pmc Research in context Evidence before this study We searched MEDLINE, the Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science, and Scopus using the terms (“rivaroxaban” OR “apixaban” OR “dabigatran” OR “edoxaban” OR “heparin” OR “enoxaparin”) AND (“extended thromboprophylaxis” OR “out-of-hospital thromboprophylaxis”) AND (“SARS-CoV-2” OR “COVID” OR “coronavirus” OR “COVID-19”) AND (“randomised” OR “clinical trials”), AND (“cost-effectiveness evaluation” OR “costs”) with no date or language restrictions. We did not find published data assessing the cost-effectiveness of thromboprophylaxis after hospitalisation due to COVID-19. Added value of this study Using the database of the MICHELLE trial, a multicentre randomised trial led by our group, where a central events committee evaluated all events, we developed a decision tree to estimate the costs and effectiveness of thromboprophylaxis with rivaroxaban 10 mg/day for 35 days versus no thromboprophylaxis in high-risk post-discharge patients for COVID-19 through an incremental cost-effectiveness analysis, for the first time. The MICHELLE trial provided high-quality evidence and combined with this cost-effectiveness analysis, helps guide informed-medical decisions. Implications of all the available evidence Extended treatment with rivaroxaban as thromboprophylaxis after hospital discharge for high-risk patients with COVID-19 is a cost-effective treatment option. The findings of this study have important implications for resource prioritization and provide a comprehensive framework to inform policymakers about better decisions in public health. Introduction COVID-19 was the most important cause of hospitalisation worldwide in 2020 and 2021, surpassing circulatory and other respiratory system diseases.1 Severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) infection leads to endothelial dysfunction, platelet activation, and stasis.2 , 3 In addition, the interplay with intense inflammation improves the hypercoagulable state, increasing the risk of thromboembolism, whether venous (venous thromboembolism -VTE) or arterial (arterial thromboembolism - ATE).4 Several studies of post-discharge patients with COVID-19 demonstrated elevated incidences of symptomatic or asymptomatic events, reaching rates between 1.7% and 7.4%.5 Thromboembolic events, a composite of venous (deep-vein thrombosis [DVT], pulmonary embolism [PE]) and arterial (myocardial infarction [MI], stroke and acute limb ischemia) complications, most of the time require hospitalisation, a prolonged treatment, decrease of quality of life and different types of sequelae, depending on the site of thrombosis. The annual costs estimated in the United States range from U$7.5 billion, in the case of DVT,6 to U$ 84 billion in those with myocardial infarction.7 Other possible consequences, such as post-thrombotic syndrome (post-DVT PTS) and heart failure (post-AMI), are responsible for worsening patient quality of life and increasing costs in a broad time horizon.8 , 9 To mitigate the risk of post-discharge VTE, the MARINER trial published in 2018 evaluated the use of rivaroxaban after hospitalisation in medically ill patients. Patients with a high risk of VTE (defined by the International Medical Prevention Registry on Venous Thromboembolism [IMPROVE] score of ≥4 or 2–3 with elevated D-dimer levels) at hospital discharge were randomly assigned to rivaroxaban 10 mg/day (or 7.5 mg/day if creatinine clearance <50 mL/min) versus placebo for 45 days after hospital discharge. Despite not achieving superiority on its primary outcome (a combination of symptomatic VTE and death due to VTE), this prophylactic anticoagulation strategy led to a 28% relative risk reduction in major and fatal thromboembolic events and a 27% relative risk reduction of symptomatic venous thromboembolism and all-cause death, reducing the global burden of death and disability from VTE.10 The MICHELLE trial published in 2021 was the first study on prophylactic anticoagulation after discharge in patients with COVID-19. In this open-label, multicentre, randomised trial (with blinded adjudication), patients hospitalised with COVID-19 at increased risk for venous thromboembolism with the International Medical Prevention Registry on Venous Thromboembolism [IMPROVE] venous thromboembolism [VTE] score of ≥4 or 2–3 with a D-dimer >500 ng/mL were randomly assigned to receive, at hospital discharge, rivaroxaban 10 mg/day or no anticoagulation for 35 days. The primary efficacy outcome was a composite of symptomatic or fatal venous thromboembolism, asymptomatic venous thromboembolism on bilateral lower-limb venous ultrasound and CT pulmonary angiogram, symptomatic arterial thromboembolism, and cardiovascular death at day 35. The primary safety outcome was major bleeding. The primary and safety analyses were carried out in the intention-to-treat population. This study led by our group demonstrated that oral rivaroxaban 10 mg per day for 35 days after discharge in patients hospitalised by COVID-19 at high risk of thromboembolism decreased the risk of events defined by the primary efficacy outcome by 67% (relative risk 0.33, 95% CI 0.12–0.90; p = 0.03). There was no statistically significant increase in the rate of minor or major bleeding.11 This current study aims to estimate the cost-effectiveness of thromboprophylaxis with rivaroxaban versus no thromboprophylaxis in high-risk patients after hospitalisation for COVID-19 from the perspective of the Brazilian public health system. Methods Model assumptions Using the database of the MICHELLE trial, a decision tree was developed to estimate the cost-effectiveness of thromboprophylaxis with oral rivaroxaban 10 mg/day for 35 days versus no thromboprophylaxis in high-risk patients post-discharge for COVID-19 through an incremental cost-effectiveness analysis. It is a unidirectional flow of events followed by different outcomes. It ends with a terminal event in which the patient may have total or partial recovery of health or death.12 At the beginning of the decision tree, patients could receive rivaroxaban or no intervention after hospital discharge. The proportion of all the thromboembolic events, such as DVT, PE, MI, stroke, and limb acute ischemia, was evaluated in the two groups. Each group had three possibilities: no event, thromboembolic event occurrence, or death. Given the low risk of significant clinical events with rivaroxaban, which would result in irrelevant impacts on both costs and effectiveness, the bleeding rate and allergic reactions were not applied in the model. This is a conservative model, with an immediate post-discharge horizon timeline of two months, because we were focused on acute adverse outcome events and their implications. It was the same follow-up time pre-specified for the MICHELLE trial. No discount rate was applied since the model has a horizon timeline of less than one year. Treeage Pro™ software 2022 (www.treeage.com) was used for the analysis. The model is demonstrated in Fig. 1 .Fig. 1 Decision tree (PE, pulmonary embolism; MI, myocardial infarction; DVT, deep-vein thrombosis). Efficacy data DVT, PE, MI, and acute limb ischemia risk ratios were extracted from the raw database of the MICHELLE trial. There was no ischemic stroke in the outcomes of that trial. All relative risks are demonstrated in Table 1 .Table 1 Thromboembolic events: relative risks. Relative Risk (95% CI) Source DVT symptomatic 0.14 (0.01–2.69) Ramacciotti et al.11 DVT asymptomatic 1.96 (0.18–21.40) Ramacciotti et al.11 PE symptomatic 0.49 (0.04–5.35) Ramacciotti et al.11 PE asymptomatic 0.25 (0.03–2.17) Ramacciotti et al.11 Fatal PE 0.14 (0.01–2.69) Ramacciotti et al.11 Myocardial infarction 0.33 (0.01–7.96) Ramacciotti et al.11 Acute limb ischemia 0.33 (0.01–7.96) Ramacciotti et al.11 Stroke N/A Ramacciotti et al.11 Medical costs The model was constructed from the Brazilian Public Health System (SUS) perspective. Only direct medical costs associated with treatment during hospitalisation for each event were evaluated. Outpatient costs were not considered in the analysis. Cost studies performed in national hospitals were consulted, and federal databases informed expenses related to each type of event.13 All costs are demonstrated in Table 2 .13 , 16 , 19 Table 2 Thromboembolic events: utility, mortality rate, and mean of costs. Utility Mortality rate (%)13 Mean of costs (USD) DVT 0.8414 2.95 151.2413 PE 0.6314 18.87 375.7913 MI 0.7615 9.68 982.3616 Acute limb ischemia 0.2817 9.21 468.6813 Stroke 0.4418 16.23 2860.2119 The price of rivaroxaban 10 mg was extracted from the List of Maximum Drug Prices from CMED-ANVISA (Health Surveillance National Agency).20 A Brazilian state excise tax (ICMS) of 18% was considered. The thromboprophylaxis with rivaroxaban 10 mg/daily for 35 days represented an expense of U$ 45.92. American Dollar was considered currency and converted in 2022 with a conversion rate of 1 USD = 5.33 Brazilian Real (BRL). Utilities and mortality rate The effectiveness of the model was measured in quality-adjusted life-years (QALYs). They are calculated by estimating the years of life for a patient after intervention and weighting each year with a utility score, which varies between 1 (perfect health) and 0 (dead). The average Brazilian utility is 0.88. We consider it for all patients with no events.21 For those who had events, we used utility studies from medical literature focused on national studies based on EQ-5D22 or SF-3623 questionnaires applied to patients who had a thromboembolic event in the last two months, the same time horizon chosen in our study. There is a significant decrease in the quality of life in conditions such as stroke18 and acute limb ischemia.17 All utilities are reported in Table 2.14 , 15 , 17 , 18 The mortality rates were obtained from the information system database on mortality from the Brazil Ministry of Health (Table 2).13 Analyses Incremental cost-effectiveness ratios (ICER) were calculated as the difference in costs in the rivaroxaban group minus control divided by the difference in health outcomes in both groups. The model's outcome was each comparator's quality-adjusted life-years (QALYs) and the cost per QALY gained. All cost-effectiveness analyses were developed using the Treeage Pro™ software 2022. The reporting in this study follows Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) statement (Supplementary Table S3).24 This study was approved by the local Institutional Review Board (IRB), CAAE 52371121.7.0000.5485. Role of funding source Science Valley Research Institute, São Paulo, Brazil, provided modest funding to access the MICHELLE Trial database. The authors have received no funding to conduct the study. Science Valley Research Institute had no role in study design, data collection, analysis, or interpretation, in the report's writing or in the decision to submit the paper for publication. Results Three hundred and eighteen patients in 14 centres in Brazil were enrolled in the primary MICHELLE trial. The mean age was 57.1 years (SD 15.2), 127 (40%) were women, 191 (60%) were men, and the mean body-mass index was 29.7 kg/m2 (SD 5.6). Rivaroxaban 10 mg per day orally for 35 days after discharge decreased the risk of events defined by the primary efficacy outcome by 67% (relative risk 0.33, 95% CI 0.12–0.90; p = 0.03). For a period of 2-month time frame, the mean cost for thromboprophylaxis with rivaroxaban was $53.37/patient, and no prophylaxis was $34.22/patient, with an incremental cost difference of $19.15. The effectiveness means obtained in the intervention group was 0.1457, while in the control group was 0.1421, determining an incremental QALY difference of 0.0036. The estimated ICER was $5385.52/QALY. Further details on the cost-effectiveness results can be found in Supplementary Table S1. The one-way deterministic sensitivity analysis was performed to assess the parameter variations. The results are disclosed in Tornado Diagram (Fig. 2 ), which demonstrates how variations in each variable affect the ICER. The bars are distributed in decreasing order of width, showing that those at the top have the most significant effect on the ICER. In contrast, variations in variables near the bottom have relatively little impact on the ICER.Fig. 2 One-way sensitivity analysis - Tornado diagram. (Price and cost in USD. PE, pulmonary embolism; MI, myocardial infarction; DVT, deep-vein thrombosis; ICER, incremental cost-effectiveness ratio; CI, confidence interval). Red bars are the ICER patterns when the parameter values increase. Blue bars are the ICER response when the parameter values decrease. The variable that most impacts directly the model is rivaroxaban's price, followed by hospitalisation costs with PE and stroke. All variables are represented with their respective confidence interval in brackets. No variable at the confidence interval's lower or upper possible values extrapolated $15,965.25. Further details on the values, ranges, and confidence intervals can be found in Supplementary Table S2. Discussion There was no consensus on anticoagulation after discharge at the beginning of the COVID-19 pandemic, and the costs of unnecessary anticoagulant prescriptions may have been very high, or the absence of that may have led to several thromboembolic preventable events. The MICHELLE trial demonstrated that rivaroxaban is effective and safe as thromboprophylaxis in high-risk patients after hospitalisation for COVID-19 and became the standard to guide thromboprophylaxis in those patients.11 A meta-analysis published in November 2022, including more than 10,000 patients, demonstrated that extended thromboprophylaxis was associated with a statistically significant reduced composite endpoint of thrombosis and all-cause mortality in patients with COVID-19 after discharge (OR 0.52; 95% CI: 0.41–0.67, p = 0.0001). Extended anticoagulant therapy was not associated with a significant increase in serious bleeding events (OR: 1.64; 95% CI: 0.95–2.82, p = 0.07), supporting the clinical benefit of post-hospitalisation thromboprophylaxis in selected COVID-19 patients.25 To evaluate the cost-effectiveness of this therapy, the incremental cost-effectiveness analysis technique was chosen because we analysed only acute thromboembolic events after hospitalisation. The incremental cost-effectiveness ratio of $5385.52/QALY demonstrated its cost-effectiveness. It means a gain of one year with the quality of life per $5385.52 expense. There are multiple cost-effectiveness analyses related to COVID-19 vaccination and treatments with remdesivir or monoclonal antibodies in the literature.26 Still, we found no studies addressing anticoagulation cost-effectiveness during or after hospitalisation. There is also a lack of DVT, PE, and limb acute ischemia costs data worldwide, including Brazil, where the MICHELLE trial was conducted. However, all data related to direct medical costs and government expenses in Brazil are published monthly on a public government website (DATASUS),13 becoming a reliable data source. Only acute events were assessed in a deterministic analysis because the horizon time was determined for two months post-discharge. Cost-effectiveness analysis takes into consideration a balance. The Tornado diagram (Fig. 2) discloses costs for the variables, and at the top of the chart, pulmonary embolism risk was the most relevant variable, followed by rivaroxaban cost. A higher relative risk of PE in the no prophylaxis group decreases the ICER. If PE were rare, it would increase the cost of PE prophylaxis, increasing ICER. The same rationale can apply to all thrombotic outcomes. In addition, the more expensive rivaroxaban, the higher the ICER. The ICER limit in Brazil is considered BRL 40,000.00/QALY (USD 7504.69/QALY), but in a pandemic scenario, the limit increases to BRL 120,000.00/QALY (USD 22,514.00/QALY). In the United States, the cost threshold is $50,000 to $150,000; in the UK, the most used threshold is £30,000 by QALY.27 In our model analysis, based on the Brazilian perspective, we met an ICER of $5385.52/QALY, which means using rivaroxaban post-discharge in patients hospitalised by COVID-19 is highly cost-effective, considering the increased limit of ICER due to pandemic. Rivaroxaban price is a variable that significantly affects the model. If the price of rivaroxaban were $68.67 (the highest variation), the ICER would be $11,822.81/QALY. Rivaroxaban's price in Brazil was extracted from the ANVISA list and cost U$1.30 for each pill, comparatively cheaper than other countries, such as the USA. Furthermore, the price may be even lower in clinical practice, lowering the ICER. If the treatment price were $22.89, with complete exemption from taxes, for example, using rivaroxaban would be cost-saving. No upper bound of all variables exceeded $15,965.10/QALY on ICER, showing that the model remains cost-effective even in the worst-case scenario. Choosing a conservative model with a short time horizon generates some limitations inherent to the model, such as only the analysis of acute complications. Significant sequelae with high cost, i.e., heart failure after MI and stroke sequelae, were not analysed. If the horizon were more prolonged, it would impact the quality of life lost and costs due to chronic effects such as post-PE illness, heart failure, and post-thrombotic syndrome. The model would be more cost-effective and may become cost-saving. Furthermore, different events could happen in the same patient (i.e., DVT plus MI), but we did not model the impact of this possible occurrence. According to Goldin et al., that validated the IMPROVE inpatients with COVID-19, approximately 45.7% of patients are classified as high-risk.28 Considering that the number of hospitalisation caused by COVID-19 in Brazil was 1,564,842 since the beginning of the pandemic until September/2022,29 around 715,133 patients would be classified as high-risk and considering the risk relative of 0.33, observed on the MICHELLE trial, almost 480,000 would have benefited from rivaroxaban use, saving expenses with thromboembolic complications and improving the quality of life. In conclusion, the extended treatment with rivaroxaban as thromboprophylaxis after hospital discharge for high-risk patients with COVID-19 is a cost-effective treatment option. The findings of this study have important implications for resource prioritization and provide a comprehensive framework to inform policymakers about better decisions in public health. Contributors CCCO, ER, and LBA conceived the trial and wrote the initial proposal. Literature search, figures, data analysis, data interpretation, and writing were performed by CCCO, MSS and RRAF. CMR, ER, VCRA, RAC, CASM, AT, ACS, JF, and RDL contributed to data interpretation, manuscript writing, review, and editing. All authors had direct access to all the data, significantly contributed to the manuscript and agreed to submit it for publication. CCCO and ER directly accessed and verified the underlying data reported in the manuscript. Data sharing statement All data used in this study are openly accessible and available through the sources listed in Table 1, Table 2. Declaration of interests ER reports grants and consulting fees from Bayer and Pfizer; grants from the Brazilian Ministry of Science and Technology; and personal fees from Aspen Pharma, Biomm Pharma, and Daiichi-Sankyo, outside the submitted work. LBA reports grants from Bayer, Pfizer and the Brazilian Ministry of Science and Technology. ACS reports consulting fees from Janssen Research & Development LLC, Bayer, Portola, Boehringer Ingelheim, Bristol-Meyers Squibb, ATLAS group and grants from Janssen and Boehringer Ingelheim. AT reports personal fees from Janssen and Recovery Force and grants from Bio Tap, Idorsia, Bristol-Myers Squibb, Novo Nordisk, Janssen, and Doasense. RDL reports grants and personal fees from Bristol-Myers Squibb, Pfizer, GlaxoSmithKline, Medtronic PLC, and Sanofi; and personal fees from Amgen, Bayer, and Boehringer Ingelheim outside the submitted work. No grants from pharmaceutical companies developing or manufacturing rivaroxaban were involved in this study. Appendix A Supplementary data Supplementary Tables S1, S2, and Cheers checklist Acknowledgements We would like to thank Science Valley Research Institute São Paulo, Brazil, for its contribution with modest funding to help access and review The MICHELLE trial database. No other funding was provided. This article has not been presented anywhere. This article has not been published. Appendix A Supplementary data related to this article can be found at https://doi.org/10.1016/j.lana.2023.100543. ==== Refs References 1 Zimmermann I.R. Sanchez M.N. Alves L.C. COVID-19 as the leading cause of hospital deaths in the Brazilian public health system in 2020 Int J Infect Dis 113 2021 162 165 34607016 2 Clerkin K.J. Fried J.A. Raikhelkar J. COVID-19 and cardiovascular disease Circulation 141 20 2020 1648 1655 32200663 3 Zheng Y.Y. Ma Y.T. Zhang J.Y. Xie X. COVID-19 and the cardiovascular system Nat Rev Cardiol 17 5 2020 259 260 32139904 4 Li P. Zhao W. Kaatz S. Latack K. Schultz L. Poisson L. Factors associated with risk of postdischarge thrombosis in patients with COVID-19 JAMA Netw Open 4 11 2021 e2135397 10.1001/jamanetworkopen.2021.35397 5 Giannis D. Allen S.L. Tsang J. Postdischarge thromboembolic outcomes and mortality of hospitalized patients with COVID-19: the CORE-19 registry Blood 137 20 2021 2838 2847 33824972 6 Mahan C.E. Holdsworth M.T. Welch S.M. Borrego M. Spyropoulos A.C. Deep-vein thrombosis: a United States cost model for a preventable and costly adverse event Thromb Haemostasis 106 3 2011 405 415 21833446 7 Bishu K.G. Lekoubou A. Kirkland E. Estimating the economic burden of acute myocardial infarction in the US: 12 Year national data Am J Med Sci 359 5 2020 257 265 32265010 8 Ashrani A.A. Heit J.A. Incidence and cost burden of post-thrombotic syndrome J Thromb Thrombolysis 28 4 2009 465 476 19224134 9 Urbich M. Globe G. Pantiri K. A systematic review of medical costs associated with heart failure in the USA (2014-2020) Pharmacoeconomics 38 11 2020 1219 1236 32812149 10 Spyropoulos A.C. Ageno W. Albers G.W. Rivaroxaban for thromboprophylaxis after hospitalization for medical illness N Engl J Med 379 12 2018 1118 1127 30145946 11 Ramacciotti E. Barile Agati L. Calderaro D. Rivaroxaban versus no anticoagulation for post-discharge thromboprophylaxis after hospitalisation for COVID-19 (MICHELLE): an open-label, multicentre, randomised, controlled trial Lancet (London, England) 399 10319 2022 50 59 34921756 12 Gupta N. Verma R. Dhiman R.K. Rajsekhar K. Prinja S. Cost-effectiveness analysis and decision modelling: a tutorial for clinicians J Clin Exp Hepatol 10 2 2020 177 184 32189934 13 DATASUS http://tabnet.datasus.gov.br/cgi/tabcgi.exe?sih/cnv/niuf.def [In pt_BR)] 14 Locadia M. Bossuyt P.M.M. Stalmeier P.F.M. Treatment of venous thromboembolism with vitamin K antagonists: patients' health state valuations and treatment preferences Thromb Haemostasis 92 6 2004 1336 1341 15583742 15 Brandão S.M.G. Rezende P.C. Rocca H.-P.B.-L. Comparative cost-effectiveness of surgery, angioplasty, or medical therapy in patients with multivessel coronary artery disease: MASS II trial Cost Eff Resour Allocation 16 1 2018 55 16 Teich V. Piha T. Fahham L. Acute coronary syndrome treatment costs from the perspective of the supplementary health system Arq Bras Cardiol 105 4 2015 339 344 26559980 17 Adam D.J. Beard J.D. Cleveland T. Bypass versus angioplasty in severe ischaemia of the leg (BASIL): multicentre, randomised controlled trial Lancet (London, England) 366 9501 2005 1925 1934 16325694 18 Marinho C. Monteiro M. Santos L. Oliveira-Filho J. Pinto E.B. Gait performance and quality of life in stroke survivors: a cross-sectional study Revista Pesquisa em Fisioterapia 8 1 2018 79 87 19 Safanelli J. Vieira L.G.D.R. de Araujo T. The cost of stroke in a public hospital in Brazil: a one-year prospective study Arq Neuro Psiquiatr 77 6 2019 404 411 10.1590/0004-282X20190059 20 ANVISA Medicines market regulation chamber-ANVISA CMED 2020 Agência Nacional de Vigilância Sanitária - Anvisa 787 https://wwwgovbr/anvisa/pt-br/pagina-inicial 21 Santos M. Cintra M.A. Monteiro A.L. Brazilian valuation of EQ-5D-3L health states: results from a saturation study Med Decis Making 36 2 2016 253 263 26492896 22 Balestroni G. Bertolotti G. EuroQol-5D (EQ-5D): an instrument for measuring quality of life Monaldi Archives for Chest Disease = Archivio Monaldi Per Le Malattie Del Torace 78 3 2012 155 159 23614330 23 Ware J. Ma K. Keller S.D. SF-36 physical and mental health summary scales: a user’s manual 8 1993 Health Assessment Lab Boston, MA 23 28 24 Husereau D. Drummond M. Augustovski F. Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) statement: updated reporting guidance for health economic evaluations BJOG 129 3 2022 336 344 35014160 25 Dai M.F. Xin W.X. Kong S. Ding H.Y. Fang L. Effectiveness and safety of extended thromboprophylaxis in post-discharge patients with COVID-19: a systematic review and meta-analysis Thromb Res 221 2023 105 112 36502592 26 Fiolet T. Kherabi Y. MacDonald C.-J. Ghosn J. Peiffer-Smadja N. Comparing COVID-19 vaccines for their characteristics, efficacy and effectiveness against SARS-CoV-2 and variants of concern: a narrative review Clin Microbiol Infect 28 2 2022 202 221 34715347 27 Neumann P.J. Cohen J.T. Weinstein M.C. Updating cost-effectiveness--the curious resilience of the $50,000-per-QALY threshold N Engl J Med 371 9 2014 796 797 25162885 28 Goldin M. Lin S.K. Kohn N. External validation of the IMPROVE-DD risk assessment model for venous thromboembolism among inpatients with COVID-19 J Thromb Thrombolysis 52 4 2021 1032 1035 34146235 29 Kenakin T. New lives for seven transmembrane receptors as Drug targets Trends Pharmacol Sci 36 11 2015 705 706 26482172
PMC010xxxxxx/PMC10288307.txt	==== Front J Geriatr Oncol J Geriatr Oncol Journal of Geriatric Oncology 1879-4068 1879-4076 Elsevier Ltd. S1879-4068(23)00161-3 10.1016/j.jgo.2023.101564 101564 Perspectives SIOG COVID-19 working group recommendations on COVID-19 therapeutic approaches in older adults with cancer Russo Chiara a⁎ Mislang Anna Rachelle b Ferraioli Domenico c Soto-Perez-de-Celis Enrique d Colloca Giuseppe e Williams Grant R. f O'Hanlon Shane g Cooper Lisa h O'Donovan Anita i Audisio Riccardo A. j Cheung Kwok-Leung k Sarrió Regina Gironés l Stauder Reinhard m Jaklitsch Michael n Cairo Clarito Jr o Gil Luiz Antonio Jr p Sattar Schroder q Kantilal Kumud r Loh Kah Poh s Lichtman Stuart M. t Brain Etienne u Kanesvaran Ravindran v Battisti Nicolò Matteo Luca w a Department of Medical Oncology, Léon Bérard, Comprehensive Cancer Centre, Lyon, France b Department of Medical Oncology, Flinders Centre for Innovation in Cancer, College of Medicine and Public Health, Flinders University, Bedford Park, SA 5042, Australia c Department of Gynaecology, Léon Bérard, Comprehensive Cancer Centre, Lyon, France d Department of Geriatrics, Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Mexico e Dipartimento di Diagnostica per Immagini, Radioterapia Oncologica ed Ematologia, Fondazione Policlinico Universitario A. Gemelli IRCCS, Rome, Italy f Institute for Cancer Outcomes and Survivorship, University of Alabama at Birmingham School of Medicine, Birmingham, AL, USA g University College Dublin, St Vincent's University Hospital, Dublin, Ireland h Department of Geriatric Medicine, Rabin Medical Center, Sackler Faculty of Medicine, Division of Aging, Department of Medicine, Tel Aviv University, Israel i Applied Radiation Therapy Trinity (ARTT), Trinity St James's Cancer Institute, Trinity College, Dublin, Ireland j Department of Surgery, Sahlgrenska Academy - University of Gothenburg, Gothenburg, Sweden k School of Medicine, University of Nottingham, Royal Derby Hospital Centre, Derby, UK l Department of Medical Oncology, Hospital Universitari i Politècnic La FE, Valencia, Spain m Department of Internal Medicine V (Haematology and Oncology), Innsbruck Medical University, Innsbruck, Austria n Brigham and Women's Hospital – Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA o National Integrated Cancer Control Program, Department of Health, Manila, Philippines p Geriatric Center for Advanced Medicine - Hospital Sirio-Libanês, São Paulo, SP, Brazil q College of Nursing – University of Saskatchewan, Saskatoon, Canada r School of Pharmacy, University of East Anglia, Norwich, UK s University of Rochester Medical Center, Division of Hematology/Oncology, Department of Medicine, James P. Wilmot Cancer Institute, Rochester, NY, USA t Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY, USA u Department of General Medical Oncology, University Hospitals Leuven, Leuven, Belgium v Division of Medical Oncology, National Cancer Centre Singapore, Singapore w Breast Unit – Department of Medicine Department, The Royal Marsden NHS Foundation Trus, Breast Cancer Research Division, The Institute of Cancer Research, London, UK ⁎ Corresponding author at: Department of Medical Oncology, Léon Bérard, Comprehensive Cancer Centre, Lyon, France. 23 6 2023 23 6 2023 10156414 11 2022 23 1 2023 21 6 2023 © 2023 Elsevier Ltd. All rights reserved. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Keywords Cancer COVID-19 SARS-CoV-2 Older patients Therapeutics Mortality ==== Body pmc
PMC010xxxxxx/PMC10288308.txt	==== Front Sleep Med Clin Sleep Med Clin Sleep Medicine Clinics 1556-407X 1556-4088 Published by Elsevier Inc. S1556-407X(23)00057-7 10.1016/j.jsmc.2023.06.012 Article COVID-19: a challenge to the safety of assisted reproduction Samama Marise MD, PhD 12∗ Entezami Frida MD, MSc 3 Rosa Daniela S. PhD 5 Sartor Amanda 24 Piscopo Rita C.C.P. MD 2 Andersen Monica L. PhD 4 Cunha-Filho Joao Sabino MD, PhD 6 Jarmy-Di Bella Zsuzsanna I.K. MD, PhD 1 1 Department of Gynecology, Federal University of São Paulo-Brazil; 2 GERA Institute of Reproductive Medicine, São Paulo-Brazil; 3 American Hospital of Paris, IVF Unit, Neuilly-Sur-Seine-France; 4 Department of Psychobiology, Federal University of São Paulo-Brazil; 5 Department of Microbiology, Immunology and Parasitology, Federal University of São Paulo-Brazil; 6 Faculdade de Medicina, Universidade Federal do Rio Grande do Sul, Porto Alegre-Brazil ∗ Corresponding Author: Marise Samama Rua Teodoro Sampaio 352/117 São Paulo, 05406-000, Brazil Tel: 55 11 991339026 23 6 2023 23 6 2023 © 2023 Published by Elsevier Inc. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background There is an increased risk of becoming pregnant through fertility treatments using assisted reproductive technology (ART) during the COVID-19 pandemic. The aim of this review is to gather comprehensive data from the existing literature on the potential risks of fertility management during the pandemic period, and outline strategies to mitigate them, with a focus on the hormonal and surgical procedures of ART. Methods A comprehensive search of the scientific literature on COVID-19 in relation to fertility was conducted in the PubMed database using the keywords “coronavirus”, “COVID-19”, “SARS-CoV-2” and “pregnancy”, “fertility”, “urogenital system”, “vertical transmission”, “assisted human reproduction”, “controlled ovarian stimulation”, “oocyte retrieval”, “in vitro fertilization”, “hormones”, “surgical procedures”, “embryos”, “oocytes”, “sperm”, “semen”, “ovary”, “testis”, “ACE-2 receptor”, “immunology”, “cytokine storm”, and “coagulation”, from January 2020-July 2022. Findings Published data on pregnancy and COVID-19, and the interaction of the urogenital system and SARS-CoV-2 is reported. The immunological and prothrombotic profiles of COVID-19 patients, and their increased risks from controlled ovarian stimulation (COS) and ART surgeries, and how these procedures could facilitate COVID-19 and/or contribute to the severity of the disease by enhancing the cytokine storm are summarized. Strategies to prevent complications during COS that could increase the risks of the disease in pre-symptomatic patients are considered. Conclusions: The impact of SARS-CoV-2 on pre-symptomatic infertile patients presents a challenge to find ways to avoid the increased hormonal, immunological, and prothrombotic risks presented by the use of COS in ART protocols during the COVID-19 outbreak. Safe ART procedures and recommendations are highlighted. Keywords Pandemic COVID-19 ovarian stimulation human IVF cytokine storm assisted reproductive technology sleep immune system ==== Body pmcThe authors have nothing to disclose.
PMC010xxxxxx/PMC10288309.txt	==== Front Mar Policy Mar Policy Marine Policy 0308-597X 1872-9460 Published by Elsevier Ltd. S0308-597X(23)00263-4 10.1016/j.marpol.2023.105730 105730 Full Length Article Fatigue at sea during and after the COVID-19 pandemic: A comparative study of two matched samples of seafarers Zhao Zhiwei ab⁎ Tang Lijun c Ma Yunlei b Wu Yueyan d Lin Shiqi b Wu Zefan b Zhou Zheng b Wang Xinyi d Wang Xinjian ef a Seafarers Research Institute, Dalian Maritime University b Transportation Engineering College, Dalian Maritime University, No. 1 Linghai Road, Dalian, China. 116026 c Plymouth Business School, University of Plymouth, Cookworthy Building, Drake Circus, Plymouth PL4 8AA, UK d College Of Sciences, Dalian Maritime University, No. 1 Linghai Road, Dalian, China. 116026 e Navigation College, Dalian Maritime University f Key Laboratory of Navigation Safety Guarantee of Liaoning Province, No. 1 Linghai Road, Dalian, China. 116026 ⁎ Corresponding author at: Seafarers Research Institute, Dalian Maritime University 23 6 2023 23 6 2023 10573019 5 2023 14 6 2023 15 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. This paper examines seafarers’ experience of fatigue during and after the pandemic. A multi-phase mixed methods research design was used, including two quantitative surveys (Nduring-pandemic=501 and Nafter-pandemic=412) and 36 in-depth interviews. Applying propensity score matching the two samples to approximate the conditions of a randomized controlled experiment, the study shows that surprisingly seafarers reported higher levels of fatigue after the pandemic. Qualitative interviews with seafarers and ship managers reveal the underlying reason – the intensified ship inspection regime together with policy and regulatory updates/revisions in the immediate aftermath of the pandemic increased seafarers’ workload and made seafarers more fatigued. The results of the two surveys also show that while fatigue risk factors differed between the two periods, fatigue risk can be managed and mitigated in both periods by implementing fatigue risk management policies and practices. Policy and management implications for improving seafarers’ occupational health and safety are discussed at the end of the paper. Keywords COVID-19 pandemic fatigue mixed methods occupational health and safety paradox theory safety inspection ==== Body pmc1 Introduction Fatigue is a major contributing factor to workplace and transport-related accidents, injuries and fatalities [32]. It has long been a safety concern in the shipping industry [12], [25], [30], [36], [5], [8]. Fatigue is caused by a number of factors, such as long working hours, high work demands, poor sleep, long periods of service on board, and poor working conditions [12], [26], [9]. During the COVID-19 pandemic, the uncertainties associated with crew changes, lack of shore leave, and the risk of contracting the virus resulted in mental health issues and worsened the problem of fatigue among seafarers at sea [6]. Unsurprisingly, the Seafarers’ Happiness Index showed that seafarers’ happiness in early 2022 fell to its lowest point of 5.85 (out of 10) since it was launched in 2015 [27]. Fortunately, 2022 also saw the gradual lifting of pandemic restrictions around the world, signalling post-pandemic normality. For seafarers, especially towards the end of 2022, shore leave and access to welfare facilities ashore were steadily restored, and the certainty and stability of crew changes returned. By the end of 2022, their happiness level had risen to 7.69, even higher than the pre-pandemic level in 2019 [27]. This increase may reflect the removal of pandemic restrictions and the associated fatigue factors, and thus it is reasonable to assume that the removal would lead to a reduction in seafarers’ fatigue levels. Nevertheless, this assumption needs to be validated empirically. In addition, it is unknown whether the transition to the post-pandemic era has any unexpected effects on seafarers’ fatigue. It is well known that tensions are inherent in organisations due to multiple and conflicting demands [21]. In fact, fatigue can be seen to reflect the tension between cost/profit and safety [26], [33], [4], which can be explained by paradox theory [21]. According to this theory, tensions are inherent in organisations because they have multiple and often incompatible goals to achieve. Furthermore, it argues that changes and transitions can bring latent tensions to the fore and create the need to manage them. As such, periods of change can provide a magnifying lens through which organisational tensions and the related management issues can be explored. While there is a large body of literature on the impact of the pandemic on seafarers, the post-pandemic transition has not been studied. Certainly the transition has ended the restrictions and is welcome as evidenced by the happiness index, but it should not be taken for granted that the transition would be problem-free. In this context, this paper reports and discusses the results of a comparative study of Chinese seafarers’ experiences of fatigue based on two surveys conducted in two phases – during and after the pandemic. To compare self-reported fatigue levels, data from the two samples (Sample A during the pandemic and Sample B after the pandemic) were matched based on respondent and work-related characteristics using propensity scores. In addition, factors associated with fatigue in the two samples were analysed separately. To ensure a robust interpretation of the survey results, qualitative interviews with seafarers and ship managers were conducted following the surveys. This mixed-methods and comparative approach helps to uncover unexpected fatigue issues at the beginning of the post-pandemic era. 2 Fatigue and the pandemic in the maritime context According to the International Maritime Organisation [11], fatigue can be defined as A state of physical and/or mental impairment resulting from factors such as inadequate sleep, extended wakefulness, work/rest requirements out of sync with circadian rhythms and physical, mental or emotional exertion that can impair alertness and the ability to safely operate a ship or perform safety-related duties. This definition reflects the most up-to-date knowledge of fatigue in the maritime domain and enumerates the key contributing factors identified in the research literature [12], [8]. The general fatigue literature tends to highlight three key factors [32], [9]: 1) homeostatic factors (i.e. poor sleep quality and/or poor sleep quantity), 2) circadian factors (e.g. shift work patterns – night shift work is likely to cause fatigue as it disrupts circadian rhythms and negatively affects sleep, and 3) task-related factors (i.e. long working hours, high workload, and work intensification). In the maritime domain, additional factors related to the working environment have also been identified [12], [8]. These include psycho-social stressors (such as separation from family, loneliness on board, multi-national crews, and limited recreational activities), physical work environment factors (such as noise, vibration, ship motion, and light in the cabin, which inevitably affect sleep and thus cause fatigue), and prolonged service which leads to longer exposure to fatigue-related factors [16]. Recognising that tensions are inherent in organisations as they have multiple and often incompatible goals to achieve, Smith and Lewis [21] develop a theory of paradox. According to this theory, multiple demands, times of change and limited resources are likely to create tensions and/or bring latent tensions to the surface. In line with this theory, the occupational health and safety literature suggests that the tension between investing in occupational health and safety and making a profit is even present, and has identified this tension as a major underlying factor in workplace injuries [14], [31] and fatigue in the maritime industry [26], [33], [4], [5]. To achieve long-term development, Smith and Lewis’ [21] paradox theory suggests that rather than denying tensions, organisations need to accept and manage them effectively. In the maritime industry, a number of international regulations have been adopted, requiring shipping companies to invest and develop capacity in fatigue management. Both the International Convention on Standards of Training, Certification and Watchkeeping for Seafarers (STCW) adopted by the IMO and the Maritime Labour Convention (MLC) adopted by the International Labour Organisation (ILO) set limits on seafarers’ work and rest hours. However, as fatigue is caused by multiple factors, it is suggested that hours of service regulations are not sufficient and that a fatigue management system approach should be adopted [9]. The system approach involves a fatigue management policy, education and awareness training, fatigue level monitoring, fatigue reporting mechanisms, and fatigue incident reporting and investigation procedures. Indeed, the International Safety Management (ISM) Code adopted by the IMO requires shipping companies to develop safety management systems (SMS) to manage safety (including fatigue) in a systematic manner. The recent COVID-19 pandemic had a prolonged impact on the maritime industry [1], [22], [23], [34], and posed challenges for fatigue management as it caused a crew change crisis and shore leave restrictions for seafarers. During the pandemic, qualitative studies [13], [19] showed that, the uncertainties associated with crew change caused stress, anxiety, depression, fatigue and even suicidal thoughts among seafarers at sea. Similarly, quantitative studies showed that a large number of seafarers suffered mental health problems and fatigue [10], [17]. For example, Hebbar and Mukesh [10] surveyed 288 seafarers, of whom 40 percent felt unhappy, 30 percent endured stress and over 15 percent felt completely fatigued. Fatigue was reported being associated with poor sleep. In [18]) survey of 72 seafarers during their ship’s call at the Port of Trieste, 30 percent of participants reported suffering from insomnia to the extent that they were concerned. Similar findings were also reported by other researchers (Kaptan & Kaptan, 2021; [19]). As such, mental health problems, such as anxiety and depression, which were exacerbated by the pandemic [16], [3], had a negative impact on seafarers’ sleep quality. Furthermore, the pandemic resulted in extended services for many seafarers and increased workload due to additional tasks such as disinfection [17], [35]. All these factors contributed to fatigue during the pandemic. As noted above, pandemic-related restrictions were largely lifted in late 2022. With crew changes, shore leave and access to welfare facilities ashore returning to normal, seafarers’ happiness had risen to high levels by the end of the year [27]. In this context, it is reasonable to formulate a hypothesis:Hypothesis the mean levels of fatigue reported by seafarers after the COVID-19 pandemic will be lower compared to the mean levels of fatigue reported by the sample during the pandemic. More broadly, this paper also draws on paradox theory [21] to examine fatigue risk factors during and after the pandemic and the related management issues. During the pandemic, although shipping companies took various measures to support seafarers [24], the pandemic-related fatigue factors, such as mental health problems, extended tour of duty, limited shore leave, and pandemic-induced additional workload, were largely beyond the control of individual companies. As such, these factors could play a significant role in causing fatigue. They reflect the tension between public health and seafarers’ occupational health and safety. In the aftermath of the pandemic, these factors have generally been removed. Nevertheless, the fatigue problem remains, since the traditional factors associated with the working environment and conditions, such as work intensification, shift work, and ship motion and vibration are still present and need to be managed. Furthermore, as suggested by the paradox theory [21], in times of change, new demands (or changes in demands) from multiple stakeholders can place constraints on limited resources and bring latent tensions to the surface. In the context of this study, it is reasonable to assume that such tensions may have an impact on fatigue management and affect seafarers’ experience of fatigue. As such, the main fatigue contributing factors would differ during and after the pandemic. To identify the differences and explore the implications for fatigue management, a research question can be formulated: what are the main fatigue contributing factors during the two periods? Before discussing the research methods, it is necessary to draw a line between during and after the pandemic in the context of this paper. After three years of implementing strict pandemic control measures, China reopened its borders and lifted travel restrictions on 8 January 2023, marking the end of the zero-COVID policy and a return to the post-pandemic normality [37]. As such, in relation to Chinese seafarers in this paper, ‘during the pandemic’ refers to the period from January 2020 to 7 January 2023, and the post-pandemic era begins on 8 January 2023. For seafarers of other nationalities, the ‘after pandemic’ start date is less clear-cut, but can be seen to start from mid-2022 when most countries began to gradually open their borders. 3 Research methods This research was conducted with two Chinese state-owned shipping companies. At the time of the research, Company A operated more than 300 container ships sailing on 357 international and domestic routes. Company B operated more than 300 dry bulk ships trading worldwide. Both companies directly employed around 10,000 officers and ratings on fixed-term contracts (5–8 years), all of whom are Chinese. They also recruited some ratings through crewing agencies on short-term tour-of-duty contracts. As state-owned companies with government financial support, the two companies provided directly employed seafarers with better social security coverage, more fringe benefits, and better working conditions than companies with other forms of ownership [7]. This research adopted a multi-phase mixed methods research design involving two quantitative surveys and 36 in-depth interviews. Both surveys used the same questionnaire, which is a revised version of the questionnaire used in the Cardiff Seafarers’ Fatigue Research Programme [20]. The questions were designed to explore the perceived fatigue levels of seafarers as well as the organisational and individual factors associated with fatigue. Survey A was conducted during the COVID-19 pandemic between November and December 2022. The questionnaire was distributed via email to all the 581 seafarers working on 25 ships of the companies and 501 of them (86.3%) participated and returned the survey questionnaire. Survey B was conducted after the pandemic between mid-January and the end of February 2023. The questionnaire emailed to all the 488 seafarers working on 21 of the companies’ ships and 412 of them (84.5%) returned the questionnaire. A total of 913 seafarers completed the questionnaire, including 442 officers (48.4%) and 438 ratings (48%). In terms of department, 386 participants (42.3%) were from the deck department and 480 (52.6%) were from the engine department. Their ages ranged from 23 to 59, with an average age of 34.6. Since the shipping companies only recruited male seafarers, all participants were male. Fatigue is an integration of subjective perception, performance and physiological functioning, making its measurement complex. To capture both the intensity and frequency of perceived physical and mental fatigue, we used the mean of five aspects of acute fatigue (see Appendix Table. A). It should be noted that physical fatigue is measured by item 4, and mental fatigue by item 5, in Table. A. In addition, a full list of the measures used in the analyses representing the demographic and work-related characteristics of the seafarers is provided in Appendix Table. B. These have been identified in previous research as factors associated with fatigue [12], [36], [8]. To provide a contextual explanation of the survey results, semi-structured interviews were conducted with 12 managers (6 from each company) and 24 seafarers (12 from each company) following Survey B in 2023. The interviews explored whether and how the lifting of the pandemic restrictions affected seafarers’ work and life on board and their experience of fatigue. The interviews were conducted online, and informed consent was obtained verbally from all the interviewees. Participants were anonymised. Ethical approval for this research was granted by the Psychology Ethics Committee of Dalian Maritime University. Interviews were audio recorded and transcribed. Analyses were conducted using the NVivo software package and organised around key emerging themes. 4 Analyses and results 4.1 Comparison of reported fatigue levels using propensity scores For the comparison between the two survey samples, a binary variable (treatment) was added to distinguish them. The 2022 sample (Survey A) was considered as the control group and the 2023 sample (Survey B) as the treatment group. The 2023 sample was then matched to the other one in order to reduce bias due to differences in respondent and work-related characteristics between the two samples. Nine variables were chosen according to the disjunctive cause criterion to match the samples [28], [29]. These selected variables were either independent of pandemic restrictions or affected the outcome of pandemic restrictions (i.e., fatigue) [16].1) Age 2) Crew number 3) Rank 4) Department 5) Experience at sea 6) Standing when on watch 7) Training on fatigue 8) Sleep periods per day 9) Rest before sailing Propensity score matching was used to estimate the average marginal effect of the removal of pandemic control on seafarers’ fatigue. The propensity score was estimated using logistic regression with option “glm” in R package MatchIt. This provided adequate balance ( Table 1), as indicated by an overall standardized mean difference of 0.0068 and standardized mean differences for all covariates below 0.13.Table 1 Sample means and balance information. Table 1 Survey A: original sample Survey A: matched sample Survey B sample Balance for matched data Mean Mean Mean Std. mean diff. (original) Std. mean diff. Variance ratio eCDF mean distance eCDF maximum distance Std. pair distance Distance 0.5662 0.5034 0.5023 0.9186 0.0068 1.0052 0.0027 0.0436 0.0034 Age 2.3964 2.4036 2.3782 -0.1944 0.0286 1.1270 0.0240 0.0473 0.8838 Crew number 2.8953 2.9891 2.9964 -0.4010 -0.0151 1.2044 0.0116 0.0291 0.1639 Rank 0.8151 0.9818 1.0036 -0.3835 -0.0223 0.9851 0.0073 0.0182 0.7779 Department 0.5568 0.5564 0.5855 -0.1358 -0.0519 1.1757 0.0242 0.0509 0.7224 Experience at sea 2.0735 2.0691 2.1382 -0.0451 -0.0446 1.1087 0.0141 0.0400 0.7392 Standing on watch 0.7996 0.7127 0.7236 0.6218 -0.0272 0.0109 0.0109 0.8054 Training on fatigue 0.6860 0.6327 0.6909 -0.0493 -0.1254 0.0582 0.0582 0.6118 Sleep periods per day 0.9773 1.0036 1.0036 -0.1570 0.0000 0.8551 0.0121 0.0182 0.5310 Rest before sailing 0.6726 0.6327 0.6073 0.1475 0.0542 0.0255 0.0255 0.6849 To estimate the treatment effect and its standard error, linear regression with acute fatigue as the outcome was then performed using SPSS version 27. The model included the treatment effect and the covariates used for the matching. The results are shown in Table 2.Table 2 Estimating the impact of the lifting of pandemic restrictions on acute fatigue. Table 2 Fatigue b (SE) Intercept 3.340(0.368) The release of pandemic control 0.275⁎⁎⁎(0.069) Age -0.156(0.049) Crew number -0.038(0.095) Rank -0.148(0.036) Department -0.109(0.063) Experience at sea -0.063(0.032) Standing when on watch 0.056(0.084) Training on fatigue -0.218⁎⁎⁎(0.075) Sleep periods per day -0.100(0.070) Rest before sailing -0.240⁎⁎⁎(0.075) The results indicate a significant treatment effect for acute fatigue (b = 0.275, SE = 0. 069, p <.001), indicating significantly higher average levels of reported fatigue after the pandemic than during the pandemic. The self-reported levels of acute fatigue, physical fatigue (measured by item 4 in Table. A) and mental fatigue (measured by item 5 in Table. A) in the two matched samples were also compared. The results (see Table 3) show that the means for acute fatigue, physical fatigue and mental fatigue were all significantly higher in the Survey B sample.Table 3 Comparison of fatigue levels between Survey A and Survey B respondents. Table 3 N Mean SD Sig Physical fatigue Survey A 275 1.4718 0.90639 0.000 Survey B 275 1.9806 0.84877 Mental fatigue Survey A 275 1.4978 0.98302 0.000 Survey B 275 1.9180 0.89554 Acute fatigue Survey A 275 1.58381 0.836166 0.033 Survey B 275 1.80968 0.686037 This suggests that seafarers reported higher levels of fatigue after the pandemic than during it. The results do not support the hypothesis. In fact, they support the opposite, that is, the mean levels of fatigue reported by seafarers after the COVID-19 pandemic are significantly higher than the mean levels of fatigue reported by the sample during the pandemic. 4.2 Fatigue factors To address the research question, stepwise regressions were conducted using SPSS version 27. Data from Survey A and Survey B (the original rather than the two matched samples) were analysed separately. All variables were added simultaneously. Table 4 and Fig. 1 show the top 3 factors (ranked by B value) associated with acute fatigue during the pandemic. Table 5 and Fig. 2 show the top 3 (direct) factors (ranked by B value) associated with acute fatigue after the pandemic. The top 2 underlying factors (ranked by B value) associated with each of the identified (direct) factors were also analysed in order to explore how seafarers’ fatigue was related to management strategies.Table 4 Stepwise regression with acute fatigue during the pandemic. Table 4Associated factors Fatigue b (SE) Supplies to vessel and crew (in port) Acute Fatigue -0.194⁎（.079) A lack of shore leave Acute Fatigue 0.183⁎（.081) Training on fatigue Acute Fatigue -0.138⁎（.059) Note. Unstandardized coefficients and standard errors, based on data from 501 respondents. ⁎ p <.05, ⁎⁎ p <.01, ⁎⁎⁎ p <.001. Fatigue factors are listed and ranked by B value. Fig.1 Visualization of the structural equation model in Table 4, showing the significant paths (n = 501). Fig.1 Table 5 Stepwise regression with acute fatigue and with the identified factors after the pandemic. Table 5 Associated factors Outcome b (SE) Part A Your state when you arrived at your most recent vessel ready for your new tour of duty Acute Fatigue 0.260⁎⁎⁎(0.052) Have difficulty getting up Acute Fatigue 0.212⁎⁎⁎(0.050) Have difficulty falling asleep Acute Fatigue 0.209⁎⁎⁎(0.058) Part B Effectiveness of fatigue risk management system Your state when you arrived at your most recent vessel ready for your new tour of duty -0.340⁎⁎⁎ (0.050) Experiences at sea Your state when you arrived at your most recent vessel ready for your new tour of duty -0.122⁎ (0.052) Training on fatigue Have difficulty falling asleep -0.624⁎⁎ (0.201) Policy on working hours Have difficulty falling asleep -0.563⁎ (0.257) Policy on working hours Have difficulty getting up 2.257⁎ (0.892) Experiences at sea Have difficulty getting up -0.242⁎⁎⁎ (0.061) Note. Unstandardized coefficients and standard errors, based on data from 412 respondents. ⁎ p <.05, ⁎⁎ p<.01, ⁎⁎⁎ p<.001. Fatigue factors are listed and ranked by B value. Fig.2 Visualization of the structural equation model in Table 5, showing the significant paths (n =412). Fig.2 Table 4 and Fig. 1 show significant negative associations between acute fatigue and two factors: training on fatigue (b =-.138, SE=.059, p=.021) and supplies to vessel and crew (in port) (b =-.194, SE=.079, p=.014), and a significant positive association between acute fatigue and a lack of shore leave (b =.183, SE=.081, p =.024). These results suggest that during the pandemic, training on fatigue, providing sufficient supplies while the ship is at berth, and allowing/arranging shore leave helped to reduce acute fatigue. While the first factor was related to fatigue risk management, the latter two were related to pandemic restrictions. Table 5 (Part A) and Fig. 2 (Columns 2 and 3) show that after the pandemic, the top 3 (direct) factors significantly associated with acute fatigue were: Your state when you arrived at your most recent vessel ready for your new tour of duty (b = 0.260, SE = 0.052, p <.001)，Have difficulty falling asleep (b=0.209，SE=0.058，p <.001) and Have difficulty getting up (b=0.212，SE=0.050，p <.001). This suggests that if seafarers arrived at their ship tired and if they felt they had difficulty falling asleep and getting up, they were more likely to feel fatigued. A further question is what underlying factors were are associated with the (direct) factors identified above. To explore this question, stepwise regressions were run with the top 3 (direct) factors as dependent variables. The results are shown in Part B of Table 5, and in Columns 1 and 2 of Fig. 2. Regarding the factor: Your state when you arrived at your most recent vessel ready for your new tour of duty, the regression results show that the top 2 underlying factors were the effectiveness of Fatigue Risk Management System (b=-0.340, SE=0.050, P <.001) and Experience at sea (b=-0.122, SE=0.052, P=.020). This suggests that the more effective the company’s Fatigue Risk Management System was and the more experienced the seafarers were, the less likely seafarers were to feel fatigued when they arrived at their vessel to start their tour of duty. In relation to the factor - Have difficulty falling asleep, the regression results show that the top 2 underlying factors were Training on fatigue (b=-0.624, SE=0.201, p=.002) and having policy on working hours (b=-0.563，SE=0.257，p=.031). This suggests that training on fatigue and having working hour policy were associated with lower frequency of having difficulty falling asleep. Regarding the factor – Have difficulty getting up, the regression results show that the top 2 underlying factors were having policy on working hours (b=-2.257，SE=0.892，p=.012) and Experience at sea (b = -0.242, SE = 0.061, p<.001). This suggests working hour policy and more experience at sea were associated with a lower frequency of feeling fatigued. In summary, in the aftermath of the pandemic, seafarers’ fatigue, through its association with sleep quality and sleep disturbance, is fundamentally related company’s fatigue risk management policies and practices (including training on fatigue, policy on working hours and effectiveness of fatigue risk management) and seafarers experience at sea. The more effective the management strategies were and the more experience they have at sea, the less likely seafarers were to report fatigue. These findings, however, do not explain why seafarers’ fatigue levels increased after the pandemic. To shed light on this issue, we next examine the interview data. 4.3 Interview findings During the pandemic, in compliance with the zero-COVID policy, Chinese shipping companies required their seafarers must wear anti-viral PPE1 when working on deck and the ship was docked. They also did not allow seafarers to take any shore leave, fearing that seafarers might catch the virus ashore or bring it back to China. Although the restrictions were officially lifted after the pandemic, seafarers reported that in many cases, Chinese port authorities continued to require seafarers to wear a N95 mask and gloves while in port and to prohibit shore leave. Similarly, Chinese shipping companies, especially large ones, discouraged their seafarers from taking shore leave, stating that seafarers would be held responsible for any consequences if they went ashore and contracted COVID-19. A chief officer reported: After three years of draconian pandemic control, we are eager to visit a port with no pandemic restrictions. At the moment, however, some Chinese ports still have some precautionary measures in place, and in some cases we are still not allowed to go ashore. We are still required by the company and the port authority to wear N95 masks and gloves when working in the port. We are very tired of this. It is not good for reducing our work pressure and it is bad for our mental health. The underlying reason is safety. COVID-19 spreads rapidly, and it is estimated that 80% of Chinese people were infected six weeks after the lockdown measures were lifted [15]. It is reasonable to assume that if one seafarer caught the virus, the whole ship would quickly be infected. Although the new strains tended to cause only mild symptoms, a large number of crew members falling ill at the same time would inevitably affect the safe operation of the ship. From this perspective, the restrictions may be understandable. However, they have been shown to exacerbate fatigue [6]. As such, the restrictions create a paradoxical tension between fatigue and safety. This tension is also reflected in safety inspections. During the pandemic, ship inspections, including port state control (PSC), flag state and company inspections, were carried out remotely. The lifting of the pandemic restrictions allowed the return of shipboard inspections. In relation to company inspection, a manager from Company A explained: During the pandemic, we did not have the opportunity to go on board and find out what was going on. Now that it is open, one of our most important tasks is to get the latest information on the situation on board as quickly as possible. This is also a kind of humanistic care for our crew. From their perspective, however, seafarers reported that different types of physical inspections were resumed at the same time and were carried out one after another at short intervals. Without any doubt, the inspection regime aims to ensure maritime safety and research shows that it has improved the maritime safety records over the years [26]. However, the intensified regime following the lifting of restrictions has also led to fatigue. A captain said: Following the COVID-19 pandemic, shipowners, flag States, port States and other relevant organisations have gradually resumed on-site ship inspections. However, as the pandemic has virtually prevented physical inspections for the past three years, all physical inspections are now being resumed and concentrated in this period. This leads to a sharp increase in our workload and pressure to prepare for all these inspections. This also disrupts our work pattern and rhythm in port, resulting in fatigue and potential safety hazards to navigation. We hope that the relevant organisations could consider coordinating the inspection time and adopt a method of combining remote inspection and on-site boarding inspection to reduce the inspection time in port and reduce the work pressure and workload of seafarers. Furthermore, post-pandemic policies and requirements for ship and port operations differ around the world. Seafarers had to quickly learn and adapt to the new safety regulatory environment and requirements, further exacerbating the fatigue problem. A third officer said: Each time the ship calls at a port, we have to learn new regulations and requirements. Although they are all based on IMO conventions, each port makes its own adjustments. As we are already very busy in port, we have to learn new rules, fill in different forms and prepare for strict inspections in a short period of time. This adds to the workload and makes us very tired. In addition, seafarers were required to learn the company’s policies, which had been revised and updated to reflect the policy changes made by the Chinese maritime authorities. A manager from Company B said: ‘We always sent the latest policies on board and required our seafarers to learn them to ensure they were up to date.’ However, the seafarers saw this as a burden, as a second engineer pointed out: Too many documents were sent to the ship, some of which were not directly related to our work. But our company required us to learn them and checked how we had learnt them. This is not really necessary. It is just a waste of time. Internet connectivity was found to alleviate fatigue problems and promote occupational health and safety during the pandemic [17]. In this study, although seafarers were provided with more free Wi-Fi data, they complained that it was mostly used to download and learn company policy updates rather than to communicate with family on shore. Thus, in this case, internet connectivity increased the demand on seafarers' work/learning time, leading to fatigue. A chief engineer complained: The IMO conventions, the guidelines and regulations issued by the Chinese maritime authorities. more and more documents that we have to learn. The company also checks how we have learnt and implemented the requirements via the wireless network on board. The consequences of non-compliance are very serious. We are only given a few tens of megabytes per month, all of which is used for the company's inspection. 5 Discussion Drawing on data from two surveys of seafarers’ experience of fatigue – Survey A during the pandemic and Survey B after the pandemic, this paper shows that, unexpectedly and alarmingly, seafarers' fatigue levels increased after the pandemic, despite the removal of pandemic-related fatigue factors. Regression analysis of Survey A data shows that, factors related to pandemic restrictions, such as lack of shore leave and insufficient supplies increased fatigue levels, while training on fatigue helped reduce fatigue levels. The results corroborate previous research findings that the pandemic posed challenges for fatigue management in the maritime industry ([10]; Kaptan & Kaptan, 2021; [17], [19]). Regression analyses of Survey B data show that fatigue risk factors changed after the pandemic. During this period, seafarers' reported fatigue levels were directly related to factors such as sleep problems (difficulty falling asleep and getting up) and whether they were tired when they arrived at the ship to start work, and these factors were in turn related to the company's fatigue risk management strategies (including training on fatigue, policy on working hours and the effectiveness of fatigue risk management) and seafarers' experiences at sea. These findings also confirm previous research that effective fatigue risk management helps to reduce fatigue levels and that working arrangements and conditions have significant implications for seafarers’ occupational health and safety [36], [9]. Although the results of the regression analyses are consistent with previous research, they do not provide an explanation as to why seafarers felt more fatigued after the pandemic. In this context, the interview results complement the survey results and shed light on the issue. They suggest that in the immediate aftermath of the pandemic, seafarers were required to continuously comply with pandemic preventive measures in some ports, to cope with concentrated ship inspections, and to learn and adapt to various newly updated port regulations and company policies. All of this increased their workload and contributed to fatigue in the name of safety. Overall, this multi-phase mixed methods study reveals paradoxes and tensions in relation to fatigue management in the shipping industry. Firstly, the removal of pandemic restrictions (and the associated fatigue risk factors) has led to seafarers reporting higher levels of fatigue. Underlying this first paradox is the second one – while the demands of complying with the intensified safety inspection regime and adapting to policy changes were intended to improve safety, they nevertheless increased seafarers' workload and reduced their rest periods. This creates a tension between safety and fatigue. The paradox of seafarers being fatigued by safety inspections has been discussed in previous research [5]. Drawing on paradox theory [21], this paper adds to the discussion by showing that a period of transition brings with it the need to adapt quickly to changes in the environment, which puts a strain on resources and brings tensions to the surface. Returning to the results of the regression analyses of the two surveys, they show that training on fatigue led to lower fatigue levels both during and after the pandemic. They also show that after the pandemic, a number of fatigue risk management measures (training on fatigue, working hour policy and fatigue risk management system) helped seafarers to get better rest, thereby reducing fatigue levels and promoting seafarers’ occupational health and safety. These findings suggest that while the paradoxical tensions that can cause fatigue cannot be eliminated, fatigue risk can be managed and mitigated through the implementation of fatigue risk management policies and practices. 6 Conclusion This paper takes the initiative to examine seafarers’ experiences of fatigue during the transition to the post-pandemic normality. Using propensity score matching, data from the two surveys conducted during and after the pandemic show that surprisingly seafarers reported higher levels of fatigue after the pandemic. Qualitative interviews with seafarers and ship managers following the surveys reveal the underlying reason – the intensified ship inspection regime together with policy and regulatory updates/revisions in the aftermath of the pandemic increased seafarers’ workload and made it difficult for them to get good rest. The results of the two surveys also show that the fatigue risk factors were different during and after the pandemic. During the pandemic, risk factors tended to be related to pandemic restrictions. After the pandemic, risk factors were more likely to be related to the company’s fatigue risk management policies and practices. This paper sheds new light on fatigue management in the shipping industry. Drawing on paradox theory [21], it shows that the transition to the post-pandemic normality exacerbates the tension between addressing safety concerns (a concentration of various ship inspections in a short period) and fatigue risk management. It also demonstrates that this tension and fatigue risk can be mitigated by adopting effective fatigue risk management policies and practices. These findings have both policy and management implications. To safeguard seafarers’ occupational health and safety in general, at the policy level, ship inspection authorities should coordinate their inspection activities to reduce disruptions to seafarers' busy schedules in port. Efforts by national maritime administrations to develop and revise maritime and port policies and regulations should be streamlined and coordinated with international organisations. At the level of company management, it is important to monitor seafarers' fatigue during periods of change and transition, to assess whether changes in demand will exacerbate stress, and to provide adequate training to manage and reduce stress and fatigue risk. According to paradox theory [21], tensions cannot be eliminated due to multiple and conflicting demands, so organisations need to accept and manage them proactively to achieve sustainable development. In the context of managing fatigue issues, as suggested by Gander et al. [9], organisations should adopt a systematic approach and implement a fatigue management system, which would include a fatigue management policy, education and awareness training, fatigue level monitoring, fatigue reporting mechanisms, and fatigue incident reporting and investigation procedures. It is worth noting that this research has limitations. It focuses on two Chinese shipping companies and the samples are not representative. However, being large and state-owned, the two companies have more financial resources to manage fatigue than smaller and non-state-owned shipping companies. If they find it difficult to manage fatigue, other companies with fewer resources are likely to find it more difficult. Future research can extend the focus to seafarers and shipping companies of other nationalities to explore the issue further. CRediT authorship contribution statement Zhiwei Zhao: Conceptualization, Methodology, Investigation, Formal analysis, Resources, Writing - original draft, Writing - review & editing. Lijun Tang: Conceptualization, Investigation, Methodology, Validation, Writing - original draft, Writing - review & editing. Yunlei Ma: Conceptualization, Methodology, Investigation, Writing – review & editing. Yueyan Wu: Methodology, Investigation, Formal analysis, Writing – review & editing. Shiqi Lin: Methodology, Formal analysis, Writing – review & editing. Zefan Wu: Investigation, Data curation, Formal analysis. Zheng Zhou: Investigation, Formal analysis. Xinyi Wang: Validation, Formal analysis. Xinjian Wang: Formal analysis, Validation. Appendix Table. A Table A The five items to measure acute fatigue. Table A1 Which of the following responses best describes your typical state during work? Scored 0 (very alert) to 4 (sleepy). 2 About how often do you feel tired at work? Scored 0 (never) to 4 (about everyday). 3 About how often do you feel sleepy at work? Scored 0 (never) to 4 (about everyday). 4 In a normal workday, how physically tired do you usually feel at the end of the working day? Scored 0 (Not at all) to 4 (Extremely). 5 On a normal working day, how mentally tired do you usually feel at the end of the working day? Scored 0 (Not at all) to 4 (Extremely). Table. B Table B Descriptive statistics of during and post pandemic samples across demographic and work-related characteristics. Table BMeasures During pandemic Post pandemic 1. Supplies to vessel and crew (in port) Yes, enough supplies No, insufficient 84.4% (422) 15.6% (78) 89.2% (364) 10.8% (44) 2. Have shore leave in port Yes, have shore leave Sometimes Almost no shore leave 2.0% (10) 2.8% (14) 95.2% (477) 78.7% (322) 20.5% (84) 8% (3) 3. Training on fatigue Yes, have fatigue training No training 71.1% (356) 28.9% (145) 70.4% (285) 29.6% (120) 4. Over contract Always take the annual leave on time Occasionally delayed Always delayed 26.5% (133) 53.3% (267) 20.2% (101) 54.6% (221) 39.6% (161) 5.8% (23) 5. Sleep before starting work on board Yes, we have the opportunity to sleep No, we don’t 61.3% (307) 38.7% (194) 66.7% (273) 33.3% (136) 6. Your state when you arrived at your most recent vessel ready for your new tour of duty Not fatigued Slightly tired Moderately tired Very tired Extremely tired 50.3% (252) 25.7% (129) 13.4% (67) 6.8% (34) 3.8% (19) 51.2% (211) 34.7% (143) 8.3% (34) 3.9% (16) 1.9% (8) 7. Have difficulty falling asleep Not at all A little Quite a bit Almost always 9.8% (49) 55.2% (277) 28.8% (144) 6.2% (31) 26.9% (109) 45.2% (183) 21.7% (88) 6.2% (25) 8. Have difficulty getting up Not at all A little Quite a bit Almost always 19.6% (98) 41.4% (207) 27.1% (136) 12.0% (60) 28% (113) 39.6% (160) 20.0% (81) 12.4% (50) 9. Ship motion disturbing sleep Not at all A little Quite a bit Very much 8.4% (42) 43.2% (217) 36.0% (180) 12.4% (62) 6.1% (25) 33.6% (137) 39.7% (162) 20.6% (84) 10. Wake up confused, disorientated, irritable Not at all A little Quite a bit Almost always 34.1% (171) 38% (190) 20.9% (105) 7.0% (35) 41.0% (167) 35.4% (144) 15.0% (61) 8.6% (35) 11. Effectiveness of Fatigue Risk Management System Not effective Somewhat effective Very effective 8.1% (41) 52.0% (261) 39.9% (199) 8.9% (21) 53.5% (124) 38.6% (91) 12. Experiences at sea (in group) less than 5 years 6-10 11-15 16-20 21-25 26-30 31-35 more than 35 43.7% (219) 25.1% (126) 15.2% (76) 8.0% (40) 1.6% (8) 2.6% (13) 1.6% (8) 2.2% (11) 53.8% (218) 19.5% (79) 12.1% (49) 6.9% (28) 3.5% (14) 3.0% (12) 0.7% (3) 0.5% (2) 13. Policy on working hours Yes, we have the policy No, we don’t 90.2% (452) 9.8% (49) 89.4% (345) 10.6% (41) 14. Standing when on watch Yes No 53.1% (266) 46.9% (235) 80.0% (328) 20.0% (82) 15. Sleep Periods per 24 hours 1 sleep period 2 sleep periods 3 or more sleep periods 12.2% (61) 70.9% (355) 17.0% (85) 11.0% (45) 79.8% (327) 9.3% (38) 16. Age (in group) less than 25 26-35 36-45 46-55 more than 56 13.4% (66) 36.3% (178) 33.0% (162) 16.9% (83) 0.4% (2) 11.8% (47) 52.9% (210) 21.4% (85) 12.1% (48) 1.8% (7) 17. Rank Officer Rating 38.3% (192) 61.7% (309) 61.9% (250) 38.1% (154) 18. Department Deck Engineering 39.5% (198) 56.5% (283) 48.0% (197) 48.0% (197) Uncited reference Bai et al., [2] Data availability The data that has been used is confidential. 1 PPE refers to personal protective equipment, consisting of four items: a pair of disposable medical gloves, an N95 mask, a hazmat suit, and a pair of protective goggles. ==== Refs References 1 Bai X. Xu M. Han T. Yang D. Quantifying the impact of pandemic lockdown policies on global port calls Transportation Research Part A: Policy and Practice 164 2022 224 241 10.1016/j.tra.2022.08.002 36034472 2 Bai X. Xu M. Han T. Yang D. Quantifying the impact of pandemic lockdown policies on global port calls Transportation Research Part A: Policy and Practice 164 2022 224 241 10.1016/j.tra.2022.08.002 36034472 3 Baygi F. Mohammadian Khonsari N. Agoushi A. Hassani Gelsefid S. Mahdavi Gorabi A. Qorbani M. Prevalence and associated factors of psychosocial distress among seafarers during COVID-19 pandemic BMC Psychiatry 21 1 2021 1 9 33388029 4 Bhardwaj S. Bhattacharya S. Tang L. Howell K.E. Technology introduction on ships: The tension between safety and economic rationality Safety Science 115 2019 329 338 5 Bhattacharya S. Tang L. Fatigued for safety? Supply chain occupational health and safety initiatives in shipping Economic and Industrial Democracy 34 3 2013 383 399 6 Brooks S.K. Greenberg N. Mental health and wellbeing of seafaring personnel during COVID-19: Scoping review Journal of Occupational Health 64 1 2022 e12361 7 Chen G. Tang L. Competing for seafaring labour: Social security and agency employment in Chinese shipping Global Networks 22 1 2022 89 102 8 Dohrmann S.B. Leppin A. Determinants of seafarers’ fatigue: A systematic review and quality assessment International Archives of Occupational and Environmental Health 90 1 2017 13 37 10.1007/s00420-016-1174-y 27804037 9 Gander P. Hartley L. Powell D. Cabon P. Hitchcock E. Mills A. Popkin S. Fatigue risk management: Organizational factors at the regulatory and industry/company level Accident Analysis & Prevention 43 2 2011 573 590 21130218 10 Hebbar A.A. Mukesh N. COVID-19 and seafarers’ rights to shore leave, repatriation and medical assistance: A pilot study International Maritime Health 71 4 2020 217 228 33394486 11 IMO. (2019). Guidelines on fatigue. https://wwwcdn.imo.org/localresources/en/OurWork/HumanElement/Documents/MSC.1-Circ.1598.pdf 12 Jepsen J.R. Zhao Z. Leeuwen W.M.A. van Seafarer fatigue: A review of risk factors, consequences for seafarers’ health and safety and options for mitigation International Maritime Health, 66(2), Art 2015 2 10.5603/IMH.2015.0024 13 Kaptan M. Olgun Kaptan B. The investigation of the effects of COVID-19 restrictions on seafarers Australian Journal of Maritime & Ocean Affairs 2021 1 13 14 Nichols T. The sociology of industrial injury 1997 Mansell 15 Parkinson, J. (2023, January 22). ‘80% of China is Infected With COVID-19’ Health Official Reports. ContagionLive. https://www.contagionlive.com/view/-80-of-china-is-infected-with-covid-19-health-official-reports 16 Pauksztat B. Andrei D.M. Grech M.R. Effects of the COVID-19 pandemic on the mental health of seafarers: A comparison using matched samples Safety Science 146 2022 105542 17 Pauksztat B. Grech M.R. Kitada M. The impact of the COVID-19 pandemic on seafarers’ mental health and chronic fatigue: Beneficial effects of onboard peer support, external support and Internet access Marine Policy 137 2022 104942 18 Pesel G. Canals M.L. Sandrin M. Jensen O. Wellbeing of a selection of seafarers in Eastern Adriatic Sea during the COVID-19 pandemic 2020 International Maritime Health 71 3 2020 184 190 33001430 19 Slišković A. Seafarers’ well-being in the context of the COVID-19 pandemic: A qualitative study Work 67 4 2020 799 809 33325430 20 Smith, A.P., Allen, P.H., & Wadsworth, E.J. K. (2006). Seafarer fatigue: The Cardiff research programme. 21 Smith W.K. Lewis M.W. Toward a theory of paradox: A dynamic equilibrium model of organizing Academy of Management Review 36 2 2011 381 403 22 Tang L. Defending workers’ rights on social media: Chinese seafarers during the COVID-19 pandemic. Industrial Relations Journal 53 2 2022 110 125 23 Tang L. The impact of 2008 financial crisis and COVID-19 pandemic on the demand and supply of seafarer officers: Evidence from China Marine Policy 145 2022 105263 24 Tang L. Abila S. Kitada M. Malecosio S. Jr Montes K.K. Seafarers’ mental health during the COVID-19 pandemic: An examination of current supportive measures and their perceived effectiveness Marine Policy 145 2022 105276 25 Tang, L., Acejo, I., Ellis, N., Turgo, N., & Sampson, H. (2013). Behind the headlines? An analysis of accident investigation reports. Seafarers International Research Centre Symposium Proceedings (2013), 1. 26 Tang L. Zhang P. Human Resource Management in Shipping: Issues, Challenges, and Solutions 2021 Routledge 27 The Mission to Seafarers. (2023). Seafarers Happiness Index: Quarter 4 2022 and Annual Review. 〈https://www.happyatsea.org/wp-content/uploads/2023/01/Seafarers-Happiness-Index-Q4-2022.pdf〉 28 VanderWeele T.J. Principles of confounder selection European Journal of Epidemiology 34 2019 211 219 30840181 29 VanderWeele T.J. Shpitser I. A new criterion for confounder selection Biometrics 67 4 2011 1406 1413 21627630 30 Wadsworth E.J. Allen P.H. McNamara R.L. Smith A.P. Fatigue and health in a seafaring population Occupational Medicine 58 3 2008 198 204 18310605 31 Walters, D., & Bailey, N. (2013). Lives in peril: Profit or safety in the global maritime industry? Springer. 32 Williamson A. Lombardi D.A. Folkard S. Stutts J. Courtney T.K. Connor J.L. The link between fatigue and safety Accident Analysis & Prevention 43 2 2011 498 515 10.1016/j.aap.2009.11.011 21130213 33 Xue C. Tang L. Walters D. Who is dominant? Occupational Health and Safety management in Chinese shipping Journal of Industrial Relations 59 1 2017 34 Zhao Z. Tang L. The impact of COVID-19 on maritime pilots: Evidence and lessons Marine Policy 153 2023 105664 35 Zhao Z. Tang L. Wu Y. Fatigue During the COVID-19 Pandemic: The Experiences of Chinese Seafarers Marine Policy 2023 105643 10.1016/j.marpol.2023.105643 36 Zhao Z. Wadsworth E. Jepsen J.R. van Leeuwen W.M. Comparison of perceived fatigue levels of seafarers and management approaches in fatigue mitigation: Case studies from two Chinese and two European shipping companies Marine Policy 116 2020 103897 37 Zhou, J., & Tian, Y.L. (2023). China reopens borders in final farewell to zero-COVID. Reuters. https://www.reuters.com/world/china/china-reopens-borders-final-farewell-zero-covid-2023-01-08/
PMC010xxxxxx/PMC10288310.txt	==== Front Vacunas (English Edition) 2445-1460 2445-1460 Published by Elsevier España, S.L.U. S2445-1460(23)00035-3 10.1016/j.vacune.2023.06.002 Original Article IMPACT OF VACCINATION AGAINST SARS-CoV-2 ON THE INCIDENCE OF INFECTION IN SCHOOL SETTINGS. IMPACTO DE LA VACUNACIÓN CONTRA SARS-CoV-2 EN LA INCIDENCIA DE INFECCIÓN EN ÁMBITO ESCOLARDiez María Teresa Herrero a⁎ Valdivieso María Inés Salado b Domínguez Sara Carbajal c Tango Marta Allué b Caballero Juan Carlos Villa b Hernández Clara Berbel b a Servicio de Medicina Preventiva y Salud Pública, Hospital Clínico Universitario de Valladolid. Avda Ramón y Cajal 3, 47003, Valladolid, Spain. b Sección de Epidemiología del Servicio Territorial de Sanidad de Valladolid, Delegación Territorial de Sanidad de Valladolid, Consejería de Sanidad Junta de Castilla y León. Avda Ramón y Cajal 6, 47003, Valladolid, Spain. c Servicio de Medicina Preventiva y Salud Pública, Hospital Universitario de Araba. José Atxotegui Kalea s/N, 01009, Vitoria-Gasteiz, Spain. ⁎ Autora de correspondencia: Hospital Clínico Universitario de Valladolid, Calle Ramón y Cajal 3, 47003, Valladolid, Spain. 23 6 2023 23 6 2023 © 2023 Published by Elsevier España, S.L.U. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background The SARS-CoV-2 pandemic affected the school-aged population because of the disease itself and due to the measures applied for prevention and control of the infection. The aim of the study was to evaluate the effect of population-based vaccination against COVID-19 on the incidence of infection in school settings. Material and Methods A retrospective descriptive study of COVID-19 cases and school outbreaks was carried out at the province level. Students, teachers and staff from different educational stages of the schools were included. The outcome measure was the incidence according to educational stage, case profile and clinic during the first of the academic year 2020/2021 versus the same period 2021/2022. Results The total incidence of SARS-CoV-2 in classrooms was 2470 cases per 100,000 population in the first trimester of the academic year 2020/2021 and 2720 cases per 100,000 population in the same period 2021/2022. The number of reported school outbreaks was 7 times higher in this second period; and the risk of infection in classrooms over 12 years of age (students and teachers) was reduced by 43.1% (vaccinated in high percentage). Conclusions This study shows a reduction in transmission of SARS-CoV-2 infection in students of higher educational stages (secondary and high school) during the first of the academic year 2021/2022 (group with high vaccination coverage at the beginning of the period) compared to the previous school year (without vaccination). Antecedentes. La pandemia por SARS-CoV-2 afectó a la población en edad escolar debido a la propia enfermedad y, al mismo tiempo, por las medidas aplicadas de prevención y control de la infección. El objetivo del estudio fue evaluar el efecto de la vacunación poblacional contra COVID-19 en la incidencia de infección en ámbito escolar. Materiales y métodos. Se realizó un estudio descriptivo retrospectivo de casos y brotes escolares por COVID-19 a nivel provincial. Se incluyó alumnado, docentes y personal de diferentes etapas educativas de los centros. La medida principal de resultados fue la incidencia acumulada en función de etapa educativa, perfil del caso y clínica durante el primer trimestre del curso 2020/2021 frente al mismo periodo 2021/2022. Resultados. La incidencia total de infección por SARS-CoV-2 en las aulas fue de 2.470 casos por cada 100.000 habitantes en el primer trimestre del curso 2020/2021 y de 2.720 casos por cada 100.000 habitantes en el mismo periodo 2021/2022. El número de brotes escolares notificados fue 7 veces mayor en este segundo periodo; y, al mismo tiempo, el riesgo de infección en las aulas de mayores de 12 años (alumnos y docentes) se redujo un 43,1% (vacunados en elevado porcentaje). Conclusiones. Este estudio muestra menor transmisión de infección por SARS-CoV-2 en los alumnos de las etapas educativas superiores (Secundaria y Bachillerato) durante el primer trimestre escolar 2021/2022 (grupo con elevada cobertura vacunal al inicio del período) respecto al curso previo (sin vacunación). Keywords COVID-19 Infections Vaccines Schools Palabras claves COVID-19 Infección Vacuna Escuelas. ==== Body pmcIntroduction The school population has been affected by the severe acute respiratory syndrome type 2 (SARS-CoV-2) coronavirus pandemic, not only directly by the disease itself, but also through the implementation of infection prevention and control measures, with physical, mental, and educational repercussions.1 The negative effect on well-being and mental health due to school closures has been reflected in a systematic review with data from 20 countries.2 In Spain, the incidence of COVID-19 infection over these 2 years was very variable. The type of variant circulating and the age groups affected also varied. The initial SARS-CoV-2 lineage was circulating during the first trimester of the 2020/2021 school year, and children were considered less susceptible to infection.3 In contrast, in the first school trimester 2021/2022, the emergence of the delta lineage variant (B.1.617.2 and AY subvariants) with higher transmission capacity resulted in higher incidences in the group of children aged 5 to 12 years. The susceptibility of the paediatric population to SARS-CoV-2 infection was comparable to that of other age groups. However, the severity of the disease is low in children under 18 years of age, in terms of hospitalisation and case fatality.4 The reopening of schools after the closure during the first wave of the pandemic made it possible to evaluate the transmission of SARS-CoV-2 infection in school settings. Several published studies have shown a lower level of COVID-19 transmission in schools compared to the family or social setting, after applying the recommended infection response measures.5 , 6 , 7 Vaccination is the best available preventive tool against severe symptomatic COVID-19 disease.8 , 9 , 10 Vaccination in Spain started on 27 December 2020 in accordance with the recommendations of the Vaccination Strategy,11 and teachers, school staff, and children aged 12–19 years had access to vaccination prior to the start of the 2021/2022 school year. This national immunisation programme provided vaccination coverage of 89.7% in the population aged 12 years and over as of 22 December 2021 (end of the school trimester).12 Despite the proven effectiveness of immunisation against COVID-19 against the severity and fatality of the infection, vaccine efficacy in reducing the number of cases has been highly questioned. Therefore, we believe it appropriate to assess the effect of population-based vaccination on the incidence of SARS-CoV-2 infection in the school setting, the representative population with and without access to vaccination, by age and time period. Objectives Primary objective: to assess the impact of population-based vaccination against SARS-CoV-2 on infection in the school setting at provincial level. Secondary objectives: to analyse COVID-19 infection (cases and outbreaks) in the school setting according to educational stage, profile (pupil/teacher/staff), and case symptomatology. To describe the differences in incidence in schools in the first trimester of the 2020/2021 school year compared to the same period 2021/2022, and to analyse the vaccination status of confirmed cases of SARS-CoV-2 infection among pupils, teachers, and school staff. Material and Methods We conducted a retrospective descriptive observational epidemiological study of school cases and outbreaks of COVID-19 infection in the province of Valladolid. The study included pupils, teachers, and staff in the second cycle of infant, primary, secondary, and baccalaureate stages in public, state-subsidised, and private schools. The first trimester of the 2021/2022 school year (10 September 2021 to 22 December 2021) was evaluated compared to the first trimester of the 2020/2021 school year (9 September 2020 to 22 December 2020). Case, outbreak, and close contact were defined according to the Action Guide for the onset of cases of COVID-19 in schools13 at regional level (asymptomatic or symptomatic case with positive diagnostic test for active infection [PDIA] and attendance at the school within the last 48 h from the onset of symptoms or the date of PDIA [invalid self-test]; outbreak as a grouping of 3 or more cases with active infection with an epidemiological link established in the last 10–14 days). And, in the educational stages, the 2nd cycle of infant education and the 1st year of primary education were considered a joint school group because they were exempt from wearing masks (GSM), following the definitions of the protocol in force in Castilla y León,13 which grouped these educational levels into a single subgroup with specific prevention and control measures because they were allowed not to wear masks and due to their greater social interaction. During this study, the prevention measures of social distancing, hand hygiene, masks, and ventilation in enclosed spaces were applied in the schools; together with the mitigation measures of close school contact tracing for confirmed cases with PDIA, following the protocol and applying quarantine for 10 days (14 days in the 2020/2021 school year) from the last contact with a confirmed GSM case and in the case of a non-GSM (not exempt from mask wearing) outbreak (in the 2020/2021 school year, quarantine was applied to all close contacts in non-GSM classrooms; however, in the 2021/2022 school year, it was applied only to unvaccinated or vulnerable close contacts).13 During the study period, vaccination in Spain included groups 6B and 6C (teachers and school staff) and group 13 (persons aged 12–19 years, born between 2002 and 2009 inclusively) following Spain's COVID-19 vaccination strategies.11 The immunisation schedule (full, partial, and unvaccinated) was defined according to the current state strategy for primary vaccination, following the recommendations and the technical data sheet of the vaccines according to age, vaccine preparation, number of doses, and intervals between doses. Data sources: a separate register was created in the Epidemiological Surveillance System of the Epidemiology Section of the Territorial Health Service of Valladolid with the notification of confirmed cases, contacts, and outbreaks identified by the COVID-19 team of the Provincial Directorate of Education. It was extended with the collection of clinical information on the cases through the primary care and hospital care systems of the public and private network of the province and the Epidemiological Surveillance Information System of Castilla y León. Vaccination against COVID-19 was collected from the VACU vaccine module registry. Data on cumulative incidence and population vaccination coverage in the province were extracted through the “Datos Abiertos” (Open Data) portal of Castilla y León14 and taking the data collected by the National Institute of Statistics as the estimated reference population for the province of Valladolid.15 Outcome measures: the main outcome measure was the calculation of the cumulative incidence of SARS-CoV-2 infection in schools in the first trimester of the academic year 2020/2021 and the same period in 2021/2022. It was then disaggregated according to the profile of the case in classrooms (pupil or teacher) and educational stage (2nd cycle infant with 1st year of primary school, 2nd to 6th year of primary school, secondary school, and baccalaureate), comparing both periods. In addition, cases and outbreaks of coronavirus infection were characterised and the vaccination status (unvaccinated, partial, or full schedule) was analysed with respect to the primary vaccination of COVID-19 cases in the 2021/2022 school year in students over 12 years of age, teachers, and school staff. Analysis of outcomes: the association of qualitative variables was analysed by comparing proportions of independent samples using Pearson's χ2 test. Fisher's exact test or the likelihood ratio test was used for variables with more than 2 categories if the number of cells with expected values less than 5 was greater than 20%. P-values < .05 were considered statistically significant. We used the statistical programmes SPSS version 15.0.1 (SPSS Inc., Chicago, USA) and EPIDAT version 4.2 (EPIDAT, Spain). This study was approved by the Valladolid West Area Drug Research Ethics Committee. Results A total of 1911 cases of active SARS-CoV-2 infection were reported in the school setting during the first quarter of the academic year 2020/2021 and 2010 cases during the same period in 2021/2022. The distribution by sex and case profile (pupil, teacher, and staff) was homogeneous in both periods, as was the proportion of cases with clinical symptoms (n = 943; 61.7% vs. n = 1254; 59.4%). Table 1 presents the general characteristics of cases and outbreaks in both periods.Table 1 General characteristics of school cases and outbreaks of COVID-19 infection during the first quarter of the school year 2020/2021 and 2021/2022. Table 1 School year 2020/2021 School year 2021/2022 Total number of cases 1911 2010 Total number of outbreaks 15 108 Average number of cases in outbreaks Range of number of cases in outbreaks 6.27 (21 to 3) 4.66 (17 to 3) Gender (% of females of the total) 1024 (54%) 1109 (53%) Case profile (count, %) Pupil 1680 (87.9%) 1786 (84.6%) Teacher 207 (10.8%) 275 (13.1%) Staff 24 (1.3%) 49 (2.3%) Educational stage of the case (count and %) 2nd cycle infants and 1st year primary school 438 (23.1%) 564 (26.9%) 2nd to 6th year primary 646 (34%) 1056 (50.9%) Secondary 587 (30.9%) 318 (15.2%) Baccalaureate 228 (12%) 155 (7.4%) Symptomatology in case (symptom count and %) 943 (61.7%) 1254 (59.4%) The cumulative incidence in classrooms (pupils and teachers) was 2470 cases per 100,000 inhabitants in the first quarter of the academic year 2020/2021 and 2720 cases per 100,000 inhabitants in the same period 2021/2022. The case counts per case profile in classrooms (pupils and teachers) according to educational stage in both trimesters of the study are presented in Fig. 1 .Figure 1 Cases of COVID-19 infection in pupils and teachers by educational stage during the first trimester of the 2020/2021 and 2021/2022 school years. Abbreviations: VAC (vaccinated), UNVAC (unvaccinated). Figure 1 The risk of SARS-CoV-2 infection increased by 51.6% in under-12 classes (unvaccinated) in the 2021/2022 study period compared to the same period in 2020/2021. However, in classes over 12 years of age (highly vaccinated) the risk of infection between these periods decreased by 43.1%. In both cases the differences were statistically significant. In the analysis by subgroups, we observe that secondary school and baccalaureate pupils had a 53.3% and 36% decrease in risk respectively in the period 2021/2202 compared to 2020/2021; in contrast, the risk was 72.5% higher in students in years 2–6 of primary school. Table 2 presents the results of the cumulative incidences by case profile and educational stage of the classes analysed in the periods of the study.Table 2 Cumulative 3-month incidences of COVID-19 infection in classrooms during the first trimester of school years 2020/2021 and 2021/2022. Table 2 School year 2020/2021 School year 2021/2022 p-value Total CI 2470/100000 inhab. 2720/100000 inhab. 0.002 Total CI per classroom Under 12 years old 2391/100000 inhab. 3625/100000 inhab. < 0.001 Over 12 years old 2578/100000 inhab. 1466/100000 inhab. < 0.001 CI in pupils 2nd cycle infants and 1st year primary 2310/100000 inhab. 2848/100000 inhab. 0.002 2nd to 6th year primary 2221/100000 inhab. 3831/100000 inhab. < 0.001 Secondary 2514/100000 inhab. 1175/100000 inhab. < 0.001 Baccalaureate 3259/100000 inhab. 2085/100000 inhab. < 0.001 CI in teachers Under 12 years old 3990/100000 inhab. 5741/100000 inhab. 0.001 Over 12 years old 1712/100000 inhab. 1888/100000 inhab. 0.555 CI: Cumulative Incidence; inhab.: inhabitants. We recorded 15 outbreaks in our study in the first period 2020/2021 versus 108 outbreaks in the first trimester of the academic year 2021/2022, and the average number of cases involved in each outbreak was higher in the first period (6.27 cases per outbreak versus 4.66 cases per outbreak). Characterisation of outbreaks according to person, time, and space (urban, semi-urban in populations with more than 20,000 inhabitants, and rural) between the two periods showed some differences. In the study period of the 2020/2021 school year, more outbreaks were reported in classrooms in 2nd cycle infants and 1st year primary (n = 7; 47%). However, in the 2021/2022 school year, the highest number of outbreaks was reported in the 2nd-6th years of primary school (n = 75; 69%). Urban schools had the highest number of cases involved in outbreaks in the second period (n = 338; 67.2%), in contrast to rural schools in the first period (n = 39; 41.5%). In both periods, pupils were the cases most involved in outbreaks. The vaccination status of cases reported in school settings (pupils, teachers, and school staff) was analysed for the 2021/2022 quarter, a period when staff, teachers, and children aged 12 years and older had access to vaccination. The vaccination schedule was complete in more than 90% of COVID-19 cases in teachers and school staff. For pupils over 12 years of age, partial or full schedule immunisation was approximately 88% of confirmed cases (p = .002). Table 3 shows the vaccination status of COVID-19 cases according to the outbreak they fell under, case profile, educational stage, and symptoms.Table 3 Vaccination status of COVID-19 cases in the school setting who had access to immunisation during the first school trimester 2021/2022 (12 years and older, teachers, and school staff). The vaccination schedule was defined according to primary vaccination as per the technical data sheet of the vaccine preparation administered. Table 3 Vaccination status p-value Unvaccinated Partial vaccination schedule Full vaccination schedule n % n % n % Outbreak case No 54 8.9 28 4.6 524 86.5 0.279 Yes 7 7.9 1 1.1 81 91 Case profile Pupil 44 11.7 22 5.9 309 82.4 0.002 Teacher 13 4.8 7 2.6 252 92.6 Staff 4 8.3 0 0.0 44 91.7 Educational stage of the case Infants and 1st primary (teachers and staff) 6 5.6 3 2.8 98 91.6 0.239 2nd to 6th primary (teachers and staff) 7 5.8 4 3.3 110 90.9 Secondary (teachers, staff, and pupils) 35 11.6 16 5.3 251 83.1 Baccalaureate (teachers, staff and pupils) 12 7.8 6 3.9 136 88.3 Symptoms in the case No 20 11.9 5 3 143 85.1 0.190 Yes 41 7.8 24 4.5 462 87.7 The supplementary material contains a graphical summary of the study (Graphical summary). Discussion This study analysed the epidemiological situation of SARS-CoV-2 infection in schools during the first trimesters of the school year in the province of Valladolid, evaluating both periods. The sample (pupils, teachers, and educational staff) can be considered homogeneous, as can the periods of interest (first trimester of the academic year 2020/2021 and 2021/2022), and the protocols, protection, and control measures applied in COVID-19 infection in the school setting during the time of the study. Therefore, the differences observed between the study periods may be due to the effect of population-based vaccination in the 2021/2022 school year; without forgetting the impact of the different circulating variants of SARS-CoV-2. During the 2020/2021 and 2021/2022 school trimesters considered, the total incidence of COVID-19 in the schools was similar. If we evaluate the total population incidence of infection in our province in the same periods of this study, we observe that in the first school trimester 2021/2022 it was 2568 cases per 100,000 inhabitants, in contrast to the previous school year when it was 4434 cases per 100,000 inhabitants. The population and school incidence results obtained for the 2021/2022 school year are consistent and compatible with those extracted in the study by Viner et al.16 which associated the prevalence of infections in the school environment with the incidence of infection in the community population. In the first trimester of the 2021/2022 school year, the highest number of cases of COVID-19 in pupils were in the 2nd to 6th year of primary school, followed by cases in the 2nd cycle of infants and 1st year of primary school; in contrast to the 2020/2021 period where they were in high school and secondary school pupils. The data described in this first period of the school year in our study are analogous to those recorded in the study by Gamboa Moreno et al.6 conducted in the Basque Country during the first wave (which saw a greater secondary attack rate in the higher educational stages such as baccalaureate compared to infants) and in the study by Alonso et al.5 The increased transmissibility of the delta variant circulating in the second period of the study was reflected in the 7-fold increase in the number of school outbreaks in the period 2021/2022 compared to the same period 2020/2021, which mostly occurred in the 2nd-6th years of primary school. In addition, potential superspreader events in the school setting were reported as outbreaks involving a high number of cases in trimesters of the school year (21 cases in one outbreak in the first trimester and 17 cases in another outbreak in the second trimester), where the social link coexisted with the school setting. Availability and access to vaccination against COVID-19 in our province for the group of teachers and school staff born between 1966 and 2003 occurred as of March 2021 (those born before 1966 had access by age group a few weeks later) and for those born between 2002 and 2009 as of August 2021, following Spain's COVID-19 Vaccination Strategy.11 In our province, the population vaccination coverage against COVID-19 (persons over 12 years of age) was 82.62% and 78.27% with 1 and 2 doses, respectively, at the beginning of the 2021/2022 school year (10 September 2020); as of 22 December 2021 it was 86.64 and 84.12% (1 and 2 doses),17 and in the subgroup of children aged 12 to 19 years in our province, the vaccination rates for the same dates were 81% for 1 dose and 37% for 2 doses; and 94% and 88% for 1 and 2 doses, chronologically. The vaccination status of students aged 12 years and older, teachers, and school staff with COVID-19 infection during the first quarter of the 2021/2022 school year was in line with the high population-based vaccination coverage at the provincial level (full or partial vaccination schedule in at least 85% of study cases, statistically significant values). Among COVID-19 cases, there was a significantly higher proportion of unvaccinated students over 12 years of age (11.7%) than teachers (4.8%) and other non-teaching school staff (8.3%). These differences did not reach statistical significance. In our study, cases of infection among teachers did not decrease in this second period with access to vaccination (vaccine approval from March 2021). We do not have sufficient data on vaccination coverage and the specific vaccination schedule received by this population subgroup. Limitations. This study has some limitations that should be considered when interpreting the results. The descriptive design of the study itself restricts the scope of the results extracted. At the same time, comparing the data between the two periods defined in a trimester conditions the representation of the total school year. It should also be borne in mind that the data were obtained from a register compiled by the Surveillance System of the Epidemiology Section of the Territorial Health Service of Valladolid, which involved pooling different sources of information, but it cannot be ruled out that confirmed cases in the school setting may have been lost. Conclusions This study shows a decrease in the incidence of SARS-CoV-2 infection in pupils in higher educational stages (secondary and baccalaureate) during the first trimester of the 2021/2022 school year (in which most were vaccinated) compared to the same period of the previous school year (in which they were unvaccinated). This is despite the increased transmissibility of the circulating delta variant, reflected in the increase in the number of outbreaks reported in schools in the first trimester of the school year compared to the previous year. However, in pupils from 2nd cycle infants to 6th year primary (not vaccinated in any of the trimesters studied), this incidence increased in both trimesters. Further studies are needed to analyse the epidemiological situation of COVID-19 infection in schools after vaccination was introduced in children aged 5–12 years and the booster doses recommended in the population together with the new infection control measures. Agreements. We would like to thank the COVID-19 team of the Provincial Directorate of Education of Valladolid for their work since the beginning of the pandemic and their collaboration with the Territorial Health Service. Ethical responsibilities. The study was evaluated and approved by the Ethics Committee of the Valladolid West Health Area. Study Ref.: 22-PI061. Funding No funding was received for this study. Authorship contribution. All the authors contributed intellectually to its elaboration, and have read and approved the final version of the submitted manuscript. Conflict of Interests The authors have no conflict of interests to declare. Appendix A Supplementary data Supplementary material Supplementary material Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.vacune.2023.06.002. ==== Refs Bibliografía 1 COVID-19 en Pediatría: valoración crítica de la evidencia. Comité/Grupo de Pediatría Basada en la Evidencia de la AEP y AEPap. 2022. 2 Viner R. Russell S. Saulle R. Croker H. Stansfield C. Packer J. School closures during social lockdown and mental health, health behaviours and well-being among children and adolescents during the first COVID-19 wave: A Systematic Review JAMA Pediatr. 176 4 2022 400 409 10.1001/jamapediatrics.2021.5840 35040870 3 de Sanidad Ministerio Consumo y Bienestar Social de España. Información científica-técnica COVID-19 en distintos entornos y grupos de personas. Actualización, 25 de marzo 2021 MSCBS 2021 [Internet]. Disponible en: https://www.sanidad.gob.es/profesionales/saludPublica/ccayes/alertasActual/nCov/documentos/Documento_GRUPOS_PERSONAS.pdf 4 European Centre for Disease Prevention and Control. COVID-19 in children and the role of school settings in transmission - second update ECDC 2021 [Internet]. Disponible en: https://www.ecdc.europa.eu/sites/default/files/documents/COVID-19-in-children-and-the-role-of-school-settings-in-transmission-second-update.pdf 5 Alonso S. Alvarez-Lacalle E. Català M. López D. Jordan I. García-García J.J. Age-dependency of the Propagation Rate of Coronavirus Disease 2019 Inside School Bubble Groups in Catalonia Spain. Pediatr Infect Dis J. 40 11 2021 955 961 10.1097/INF.0000000000003279 34321438 6 Gamboa Moreno E. Garitano Gutiérrez I. Portuondo Jiménez J. Cabrera Rodríguez A. Aldeguer Corbi J. Grupo de investigación de la Red de Vigilancia Casos y Contactos del COVID -19. Baja transmisión del SARS-CoV-2 en el ámbito escolar: estudio poblacional en Euskadi Rev. Esp Salud Pública. 95 2021 e202112196 7 Viner R.M. Mytton O.T. Bonell C. Melendez-Torres G.J. Ward J. Hudson L. Susceptibility to SARS-CoV-2 infection among children and adolescents compared with adults: a Systematic Review and meta-analysis JAMA Pediatr. 175 2 2021 143 156 10.1001/jamapediatrics.2020.4573 32975552 8 Powell A.A. Kirsebom F. Stowe J. McOwat K. Saliba V. Ramsay M.E. Effectiveness of BNT162b2 against COVID-19 in adolescents Lancet Infect Dis. 22 5 2022 581 583 10.1016/S1473-3099(22)00177-3 35325619 9 Glatman-Freedman A. Hershkovitz Y. Kaufman Z. Dichtiar R. Keinan-Boker L. Bromberg M. Effectiveness of BNT162b2 Vaccine in Adolescents during Outbreak of SARS-CoV-2 Delta Variant Infection, Israel, 2021 Emerg Infect Dis. 27 11 2021 2919 2922 10.3201/eid2711.211886 34570694 10 Tartof S.Y. Slezak J.M. Fischer H. Hong V. Ackerson B.K. Ranasinghe O.N. Effectiveness of mRNA BNT162b2 COVID-19 vaccine up to 6 months in a large integrated health system in the USA: a retrospective cohort study Lancet. 398 10309 2021 1407 1416 10.1016/S0140-6736(21)02183-8 34619098 11 de Sanidad Ministerio Consumo y Bienestar Social de España. Grupo de Trabajo Técnico de Vacunación COVID-19. Estrategia de vacunación frente a COVID-19 en España MSCBS 2021 [Internet].Disponible en: https://www.sanidad.gob.es/profesionales/saludPublica/ccayes/alertasActual/nCov/vacunaCovid19.htm 12 Ministerio de Sanidad, Consumo y Bienestar Social de España. Centro de Coordinación de Alertas y Emergencias Sanitarias. Vacuna Covid-19. Cuadro de mando resumen de datos de vacunación. MSCBS; 2021–2022. [Internet]. Disponible en: https://www.sanidad.gob.es/profesionales/saludPublica/ccayes/alertasActual/nCov/pbiVacunacionMobile.htm 13 Junta de Castilla y León. Portal de Educación. Dirección Provincial de Valladolid. COVID 19. Detección y seguimiento en centros educativos. Guía de actuación ante la aparición de casos de COVID-19 en centros educativos en Castilla y León Versión 1 de 9 septiembre de 2020, Versión 2 de 7 de octubre de 2020, Versión 3 de 26 de noviembre de 2020 y Versión 4 de 14 de septiembre de 2021. [Internet]. Disponible en: https://www.educa.jcyl.es/dpvalladolid/es/informacion-especifica-dp-valladolid/documentos-administrativos/covid-19-deteccion-seguimiento-centros-educativos. 14 Junta de Castilla y León. Datos abiertos de Castilla y León. JCyL; 2021–2022. [Internet]. Disponible en: https://datosabiertos.jcyl.es/web/es/datos-abiertos-castilla-leon.html 15 Instituto Nacional de Estadística. Cifras oficiales de población resultantes de la revisión del Padrón municipal a 1 de enero. Valladolid: Población por municipios y sexo. INE; 2020–2022. [Internet]. Disponible en: https://www.ine.es/jaxiT3/Tabla.htm?t=2904 16 Viner R. Waddington C. Mytton O. Booy R. Cruz J. Ward J. Transmission of SARS-CoV-2 by children and young people in households and schools: A meta-analysis of population-based and contact-tracing studies J Infect. 84 3 2022 361 382 10.1016/j.jinf.2021.12.026 34953911 17 Consejería de Transparencia, Ordenación del Territorio y Acción Exterior con la información suministrada por la Consejería de Sanidad de Castilla y León. Situación epidemiológica del coronavirus (COVID-19) en Castilla y León. JCyL; 2020–2022. [Internet]. Disponible en: https://analisis.datosabiertos.jcyl.es/pages/coronavirus/?seccion=vacunaciones-suministros.
PMC010xxxxxx/PMC10288311.txt	==== Front Vacunas Vacunas Vacunas 1576-9887 1578-8857 Elsevier España, S.L.U. S1576-9887(23)00055-9 10.1016/j.vacun.2023.06.005 Original Epidemiological study of vaccination against SARS-CoV-2 and its impact on COVID-19 progression in a cohort of patients in gran Canaria de Arriba Fernández Alejandro 12⁎ Bilbao José Luis Alonso 3 Francés Alberto Espiñeira 3 Mora Antonio Cabeza 3 Pérez Ángela Gutiérrez 3 Barreiros Miguel Ángel Díaz 3 1 Hospital General de Fuerteventura, 35600 Puerto del Rosario, Spain 2 Research Institute of Biomedical and Health Sciences, University of Las Palmas de Gran Canaria, 35001 Las Palmas de Gran Canaria, Spain 3 Gerencia de Atención Primaria de Gran Canaria, 35006, Las Palmas de Gran Canaria, Spain ⁎ Corresponding author. 23 6 2023 23 6 2023 17 11 2022 30 5 2023 17 6 2023 © 2023 Elsevier España, S.L.U. All rights reserved. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objectives. We analyzed the impact of age, sex, vaccination against COVID-19, immunosuppressive treatment, and comorbidities on patients' risk of requiring hospital admission or of death. Methods. Population-based observational retrospective study conducted on a cohort of 19,850 patients aged 12 years or more, who were diagnosed with COVID-19 between June 1st and December 31st, 2021, in the island of Gran Canaria. Results. Hypertension (18.5%), asthma (12.8%) and diabetes (7.2%) were the most frequent comorbidities; 147 patients died (0.7%). The combination of advanced age, male sex, cancer, coronary heart disease, immunosuppressive treatment, hospital admission, admission to the intensive care unit, mechanical ventilation and lack of complete COVID-19 vaccination or booster dose was strongly predictive of mortality (p < 0.05); 831 patients required hospital admission and it was more frequent in men, older age groups, and patients with cancer, diabetes, arterial hypertension, chronic obstructive pulmonary disease, congestive heart failure or immunosuppressive treatment. The COVID-19 vaccine booster dose was associated with a lower risk of death ([OR] 0.11, 95% CI 0.06–0.21, p < 0.05) or hospital admission ([OR] 0.36, 95% CI 0.29–0.46, p < 0.05). Conclusions. Cancer, coronary heart disease, and immunosuppressive treatment were associated with increased COVID-19 mortality. More complete vaccination was associated with lower risk of hospital admission or death. Three doses of the SARS-CoV-2 vaccine were highly associated with the prevention of death and hospital admission in all age groups. These findings suggest that COVID-19 vaccination can help bring the pandemic under control. Objetivos. Analizamos el impacto de la edad, el sexo, la vacunación frente a la COVID-19, el tratamiento inmunosupresor y las comorbilidades en el riesgo de los pacientes de precisar ingreso hospitalario o de fallecer. Métodos. Estudio retrospectivo observacional de base poblacional realizado sobre una cohorte de 19.850 pacientes de 12 años o más, que fueron diagnosticados de COVID-19 entre el 1 de junio y el 31 de diciembre de 2021, en la isla de Gran Canaria. Resultados. La hipertensión arterial (18,5%), el asma (12,8%) y la diabetes (7,2%) fueron las comorbilidades más frecuentes; Fallecieron 147 pacientes (0,7%). La combinación de edad avanzada, sexo masculino, cáncer, cardiopatía coronaria, tratamiento inmunosupresor, ingreso hospitalario, ingreso en unidad de cuidados intensivos, ventilación mecánica y la falta de vacunación completa contra el COVID-19 o dosis de refuerzo fue fuertemente predictiva de mortalidad (p < 0,05); 831 pacientes requirieron ingreso hospitalario y fue más frecuente en hombres, grupos de mayor edad y pacientes con cáncer, diabetes, hipertensión arterial, enfermedad pulmonar obstructiva crónica, insuficiencia cardiaca congestiva o tratamiento inmunosupresor. La dosis de refuerzo contra la vacuna del COVID-19 se asoció con un menor riesgo de muerte ([OR] 0.11, IC 95% 0.06–0.21, p < 0,05) o ingreso hospitalario ([OR] 0.36, IC 95% 0.29–0.46; p < 0,05). Conclusiones. El cáncer, la enfermedad coronaria y el tratamiento inmunosupresor se asociaron con una mayor mortalidad por COVID-19. Una vacunación más completa se asoció con un menor riesgo de hospitalización o muerte. Tres dosis de la vacuna contra el SARS-CoV-2 se asociaron a una mayor prevención de la muerte y el ingreso hospitalario relacionados con la COVID-19 en todos los grupos de edad. Estos hallazgos sugieren que la vacunación contra el COVID-19 puede ayudar a controlar la pandemia. Keywords COVID-19 SARS-CoV-2 Vaccines Hospitalization Mortality Palabras clave COVID-19 SARS-CoV-2 Vacunas Hospitalización Mortalidad ==== Body pmcIntroduction The respiratory infection caused by SARS-CoV-2 was first documented by the end of December 2019 in Wuhan [1], from where it spread globally, and caused a pandemic with unprecedented consequences [2]. As of June 22nd, 2022, there have been more than 545 million cases and more than 6,3 million deaths worldwide [3]. Although most patients with SARS-CoV-2 infections develop mild to moderate symptoms, those with severe respiratory failure requiring admission to the intensive care unit (ICU) are at a higher risk of morbidity and show higher mortality rates. [4., 5., 6.]. Several cohort studies have been published on the characteristics and outcomes of the COVID-19 pneumonia in such critical patients [4]. However, most of the studies published up to date have been conducted with patient cohorts from North America [6,7] and China [8], which may not represent the overall picture. In addition, many studies include a relatively small number of patients, which may hinder estimations of the outcome and burden of disease in these patients [9]. The characteristics of COVID-19 patients that require hospitalization in Gran Canaria are not well known. This study analyzed the impact of COVID-19 patients' age, sex, COVID-19 vaccination, immunosuppressive therapy, and comorbidities on the risk of requiring hospital or ICU admission and on the risk of death. Our hypothesis was that patients admitted for COVID-19 would show high morbidity and mortality rates and that pre-existing comorbid conditions would be associated with a high risk of death. Methods Design. Population-based, observational retrospective cohort study. Study area. A cohort of 19,850 patients who lived in the island of Gran Canaria, Spain. 876.200 people resided in this island at the time of the study. Eligibility criteria. The inclusion criteria were patients aged 12 years or more, who were diagnosed with COVID-19 between June 1st and December 31st, 2021, in the island of Gran Canaria. The exclusion criteria was as follows: age < 12 years. Definitions. Patients were classified as suffering from: diabetes, if they had basal glycemia ≥ 126 mg/dl or were on anti-diabetes treatment; obesity, if they had BMI ≥ 30 kg/m2; hypertension, if they had diastolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg or were on anti-hypertension treatment. The participants were defined as known OSAS when there was a previous sleep study and/or the initiation of treatment documented by a physician. Confirmed COVID-19 cases. Patients who met the clinical criteria for suspected COVID-19 and showed positive results in AIDT (active infection diagnostic test); or asymptomatic patients with positive AIDT plus negative or not undertaken IgG-test. Suspected COVID-19 cases: Patients with acute respiratory infection of sudden onset of any degree of severity, who presented with fever, cough or shortness of breath, among other signs. Further signs or symptoms like odynophagia, anosmia, ageusia, muscle pain, diarrhea, chest pain, headache and others were also considered as symptoms of suspected SARS-CoV-2 infection, depending on the doctor's criterion. Severe COVID-19 progression was defined as the need for hospital admission, ICU admission or mechanical ventilation. Complete vaccination schedule. Patients were considered to be fully vaccinated if 1) they had received 2 doses of the vaccine separated by a minimum of: 19 days if the first dose was BNT162b2 mRNA (Pfizer-BioNTech), 21 days if it was ChAdOx1 nCoV-19 (AstraZeneca-University of Oxford), or 25 days if it was mRNA-1273 (Moderna); and 2) if the minimum time elapsed since the last dose was: 7 days if the last dose was Pfizer, or 14 days if it was AstraZeneca or Moderna. Patients were also considered to be fully vaccinated if they had received one dose of Ad26.COV2.S (Janssen) more than 14 days before. Patients up to 65 years old, were also considered as fully vaccinated if they had passed the disease and subsequently received a dose of any of the vaccines, after the corresponding mentioned period for the second dose. Subjects vaccinated with a heterologous schedule consisting of a first dose of AstraZeneca and a second dose of an mRNA vaccine were considered as fully vaccinated after 7 days, if the second dose was Pfizer or 14 days if it was Moderna [10]. Variables. Personal history (asthma, cancer, dementia, diabetes, coronary heart disease, chronic obstructive pulmonary disease (COPD), auricular fibrillation (AF), hypertension, congestive heart failure (CHF), obesity), date of first, second and third doses of COVID-19 vaccine, type of COVID-19 vaccine (Pfizer, Moderna, AstraZeneca, Janssen), mechanical ventilation, hospital admission, ICU admission, immunosuppressive therapy, death. Data source and collection. The identification data of all patients who were vaccinated against COVID-19 in Gran Canaria (from December 28th, 2020, to December 31st, 2021) were obtained from REGVACU (the registry of vaccination against COVID-19 in Spain). The identification data of all COVID-19 cases in Gran Canaria that were notified to the Epidemiological Surveillance Network of the Canary Islands (REVECA), were obtained from the General Directorate of Public Health (DGSP) (period: June 1st, 2021, to December 31st, 2021). Post-vaccination COVID-19 cases reported to the DGSP were identified by combining both databases. The clinical information of patients diagnosed with COVID-19 was obtained from their Primary Care electronic medical records (DRAGO AP). DRAGO is the healthcare management system of the Canary Islands. Statistical analysis. A descriptive analysis of the results was carried out using frequency and percentages for categorical variables; and mean and standard deviation (SD) for analytical determinations or quantitative variables. Bivariate analysis of qualitative variables was carried out with the χ2 test, using the Likelihood Ratio when necessary. In addition to the bivariate analysis, a multivariable logistic regression model adjusting for predefined covariates was used to estimate the propensity scores for cohort participants. The models were used to determine the predictive values of death and hospitalization, which were defined as the dependent categorical variables in the analysis, adjusted by age, sex, immunosuppressive treatment, type of COVID-19 vaccine, complete vaccination schedule, booster dose (3rd dose) by age and comorbidities, including diabetes, coronary heart disease, auricular fibrillation, hypertension, COPD, asthma, CHF, cancer, obesity, OSAS and dementia. Statistical significance was established at 5% (p < 0.05), and the level of confidence was set at 95%. Data were analyzed with the Statistical Package for the Social Sciences (SPSS) v20 and Microsoft® Excel (2010). Informed Consent Statement. This study was approved by the Ethics Committee for Research of the University Hospital of Gran Canaria Dr. Negrín (registration number 2021–355-1 COVID19) and it was compliant with the local laws and regulations, the Declaration of Helsinki, and the Good Clinical Practices. Patient consent was waived due to anonymization/dissociation of patient data and the results did not affect the clinical management of patients. Results The study included 19,850 patients diagnosed with COVID-19 between June and December, 2021, of whom 10,505 (52.9%) were women. The mean age was 40.7 years (SD 17.7), women being older than men on average (41.1 years vs. 40.2 years). The predominant age group in the sample was 18 to 49 years; 35% of patients presented some risk factor; the most frequent one was hypertension (18.5%), followed by asthma (12.8%) and diabetes (7.2%); 4.2% of patients were admitted to hospital; 1.1% required admission to the ICU, 0.3% required mechanical ventilation and 0.7% died. Between June 1st and December 31st, 2021, a total of 513,295 subjects were vaccinated in Gran Canaria. The mean time elapsing between the completion of the vaccination schedule and the diagnóstico of COVID-19 was 135.7 days (SD 64.7). The mean age of hospitalized COVID-19 patients was 60.5 years (interquartile range 44–67 years). More than half of them were men (54.8%, 455/831); 85.7% (712/831) were discharged to home; and 14.3% (119/831) died. The mean age of COVID-19 patients admitted to the ICU was 55.3 years (interquartile range 46–76 years). More than half of them were men (67.3%, 140/208); 85.6% (178/208) were discharged to home; and 14.4% (30/208) died. In survivors discharged to home and deceased patients, the proportion of subjects that needed ICU care or invasive mechanical ventilation was higher among 50-or-older ones than among 12–17 or 18–49 year-old ones; 0.46% (14/3051) of patients who had received a booster dose died, as compared to 0.80% (133/16,652) of those who had not (< 0.05). The mean hospital stay was 12.5 days (median 8 days) with a maximum of 130 days. The mean ICU stay was 13.5 days (median 9 days) with a maximum of 83 days. Subjects who had received mRNA vaccines (Pfizer/ BioNTech or Moderna) were at lower risk of needing hospital admission, ICU care or mechanical ventilation (p < 0.05) than those who were not vaccinated. No differences were found between the Janssen and Astrazeneca vaccines (p > 0.05). Table 1 illustrates the impact of patient's age, sex, COVID-19 vaccination, immunosuppressive treatment or comorbidities on the risk of requiring hospital admission, ICU care or mechanical ventilation. The risk of death was lower in patients vaccinated with Moderna or Janssen (p < 0.05) than in those who were not vaccinated (Table 2 ).Table 1 Bivariate analysis. Associations with hospital admission, admission to the ICU and mechanical ventilation. Table 1Variables N cases N admitted to hospital Odds Ratio (I.C) N admitted to the ICU Odds Ratio (I.C) N Mechanical ventilation Odds Ratio (I.C) Sex Men 9344 455 1 140 1 35 1 Women 10,506 376 0.73 (0.63–0.83) 68 0.43 (0.32–0.57) 25 0.63 (0.38–1.06) Complete COVID-19 vaccination Si 13,676 468 0.57 (0.49–0.65) 99 0.41 (0.31–0.53) 32 0.39 (0.24–0.65) No 6174 363 1 109 1 28 1 Age groups 12–49 13,884 249 0.67 (0.56–0.80) 74 0.18 (0.12–0.27) 29 0.35 (0.16–0.77) 50–69 4619 294 0.25 (0.21–0.30) 95 0.70 (0.48–1.03) 23 0.84 (0.37–1.88) > = 70 1347 288 1 39 1 8 1 Type of vaccine (2° dose) or mono-dose of Janssen Pfizer 8294 272 0.54 (0.46–0.64) 37 0.25 (0.17–0.36) 14 0.33 (0.17–0.61) Moderna 2379 46 0.32 (0.23–0.43) 14 0.33 (0.19–0.58) 3 0.24 (0.07–0.79) AstraZeneca 1613 78 0.81 (0.63–1.05) 29 1.02 (0.67–1.54) 10 1.20 (0.59–2.44) Janssen 1390 72 0.88 (0.67–1.13) 19 0.77 (0.47–1.26) 1 0.14 (0.02–1.01) Not vaccinated 6174 363 1 109 1 32 1 Asthma Yes 2548 101 1.4 (0.69–2.69) 15 1 10 1.4 (0.69–2.69) No 17,302 730 1 193 0.53 (0.31–0.89) 50 1 Cancer Yes 627 104 5.06 (4.05–6.3) 16 2.6 (1.55–4.35) 59 0.52 (0.07–3.75) No 19,223 627 1 192 1 1 1 Dementia Yes 49 26 16.1 (9.75–26.7) 0 0.99 (0.990.99) 0 1 (1–1) No 19,654 805 1 208 1 60 1 Diabetes Yes 1429 220 5.3 (4.50–6.25) 40 3.12 (2.21–4.44) 11 2.91 (1.51–5.61) No 18,421 611 1 168 1 49 1 Coronary heart disease Yes 502 104 6.69 (5.33–8.41) 13 2.61 (1.48–4.61) 2 1.33 (0.32–5.46) No 19,348 727 1 195 1 58 1 Chronic obstructive pulmonary disease Yes 299 72 7.85 (5.97–10.34) 10 3.38 (1.77–6.45) 5 6.03 (2.40–15.17) No 19,551 759 1 198 1 55 1 Atrial fibrillation Yes 282 72 8.50 (6.44–11.2) 9 3.21 (1.63–6.32) 2 2.4 (0.58–9.89) No 19,568 759 1 199 1 58 1 Hypertension Yes 3669 406 4.61 (4–5.31) 94 3.71 (2.81–4.88) 20 2.21 (1.29–3.79) No 16,181 425 1 114 1 40 1 Congestive heart failure Yes 177 63 13.6 (9.92–18.66) 3 1.64 (0.52–5.17) 1 1.89 (0.26–13.71) No 19,673 678 1 205 1 59 1 Obesity Yes 262 17 1.6 (0.97–2.63) 3 1.1 (0.35–3.45) 1 1.27 (0.18–9.19) No 19,588 814 1 205 1 59 1 Booster Yes 3065 144 1.16 (0.96–1.39) 16 0.45 (0.27–0.76) 7 0.72 (0.33–1.59) No 16,785 687 1 192 1 53 1 Table 2 Bivariate analysis. Associations with mortality in the 147 deaths in a 19.850 population. Table 2Variables N cases N deaths (%) Odds Ratio (I.C) P value Sex Men 9344 80 (0.86) 1 0.073 Women 10,506 67 (0.64) 0.74 (0.54–1.03) Complete COVID-19 vaccination Yes 13,676 99 (0.72) 0.93 (0.66–1.32) 0.684 No 6174 48 (0.78) 1 Age groups 18–49 12,473 7 (0.06) 0.01 (0.00–0.01) 0.000 50–69 4596 39 (0.85) 0.11 (0.07–0.15) 0.000 > = 70 1339 101 (7.54) 1 Ref Type of vaccine (2° dose) Pfizer 8294 80 (0.96) 1.24 (0.87–1.78) 0.235 Moderna 2379 8 (0.34) 0.43 (0.20–0.91) 0.028 AstraZeneca 1613 8 (0.50) 0.64 (0.30–1.35) 0.237 Janssen 1390 3 (0.22) 0.28 (0.09–0.89) 0.031 Not vaccinated 6174 48 (0.78) 1 Ref Booster (3rd dose) Yes 3065 14 (0.46) 0.58 (0.33–0.99) 0.046 No 16,785 133 (0.79) Asthma Yes 2548 18 (0.71) 0.95 (0.58–1.55) 0.830 No 17,302 129 (0.75) 1 Cancer Yes 627 46 (7.33) 14.99 (10.5–21.5) 0.000 No 19,223 101 (0.53) 1 Dementia Yes 49 15 (30.61) 45.6 (24.9–83.3) 0.000 No 19,654 132 (0.67) 1 Diabetes Yes 1429 51 (3.57) 7.1 (5.01–10) 0.000 No 18,421 96 (0.52) 1 Coronary heart disease Yes 502 42 (8.37) 16.73 (11.56–24.22) 0.000 No 19,348 105 (0.54) 1 Chronic obstructive pulmonary disease Yes 299 20 (6.69) 10.96 (6.74–17.82) 0.000 No 19,551 127 (0.65) 1 Atrial fibrillation Yes 282 22 (7.8) 13.16 (8.23–21.05) 0.000 No 19,568 125 (0.64) 1 Hypertension Yes 3669 98 (2.67) 9.04 (6.4–12.76) 0.000 No 16,181 49 (0.30) 1 Congestive heart failure Yes 177 24 (13.56) 24.93 (15.66–39.71) 0.000 No 19,673 123 (0.63) 1 Obesity Yes 262 2 (0.76) 1.03 (0.25–4.19) 0.965 No 19,588 145 (0.74) 1 Mechanical ventilation Yes 60 10 (16.7) 28.69 (14.25–57.74) 0.000 No 19,790 137 (0.69) 1 Admission to the ICU Yes 208 30 (14.4) 28.13 (18.34–43.12) 0.000 No 19,642 117 (0.60) 1 Hospital admission Yes 831 119 (14.3) 113.36 (74.6–172.26) 0.000 No 19,019 28 (0.15) 1 Immunosupressive treatment Yes 330 19 (5.76) 9.26 (5.64–15.18) 0.000 No 19,520 128 (0.66) 1 A multivariate logistic regression analysis (Table 3, Table 4 ) revealed that older patients, men, subjects with personal history issues (cancer or coronary heart disease), and those under immunosuppressive treatment were more likely to develop severe post-vaccine COVID-19 (often requiring hospital admission, ICU care or mechanical ventilation) or to die (p < 0.05). The COVID-19 vaccine booster dose was associated with a lower risk of death ([OR] 0.11, 95% CI 0.06–0.21, p < 0.05) or hospital admission ([OR] 0.36, 95% CI 0.29–0.46, p < 0.05). No association was found with asthma, obesity, hypertension or diabetes (p > 0.05).Table 3 Death according to gender and age, and association with obesity, diabetes, hypertension, cancer, coronary heart disease, COPD, CHD and dementia in 110,726 in-patients positive for SARS-CoV-2. Table 3Variable Death and multivariate analysis adjusted (95% CI) P value Sex Women 0.66 (0.43–1.02) 0.062 Men 1 (Ref.) Age Years 1.11 (1.09–1.13) 0.000 Asthma Yes 1.14 (0.59–2.20) 0.693 No 1 (Ref.) Cancer Yes 3.85 (2.35–6.33) 0.000 No 1 (Ref.) Diabetes Yes 0.81 (0.51–1.28) 0.363 No 1 (Ref.) Coronary heart disease Yes 1.56 (0.90–2.68) 0.110 No 1 (Ref.) Chronic obstructive pulmonary disease Yes 0.60 (0.30–1.23) 0.163 No 1 (Ref.) Atrial fibrillation Yes 0.95 (0.49–1.86) 0.884 No 1 (Ref.) Hypertension Yes 1.10 (0.69–1.74) 0.695 No 1 (Ref.) Congestive heart failure Yes 1.33 (0.65–2.73) 0.435 No 1 (Ref.) Obesity Yes 0.70 (0.14–3.50) 0.663 No 1 (Ref.) Dementia Yes 1.91 (0.76–4.80) 0.171 No 1 (Ref.) Immunosupressive treatment Yes 3.72 (1.91–7.27) 0.000 No 1 (Ref.) Complete vaccination Yes 0.72 (0.45–1.15) 0.164 No 1 (Ref.) Booster (3rd dose) Yes 0.11 (0.06–0.21) 0.000 No 1 (Ref.) Hospital admission Yes 13.39 (7.99–22.42) 0.000 No 1 (Ref.) Admission to the ICU Yes 1.93 (1.03–3.64) 0.041 No 1 (Ref.) Mechanical ventilation Yes 4.53 (1.64–12.53) 0.004 No 1 (Ref.) Table 4 Hospitalization according to gender and age, and association with obesity, diabetes, hypertension, cancer, coronary heart disease, COPD, CHD and dementia in 110,726 in-patients positive for SARS-CoV-2. Table 4Variable Death and multivariate analysis adjusted (95%CI) P value Sex Men 0.66 (0.56–0.77) 0.000 Women 1 (Ref.) Age years 1.07 (1.06–1.08) 0.000 Asthma Yes 1.15 (0.90–1.47) 0.257 No 1 (Ref.) Cancer Yes 1.54 (1.18–2.03) 0.002 No 1 (Ref.) Diabetes Yes 1.61 (1.31–1.98) 0.000 No 1 (Ref.) Coronary heart disease Yes 1.21 (0.90–1.61) 0.200 No 1 (Ref.) Chronic obstructive pulmonary disease Yes 1.72 (1.23–2.41) 0.002 No 1 (Ref.) Atrial fibrillation Yes 1.33 (0.94–1.89) 0.113 No 1 (Ref.) Hypertension Yes 1.38 (1.15–1.66) 0.001 No 1 (Ref.) Congestive heart failure Yes 1.89 (1.26–2.83) 0.002 No 1 (Ref.) Obesity Yes 1.29 (0.74–2.24) 0.379 No 1 (Ref.) Dementia Yes 1.98 (1.07–3.65) 0.030 No 1 (Ref.) Immunosupressive treatment Yes 4.62 (3.32–6.43) 0.000 No 1 (Ref.) Complete vaccination Yes 0.20 (0.17–0.24) 0.000 No 1 (Ref.) Booster (3rd dose) Yes 0.36 (0.29–0.46) 0.000 No 1 (Ref.) Discussion This study is the study with the largest number of subjects up to date, to describe the clinical and epidemiological characteristics of hospitalized COVID-19 patients in Gran Canaria. The data, corresponding to the last 7 months (June to December 2021) illustrate the new reality of the disease in a population with high vaccination rates. The main findings of our study showed that more complete vaccination was associated with less frequent risk of death or hospital admission. These findings suggest that COVID-19 vaccination can help in bringing the pandemic under control. Factors associated with greater probability of requiring hospital admission included: older age, male sex, diabetes, hypertension, cancer, CHF, COPD, and immunosuppressive treatment. These findings can help healthcare professionals identify patients at higher risk of hospitalization, who may require closer monitorization and care, and those who may benefit from specific preventive or therapeutic interventions. Among patients who required hospital admission, mortality was 14.3%. The mortality rate in this study was lower than the rates reported for other hospitalized patient cohorts, in earlier studies (approximately 15% to > 20%) [10., 11., 12., 13., 14., 15.]. Our results showed that 25% (95%CI 22.2–28.1%) of hospitalized patients required ICU admission. The mortality rate for these patients was 14.5% (95%CI 10.3–19.8%). These findings are in agreement with those of a meta-analysis published by Rodríguez et al., in which 20.3% (95%CI 10.0–30.6%) required ICU admission and the mortality rate was 13.9% (95%CI 6.2–21.5%) [16]. In our analysis, the magnitude of the risk of hospitalization for COVID-19 was lower in patients with asthma, obesity, dementia, auricular fibrillation, and coronary heart disease than in those with other medical conditions (e.g., COPD). This finding is in line with the results of Aveyard et al., who showed that the risk of severe COVID-19 in people with asthma was relatively low. Subjects with COPD appeared to have a moderately higher risk of suffering a severe illness or requiring hospital admission, but their risk of death from COVID-19 at the height of the pandemic was generally lower than the normal risk of death from any cause [17]. Although it is considered a risk factor for the acquisition of COVID-19, the role of immunosuppression after a SARS-CoV-2 infection has not been extensively studied. In this study, receiving immunosuppressive therapy before COVID-19 diagnóstico was identified as a unique risk factor for hospital admission, ICU admission, mechanical ventilation, and death from COVID-19; in agreement with Akama-Garren et al. [18], who found that the use of immunosuppressive treatment could be associated with a slightly increased risk of severe COVID-19 or death. These findings demonstrate that COVID-19 is more severe in patients who are already taking immunosuppressive medication and emphasize the need of providing aggressive monitoring and supportive care to immunocompromised patients diagnosed with COVID-19 [19]. Subjects with comorbidities and older subjects (who often present comorbidities) are especially vulnerable to acquire acute COVID-19 infection and to meet the criteria for severity during the acute phase, with the consequent aftereffects for survivors. Increased morbidity and mortality in older patients and in patients with comorbidities have been associated with both comorbidities and frailty, which entail poorer immune response [20]. There may also be protective factors for the post-COVID-19 condition, since the results of a recent study suggested that vaccines may offer protection [21]. In our study, receiving two doses of the COVID-19 vaccine was associated with a decreased risk of death. Arbel et al. demonstrated that participants who had received a booster dose at least 5 months after a second dose of Pfizer-BioNTech had 90% less short-term Covid-19 mortality than participants who did not receive a booster [22]. Complete COVID-19 vaccination was significantly less frequent than no-vaccination among patients with outcomes of hospital admission, mechanical ventilation or death. These findings are consistent with the literature [23,24]. A systematic review revealed that patients undergoing cancer treatment, such as chemotherapy, had a higher risk of death from COVID-19 [25]. In our study, the overall prevalence of active cancer as a comorbidity was 3.2%, and it was an independent factor associated with mortality in a multivariate analysis. Other authors like Xiaochen Li et al. provided substantial statistical evidence for the value of coronary heart disease as a predictor of COVID-19 like a risk factor for severe cases on admission [26]. Our findings showed that diabetes and hypertension were comorbidities associated with an increased risk of hospitalization. These findings are consistent with those of Cascella et al. who concluded that 49% of the cases that required ICU admission for COVID-19 suffered from pre-existing comorbidities; and with the results of Mughal et al. who showed that comorbidities such as obesity, diabetes or hypertension increased the severity and the mortality rates (10.5% with comorbidity vs. 0.9% without comorbidity) [27,28]. CHF was associated with an increased risk of hospital admission in a multivariate analysis. Angeli et al. also found that this condition was an independent predictor of adverse prognosis and death in COVID-19 patients [29]. The severity of COVID-19 has changed through the successive epidemic waves, likely due to increasing population immunity (caused both by vaccination and by ongoing virus circulation) and possibly to a different intrinsic virulence of SARS-CoV-2 variants. While Delta variant was generally (though inconsistently) associated to increased risk of severe disease compared to the previously dominant Alpha variant, results from different countries have pointed to a lower severity of Omicron. On the other hand, vaccine effectiveness against severe COVID-19 with Delta variant was found well preserved compared to Alpha, but evidence for severe COVID-19 with Omicron is less consistent. There are contextual factors that may affect the estimates of variant severity as well as the variant-specific vaccine effectiveness, such as the intensity of previous circulation of other SARS-CoV-2 variants in the territory or particular characteristics of the COVID-19 vaccination rollout [30]. In the period in which this study was carried out, the Alpha variant was the dominant one in Gran Canaria until week 22 of the year 2021, until later the Delta variant became dominant until week 50 of the same year. Finally, the Ómicron variant of COVID-19 was the dominant one in Gran Canaria, accounting for 54.8% of infections at the end of 2021 and coinciding with the final period of this study [31]. Our study has some limitations. The description of severe cases is limited by the reduced number of patients in this category. There are also certain epidemiological limitations to be considered when interpreting the data like the heterogeneity of samples in terms of age (younger vs. older age groups), or severity of COVID-19 (patients with mild forms vs. patients admitted to hospital or to the ICU). The main strength of this study is the size of the sample, which consisted of a large number of participants, much higher than most of the Spanish studies on this subject. Receiving a complete vaccination schedule against SARS-CoV-2 and a booster dose effectively reduced hospitalization and death from COVID-19. These findings highlight the benefits of providing SARS-CoV-2 immunization including a support dose for a complete vaccination. In conclusion, cancer, coronary heart disease, and immunosuppressive treatment were associated with increased COVID-19 mortality. More complete vaccination was associated with lower risk of hospital admission and death. Two doses of the SARS-CoV-2 vaccine were highly effective in preventing COVID-19-related deaths and hospital admission in all age groups. Data Availability Statement. The data are not publicly available due to privacy or ethical reasons. Data are available from the management of Primary Care of Gran Canaria, Spain, for researchers who meet the criteria for access to confidential data. Conflicts of interest The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. Sources of support. We would like to thank the following entities for collaboration and funding: Fundación DISA and Fundación Española de Calidad Asistencial, without whom this study would not have been carried out. We would also like to thank to all the people who voluntarily participated in the study. Authors' participation. All the authors participated in the design of the study, data collection and preparation of the manuscript, and they declare that they approve its final version and are publicly responsible for its content. ==== Refs References 1. Karami Z. Knoop B.T. Dofferhoff A.S.M. Blaauw M.J.T. Janssen N.A. van Apeldoorn M. Few bacterial co-infections but frequent empiric antibiotic use in the early phase of hospitalized patients with COVID-19: results from a multicentre retrospective cohort study in The Netherlands Infect Dis (Lond) 53 2021 102 110 10.1080/23744235.2020.1839672 33103530 2. Rawson T.M. Moore L.S.P. Castro-Sanchez E. Charani E. Davies F. Satta G. COVID-19 and the potential long-term impact on antimicrobial resistance J Antimicrob Chemother 75 2020 1681 1684 10.1093/jac/dkaa194 32433765 3. WHO Coronavirus (COVID-19) Dashboard. Available online: https://covid19.who.int (accessed on 22 June 2022). 4. Nassar Y. Mokhtar A. Elhadidy A. Elsayed M. Mostafa F. Rady A. Outcomes and risk factors for death in patients with coronavirus disease-2019 (COVID-19) pneumonia admitted to the intensive care units of an Egyptian University Hospital. A retrospective cohort study J Infect Public Health 14 10 2021 1381 1388 10.1016/j.jiph.2021.06.012 34215561 5. Grimaldi D. Aissaoui N. Blonz G. Carbutti G. Courcelle R. Gaudry S. Characteristics and outcomes of acute respiratory distress syndrome related to COVID-19 in Belgian and French intensive care units according to antiviral strategies: the COVADIS multicentre observational study Ann Intensive Care 10 1 2020 131 10.1186/s13613-020-00751-y 33025225 6. Flythe J.E. Assimon M.M. Tugman M.J. Chang E.H. Gupta S. Shah J. Characteristics and outcomes of individuals with pre-existing kidney disease and COVID-19 admitted to intensive care units in the United States Am J Kidney Dis 77 2 2021 190 203.e1 10.1053/j.ajkd.2020.09.003 32961244 7. Auld S.C. Caridi-Scheible M. Blum J.M. Robichaux C. Kraft C. Jacob J.T. ICU and ventilator mortality among critically ill adults with coronavirus disease 2019 Crit Care Med 48 9 2020 e799 e804 10.1097/CCM.0000000000004457 32452888 8. Zhang H. Zhang Y. Wu J. Li Y. Zhou X. Li X. Risks and features of secondary infections in severe and critical ill COVID-19 patients Emerg Microbes Infect 9 1 2020 1958 1964 10.1080/22221751.2020.1812437 32815458 9. Nadkarni A. Alderson S. Collett L. Maiden M. Reddi B. Sundararajan K. Impact of COVID-19 on an Australian intensive care unit: lessons learned from South Australia Intern Med J 50 9 2020 1146 1150 10.1111/imj.14963 32761863 10. Ministerio de Sanidad. Estrategia de Vacunación Frente a COVID-19 En España. Available online: https://www.mscbs.gob.es/ profesionales/saludPublica/prevPromocion/vacunaciones/covid19/docs/COVID19_Actualizacion8_EstrategiaVacunacion.pdf (accessed on 11 May 2022). 11. Docherty AB, Harrison EM, Green CA, Hardwick HE, Pius R, Norman L, et al. ISARIC4C Investigators. Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study. BMJ. 2020 May 22;369:m1985. DOI:10.1136/bmj.m1985. 12. Petrilli CM, Jones SA, Yang J, Rajagopalan H, O'Donnell L, Chernyak Y, et al. Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study. BMJ. 2020 May 22;369:m1966. DOI:10.1136/bmj.m1966. 13. Matsunaga N. Hayakawa K. Terada M. Ohtsu H. Asai Y. Tsuzuki S. Clinical Epidemiology of Hospitalized Patients With Coronavirus Disease 2019 (COVID-19) in Japan: Report of the COVID-19 Registry Japan Clin Infect Dis 73 11 2021 e3677 e3689 10.1093/cid/ciaa1470 32986793 14. Wiersinga W.J. Rhodes A. Cheng A.C. Peacock S.J. Prescott H.C. Pathophysiology, transmission, diagnosis, and treatment of coronavirus disease 2019 (COVID-19): a review JAMA 324 2020 782 793 10.1001/jama.2020.12839 32648899 15. Zhou F. Yu T. Du R. Fan G. Liu Y. Liu Z. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study Lancet 395 2020 1054 1062 10.1016/S0140-6736(20)30566-3 32171076 16. Rodriguez-Morales A.J. Cardona-Ospina J.A. Gutiérrez-Ocampo E. Villamizar-Peña R. Holguin-Rivera Y. Escalera-Antezana J.P. Clinical, laboratory and imaging features of COVID-19: A systematic review and meta-analysis Travel Med Infect Dis 34 2020 101623 10.1016/j.tmaid.2020.101623 17. Aveyard P. Gao M. Lindson N. Hartmann-Boyce J. Watkinson P. Young D. Association between pre-existing respiratory disease and its treatment, and severe COVID-19: a population cohort study Lancet Respir Med 9 8 2021 909 923 10.1016/S2213-2600(21)00095-3 33812494 18. Akama-Garren E.H. Li J.X. Prior immunosuppressive therapy is associated with mortality in COVID-19 patients: A retrospective study of 835 patients J Med Virol 93 10 2021 5768 5776 10.1002/jmv.27105 34042195 19. Remy K.E. Mazer M. Striker D.A. Ellebedy A.H. Walton A.H. Unsinger J. Severe immunosuppression and not a cytokine storm characterizes COVID-19 infections JCI Insight 5 17 2020 e140329 10.1172/jci.insight.140329 20. Sociedad Española de Médicos Generales y de Familia (SEMG) Guía clínica para el paciente Long COVID/COVID. 2021 [consultado 23 Dic 2021]. Available online: https://www.semg.es/index.php/consensos-guias-y-protocolos/363-guia-clinica-para-la-atencion-al-paciente-long-covid-covid-persistente. 21. Antonelli M. Penfold R.S. Merino J. Sudre C.H. Molteni E. Berry S. Risk factors and disease profile of post-vaccination SARS-CoV-2 infection in UK users of the COVID Symptom Study app: a prospective, community-based, nested, case-control study Lancet Infect Dis 22 1 2022 43 55 10.1016/S1473-3099(21)00460-6 Jan 34480857 22. Arbel R. Hammerman A. Sergienko R. Friger M. Peretz A. Netzer D. BNT162b2 Vaccine Booster and Mortality Due to Covid-19 N Engl J Med 385 26 2021 2413 2420 10.1056/NEJMoa2115624 34879190 23. Tenforde M.W. Self W.H. Adams K. Gaglani M. Ginde A.A. McNeal T. Association Between mRNA Vaccination and COVID-19 Hospitalization and Disease Severity JAMA. 326 20 2021 2043 2054 10.1001/jama.2021.19499 34734975 24. Lopez Bernal J. Andrews N. Gower C. Robertson C. Stowe J. Tessier E. Effectiveness of the Pfizer-BioNTech and Oxford-AstraZeneca vaccines on covid-19 related symptoms, hospital admissions, and mortality in older adults in England: test negative case-control study BMJ. 373 2021 n1088 Published 2021 May 13 10.1136/bmj.n1088 25. Yekedüz E. Utkan G. Ürün Y. A systematic review and meta-analysis: the effect of active cancer treatment on severity of COVID-19 Eur J Cancer 141 2020 92 104 10.1016/j.ejca.2020.09.028 33130550 26. Li X. Xu S. Yu M. Wang K. Tao Y. Zhou Y. Risk factors for severity and mortality in adult COVID-19 inpatients in Wuhan J Allergy Clin Immunol 146 1 2020 110 118 10.1016/j.jaci.2020.04.006 32294485 27. Cascella M, Rajnik M, Cuomo A, Dulebohn S, Di Napoli R. Features, Evaluation, and Treatment of Coronavirus (COVID-19). In: StatPearls. Treasure Island (FL): StatPearls Publishing; January 9, 2023. 28. Mughal M.S. Kaur I.P. Jaffery A.R. Dalmacion D.L. Wang C. Koyoda S. COVID-19 patients in a tertiary US hospital: Assessment of clinical course and predictors of the disease severity Respir Med 172 2020 106130 10.1016/j.rmed.2020.106130 29. Angeli F. Marazzato J. Verdecchia P. Balestrino A. Bruschi C. Ceriana P. Joint effect of heart failure and coronary artery disease on the risk of death during hospitalization for COVID-19 Eur J Intern Med 89 2021 81 86 10.1016/j.ejim.2021.04.007 33933339 30. Varea-Jiménez E, Aznar Cano E, Vega-Piris L, Martínez Sánchez EV, Mazagatos C, García San Miguel Rodríguez-Alarcón L, et al. Comparative severity of COVID-19 cases caused by Alpha, Delta or Omicron SARS-CoV-2 variants and its association with vaccination. Enferm Infecc Microbiol Clin (Engl Ed). 2023 Feb 1:S2529-993X(23)00039–4. DOI:10.1016/j.eimce.2022.11.021 31. Ministerio de Sanidad. Actualización de la situación epidemiológica de las variantes de SARS-CoV-2 en España a 14 de junio de 2022. Consultado el 30 de mayo de 2023. Disponible en: https://www.sanidad.gob.es/profesionales/saludPublica/ccayes/alertasActual/nCov/documentos/COVID19_Actualizacion_variantes_20220614.pdf
PMC010xxxxxx/PMC10288312.txt	==== Front Rev Port Cardiol Rev Port Cardiol Revista Portuguesa De Cardiologia 0870-2551 2174-2030 Published by Elsevier España, S.L.U. on behalf of Sociedade Portuguesa de Cardiologia. S0870-2551(23)00350-5 10.1016/j.repc.2023.06.003 Article Reply to: RAAS inhibitors in COVID-19: not all are created equal! Resposta a: Inibidores de RAAS no COVID-19: nem todos são iguais!Gonçalves Jorge ab⁎ Santos Catarina D. a Fresco Paula ab Fernandez-Llimos Fernando ab a Laboratório de Farmacologia, Faculdade de Farmácia, Universidade do Porto, Portugal b Mechanistic Pharmacology and Pharmacotherapy Unit, UCIBIO-i4HB, Faculty of Pharmacy, University of Porto, Porto, Portugal ⁎ Corresponding author 23 6 2023 23 6 2023 © 2023 Published by Elsevier España, S.L.U. on behalf of Sociedade Portuguesa de Cardiologia. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc
PMC010xxxxxx/PMC10288313.txt	==== Front Vacunas (English Edition) 2445-1460 2445-1460 Published by Elsevier España, S.L.U. S2445-1460(23)00034-1 10.1016/j.vacune.2023.06.001 Original Article Factores asociados a la intención de vacunación contra el COVID-19 en Popayán, Cauca, Colombia Associated factors in the intention of vaccination against COVID-19, in Popayán, Cauca Colombia Valencia Omar Andrés Ramos a⁎ Gonzalez Yuliana Buitrón b Daza Jorge Sotelo c Villaquiran Andrés Felipe a a Departamento de Fisioterapia. Universidad del Cauca. Cauca –, Colombia. b Vicerrectoría de Investigaciones. Universidad del Cauca. Cauca –, Colombia. c Departamento de Enfermería. Universidad del Cauca. Cauca –, Colombia. ⁎ Autor para correspondencia. 23 6 2023 23 6 2023 © 2023 Published by Elsevier España, S.L.U. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Vaccination against the Covid-19 pandemic, decreed by the WHO in 2020, has shown in the initial trials an admissible efficacy for the scientific community, but with many doubts and concerns for the communities, developing the phenomenon known as vaccine hesitancy. Objective: to understand the factors associated with the intention or rejection of vaccination against COVID-19 in the city of Popayán in the year 2022. Methodology: Cross-sectional descriptive-analytical study, carried out between August 2021 and March 2022; with a non-probabilistic sampling, for convenience, with a sample size of 993 people; A questionnaire-type survey was applied in person and virtually to know the intention of vaccination, knowledge and perceptions. Results: The surveyed population was characterized as 56.19% female, 49.24% between 18 and 28 years old; 23.16% state that they do not intend to be vaccinated against COVID-19, the main reasons being: not being well informed 56.29%, ineffective vaccine 54.8% and that the vaccine weakens the immune system 27 ,5%; as well as the low confidence with the Vaccination Plan and with the pharmaceutical companies that produce the vaccine. Conclusion: The intention to vaccinate against COVID-19 is determined not only by the technical-administrative dynamics of the immunization program and the health system, variables of the context and the perception of risk, add up to explain the vaccination processes. La vacunación contra la pandemia Covid -19 decretada por la OMS en el 2020, ha demostrado en los ensayos iniciales una eficacia admisible para la comunidad científica, pero con muchas dudas e inquietudes para las comunidades, desarrollando el fenómeno conocido como vacilación vacunal. Objetivo: comprender los factores asociados a la intención o rechazo de vacunación contra el COVID-19 en la ciudad de Popayán en el año 2022. Metodología: Estudio transversal descriptivo-analítico, realizado entre agosto de 2021 y marzo de 2022; con un muestreo no probabilístico, por conveniencia, con un tamaño muestral de 993 personas; se aplicó una encuesta tipo cuestionario de forma presencial y virtual para conocer la intención de vacunación, conocimientos y percepciones. Resultados: La población encuestada se caracterizó por ser de género femenino 56,19%, se encuentra entre los 18 y 28 años el 49,24%; manifiestan no tener intención de vacunarse contra el COVID-19 es del 23,16%, siendo las principales razones: no estar bien informados 56,29%, vacuna no efectiva el 54,8% y que la vacuna debilita el sistema inmunológico el 27,5%; así como también se expresa la baja confianza con el Plan de vacunación y de las farmacéuticas que producen la vacuna. Conclusión: La intención de vacunación contra la COVID-19, está determinada no solo por las dinámicas técnico-administrativas del programa de inmunización y del sistema de salud, variables del contexto y de la percepción del riesgo, se suman para explicar los procesos de vacunación. Keywords Vaccination Coronavirus Infections (Covid-19) Risk factors Colombia Palabras clave Vacunación Infección por coronavirus (Covid-19) factores de riesgo Colombia ==== Body pmcIntroduction In 2019, severe acute respiratory syndrome 2 (SARS-CoV-2) emerged in Wuhan China, a disease that the WHO1 declared a pandemic in 2020 and in which immunisation has been shown to be an effective strategy to reduce the spread, severity of symptoms and number of deaths.2 , 3 According to the National Institute of Health (INS) of Colombia,4 by May 2022, more than 6 million confirmed cases were reported, of which approximately 139,000 died. As a consequence of the severity of the disease, a vaccination programme for the coronavirus was initiated worldwide, which seeks to reduce related deaths, hospitalisations for severe cases, post-covid sequelae and transmission of the disease. In Colombia, a national vaccination programme against COVID-19 was initiated for all inhabitants over 16 years of age, with a gradual strategy framed in two phases of progressive prioritisation.5 The vaccines mentioned have demonstrated WHO-approved efficacy in their initial trials, acceptable to the scientific community, but with many doubts and concerns for communities, who, in some cases, enter into a phenomenon known as vaccine hesitancy, which refers to "delay in acceptance or refusal of vaccination despite the availability of vaccination services". Vaccine hesitancy is complex and context-specific, and varies according to time, place and vaccines.6 This phenomenon may be influenced by a number of variables, including health and immunisation literacy, confidence in vaccine efficacy and safety, reliability and competence of services, among others6. According to WHO,7 vaccine hesitancy is among the top 10 causes threatening global health security and according to Nossier8 is one of the greatest threats to vaccination programmes against COVID-19. A study conducted by the National Administrative Department of Statistics (DANE) in 2020, applied to 24 Colombian cities, inquired about vaccination intention for COVID-19 and found that the intention of the population of Popayán was 61.1%,9 however, the reasons for vaccination intention in the city of Popayán were unknown. The above results suggest that, even with the availability of COVID-19 vaccines, a part of the population is expected not to be vaccinated. Unwillingness to receive the vaccine remains a major challenge to achieving the required vaccination coverage for population immunity, making it necessary to understand the factors associated with intention or refusal to vaccinate against COVID-19 in the city of Popayán in 2022. Methodology A descriptive-analytical cross-sectional study was conducted, with non-probabilistic convenience sampling. The sample size was calculated with the Epidat 3.1 programme, with a population of 210,134 inhabitants between 18 and 59 years of age, for the city of Popayán (DANE, 2019), with a significance level of 99%, expected proportion of 50% and estimation error of 5%, resulting in a sample size of 894 people. A loss rate of 10% (99 persons) was assumed in the event of any inconvenience or difficulty in the data entry, resulting in a final sample size of 993 persons. Fieldwork was conducted from 1 August 2021 to 1 March 2022. Information was collected through the construction of a survey consisting of 3 dimensions: socio-demographic characteristics, intention to vaccinate, and knowledge and perceptions about vaccination. The survey was applied in person at the points of greatest affluence in the city of Popayán (shopping centres, market galleries, sports venues and the city centre) and virtually through a Google form shared in specific groups of social networks in the city. The inclusion criteria to be part of the study were to be a resident of the city of Popayán and to be between 18 and 59 years of age; people who had already been vaccinated against COVID-19 at the time of the study and those who did not complete the survey in full were excluded from the study. The statistical package Stata 15 was used for data management. A descriptive analysis of the data was carried out according to the type of variable. For quantitative variables, measures of central tendency and measures of dispersion were used, taking into account their distribution. Categorical variables, both nominal and ordinal, were described through absolute and relative frequencies. For bivariate analysis, continuous variables were compared with respect to the dependent variable (intention to vaccinate) using the Student's t-test if the distribution was normal or the Wilcoxon rank sum test if the distribution was non-normal. For categorical variables, the chi-square test or Fisher's exact test was applied, as appropriate. Subsequently, separate binary logistic regressions were estimated, examining different independent variables (socio-demographic data, knowledge and confidence about vaccination for COVID-19) with the dependent variable (intention to vaccinate). Multiple logistic regression was then applied. Independent variables with p ≤ .25 were included in the initial model and continued with the forward selection option (conditional - likelihood ratio statistic), at a significance level of .05. Finally, the Hosmer-Lameshow test was applied to demonstrate the goodness-of-fit of the model obtained. In terms of ethical considerations, the study was categorised as minimal risk research. The principles of bioethics were applied as established by the Declaration of Helsinki, resolution no. 8430 of 1993 of the Colombian Ministry of Health, law 1581 of 2012 and its regulatory decree 1377 of 2013. A document was shared with the participants with the description and purpose of the research, the procedure to be used, the use that would be made of the information, the risks and benefits, the voluntary participation and the mechanisms with which the confidentiality of the information was guaranteed. The study was approved by the research ethics committee of a higher education institution. Results The number of individuals who responded to the questionnaire was 1,067, of which 993 individuals met all the inclusion criteria. 49.24% (n = 489) of the study population were between 18 and 28 years old, 27.09% (n = 269) between 29 and 39 years old, 14.10% (n = 140) between 40 and 50 years old and 9.57% (n = 95) between 51 and 59 years old (socio-demographic characteristics are described in table 1 ).Table 1 Socio-demographic characteristics Table 1Variable Frequency % Variable Frequency % Gender Civil status Female 558 56.19 Married-partnership 306 30.82 Male 435 43.81 Single-widowed 687 69.18 Stratum Income 1 334 33.64 Higher 1lmlw 691 69.59 2 374 37.66 Lower 1lmlw 302 30.41 3 208 20.95 4 72 7.25 Education 5 5 .50 None 14 1.41 Primary 87 8.76 Secondary 265 26.69 Health system Tec-technological 256 25.78 Contributively 388 39.07 Graduate 371 37.36 Subsidiary 549 55.29 Special 27 2.72 Residence No affiliation 29 2.92 Rural 226 22.76 Urban 767 77.24 In Popayán, the study population that does not intend to be vaccinated against COVID-19 is 23.16% (n = 230). A total of 24.97% (n = 248) have tested positive for COVID-19. When asked about the reasons for their intention not to vaccinate, they reported not being well informed about vaccination (56.29%), being afraid of needles (24.17%), not having received adequate medical information (45.42%), not considering the vaccine effective (54.88%), considering the probability of contracting COVID-19 without vaccinating to be low (26.49%) or considering the COVID-19 vaccine as harmful to their health (28.20%) (see Table 2, Table 3 ).Table 2 Intention to vaccinate Table 2Variable Frequency % Variable Frequency % Are you thinking of vaccinating? Do you believe the vaccine is effective? No 230 23.16 No 545 54.88 Yes 763 76.84 Yes 448 45.12 Are you well informed? Do you have a medical recommendation to not vaccinate? No 559 56.29 No 960 96.68 Yes 434 43.71 Yes 33 3.32 Have you tested positive in a COVID test? If you do not vaccinate, what is the probability of acquiring COVID ? No 745 75.03 High 730 73.51 Yes 248 24.97 Low 263 26.49 Are you afraid of needles? Severity of contracting COVID No 753 75.83 High 763 76.84 Yes 240 24.17 Low 230 23.16 Have you received medical information? Is vaccinating harmful to your health? No 451 45.42 No 713 71.80 Yes 542 54.58 Yes 280 28.20 Table 3 Awareness and perceptions about vaccination Table 3Variable Frequency % Variable Frequency % Immune system weakness It may cause cancer No 719 72.41 No 917 92.35 Yes 274 27.59 Yes 76 7.65 You can get COVID from the vaccine Control devices may be implanted No 719 72.41 No 886 89.22 Yes 274 27.59 Yes 107 10.78 May change DNA Recommend to a relative to apply the vaccine No 845 85.10 No 256 25.78 Yes 148 14.90 Yes 737 74.22 May lead to allergies Confidence in the health personnel No 537 54.08 High 571 57.50 Yes 456 45.92 Low 422 42.50 May cause death Confidence in the national vaccination plan No 828 83.38 High 367 36.96 Yes 165 16.62 Low 626 63.04 May cause other illness Confidence in the pharmaceutical companies No 672 67.67 High 575 57.91 Yes 321 32.33 Low 418 42.09 The results of simple logistic regressions showed that considering the COVID-19 vaccine as harmful to health, considering the vaccine as a weakening of the immune system, considering it as an immunobiological capable of causing death and considering COVID-19 as a disease of low severity to health are risk factors for not intending to be vaccinated against COVID-19. Similarly, low trust in health personnel, in the national vaccination plan and in pharmaceutical companies were associated with non-vaccination intention. In addition, the odds of recommending to the family not to receive the COVID-19 vaccine in people who do not intend to vaccinate are 43.7 times that of people who do intend to vaccinate (OR: 43.74 CI: 28.92-66.14 P < .05). Furthermore, the intention not to vaccinate against COVID-19 decreases with increasing stratum (intention not to vaccinate versus stratum 1 is 49% lower in stratum 2 (OR:0.51 CI: 0.36-0.72 P < 0.05), 58% lower in stratum 3 (OR:0.42 CI: 0.27-065 P < 0.05) and 75% lower in stratum 4 (OR:0.25 CI: 0.11-0.54 P < 0.05). Similarly, in people who have tested positive for PCR, the chance of not intending to be vaccinated is reduced by 36% (OR:0.06 CI:0.44-0.93 P < .05), in contrast to people who have not tested positive for COVID-19 (table 4 ).Table 4 Bivariate logistic regression Table 4Variable OR Confidence interval p Stratum 1 Reference .00 2 .51 .36–.72 .00 3 .42 .27–.65 .00 4 .25 .11–.54 5 1 Sex Female Reference Male 1.07 .80–1.44 .62 Are you well informed about vaccines in general? Yes Reference No 1.46 1.08–1.95 .01 Are you well informed about this vaccine? Yes Reference No 2.50 1.81–3.46 .00 Fear of needles? No Reference Yes 1.14 .81–1.60 .43 Have you tested positive for COVID? No Reference Yes .64 .44–.93 .00 Would you recommend a relative receives the vaccine? Yes Reference No 43.74 28.92–66.14 .00 Is the vaccine harmful for one’s health? No Reference Yes 28.90 19.56–42.70 .00 Is the vaccine serious for health? High Reference Low 13.18 9.29–18.68 .00 Probability of contracting COVID-19 after applying the vaccine? High Reference Low 14,19 10,02–20,09 0,00 Can the vaccine lead to death? No Reference Yes 16.05 10.84–23.76 .00 Can the vaccine change your DNA? No Reference Yes 11.45 7.74–16.95 .00 Do you think that after the vaccine a control device may be implanted? No Reference Yes 11.05 7.05–17.32 .00 Can the vaccine lead to allergies? No Reference Yes 2,00 1.48–2.70 .00 Can the vaccine weaken the immune system? No Yes 9.12 6.55–12.70 .00 Is the vaccine effective? Yes Reference No 12.64 7.89–20.25 .00 Do you trust the health personnel? High Reference Low 8.26 5.80–11.75 .00 Do you trust the national vaccination plan ? High Reference Low 7.38 4.69–11.63 .00 Do you trust the pharmaceutical companies? High Reference Low 15.94 10.58–24.01 .00 The results of the multivariate logistic regression showed that after controlling for sociodemographic variables, beliefs about the severity of the disease, the consequences of vaccination such as contracting the disease, death and weakening the immune system are risk factors for not intending to vaccinate. Similarly, not considering the vaccine effective, having low trust in health workers and considering the likelihood of contracting COVID-19 to be low were described as additional risk factors (table 5 ).Table 5 Multivariate logistic regression Table 5Vaccination intention Odds Ratio (95% Conf. Interval) p Would you recommend it to a family member? 7.179153 4.066311 12.67494 .000 Is the vaccine damaging to health? 5.94329 3.216886 10.9804 .000 Is contracting COVID-19 a serious health threat? 5.482954 2.964259 10.14175 .000 What is the probability of contracting COVID-19? 3.389196 1.872784 6.133463 .000 Can the vaccine cause death? 2.434627 1.276974 4.641762 .007 Do you trust the health personnel? 1.753201 0.9742801 3.154857 .061 Stratum .6480079 .4796884 .8753896 .005 Can it weaken the immune system? 2.299955 1.267476 4.173484 .006 Is the vaccine effective? 2.236907 1.147233 4.361587 .018 1. Chi-square Hosmer and Lemeshow = 5.32, p = .722. Discussion According to the present study, there is a high prevalence of people who do not intend to be vaccinated (23.1%), this situation was similar to the study by Ruiz,10 where it is evident that 14.8% of respondents have a limited intention to be vaccinated and 23% say they feel insecure; in the same way, Viswanath et al. 11 found that 50.3% of respondents stated that they were unlikely to be vaccinated, showing resistance to the vaccination process that is being implemented worldwide. According to the records obtained, the female gender had greater participation in the surveys carried out in the city of Popayán, as well as in the studies carried out in France,12 the United Kingdom13 and the United States,10 where more than 50% of women showed greater participation in issues related to vaccination against COVID-19, finding that men were less likely to participate, as they are less motivated by the vaccine.12 Likewise, Detoc et al. 12 found in their research that those who are most interested in participating in this type of study have a technical-technological or university level of education, which is similar to what was found in the present study. Similarly, the results from Popayán showed that socioeconomic factors, level of education and high income are associated with a greater intention to be vaccinated, similar to the study by Nikolovskiid et al.14 in the United States, where it was found that poor populations and those with a low level of education were less willing to be vaccinated. According to DANE figures, Popayán has 328,129 inhabitants. According to official records, by May 2022, 93 per cent of the population received the first dose and 81.4 per cent received the second dose, leaving approximately 61,193 people in the city who have not yet been fully vaccinated.15 Despite the great insistence by the National Government to promote vaccination and the great media coverage of COVID-19, there is still a significant number of inhabitants who have not been vaccinated. Furthermore, the Ministry of Health and Social Protection, by means of resolution no. 350 of 2022,16 established new guidelines regarding the use of face masks, making the use of this personal protection element in open spaces not required in regions where at least 70% of the population had a complete vaccination schedule, meaning the first and second doses, without boosters. According to INS figures,4 Popayán met the percentage for the implementation of the measure, however, the department of Cauca did not reach the required threshold. By May 2022, only 41.7% of the total population of the department has the complete scheme; a situation that generates uncertainty about the emergence of a new epidemiological peak due to the constant displacement of inhabitants from rural areas to the departmental capital, increasing the risk of exposure of the unvaccinated and those who do not intend to be vaccinated and, consequently, the increase in morbidity and mortality due to COVID-19. According to the study by Ruiz and Bel,10 there are at least 4 reasons for vaccine hesitancy: concerns about the side effects of the vaccine, concerns about allergic responses to the vaccine, doubts about the efficacy of the vaccine, and concerns about developing immunity through infection. In the present study, 54% of the population do not consider the vaccine to be effective and a high percentage have concerns about side effects, including DNA alteration, generation of other diseases and even death. Similarly, it should be noted that one of the reasons that least supports non-vaccination is fear of needles, with a representation of only 11% in the study by Ruiz and Bell10 and 24.1% in the present study. Regarding knowledge and perceptions about vaccination against COVID-19, 27.59% of the study population believe that they can contract the disease through vaccination, a figure that coincides with the study by Sherman et al.,13 where the population believes that a vaccine against the coronavirus could transmit the virus. Only 32.33% of the study population believe that the vaccine may cause other diseases, which is lower, compared to the perception found in the aforementioned study, where the concern about experiencing side effects from a coronavirus vaccine is higher. Regarding the perception of the surveyed population of the seriousness of contracting COVID-19, in this research it was quite high at 76.84%, which is in line with the data from the article by Sherman et al,.13 where 73.4% believe that the coronavirus represents a significant to important risk to people in the UK. The intention not to be vaccinated against COVID-19 decreases with increasing stratum; living in rural areas where the socio-economic strata are low increases the fear of adverse effects that could be caused by the vaccine.17 As these areas are vulnerable and difficult to access due to their geographic barriers, their population does not have useful coverage that provides adequate and timely information on promotion and preven tion services, within which immunisation strategies against the virus are outline. On the other hand, the present study shows an association between people who do not intend to be vaccinated and those who do not recommend vaccination to their friends and family members. According to the study by Urrunaga-Pastor et al.,17 the recommendation of friends and family has a positive influence on the intention to vaccinate, but is associated with a higher prevalence of adverse effects. This situation suggests that excessive exposure to false and fatalistic news may generate resistance to vaccination and lead to vaccination refusal. The results of the present study indicate that 63.04% of the surveyed population has low confidence in the implementation of the national vaccination plan, contrary to what was found in the study by Sherman et al.,13 where the population indicated confidence in the National Health Service to manage the coronavirus pandemic in the UK. According to Urrunaga-Pastor et al.,17 the low trust in government entities could be due to the fact that the measures adopted have been affected by political decisions or groups, which has led to clandestine vaccination outside the context of a clinical trial and influence peddling. Conclusion The intention to vaccinate against COVID-19 is determined not only by the technical-administrative dynamics of the immunisation programme and the health system, but also by the variables of the context and the perception of risk, which add up to explain the vaccination processes. It should be noted that this research has the limitations inherent to cross-sectional studies, including the impossibility of establishing causal relationships, among other aspects, such as the lack of approximation of the population's perception of the appearance of new strains of the virus, since it is a phenomenon that generates changes in its infectious capacity, clinical evolution and prognosis, which becomes a factor that reduces the effectiveness of current vaccines and can consequently reduce the population's confidence both in the application of the first dose and in completing vaccination schedules. It is suggested that health authorities refer to variables such as those expressed in this study, in order to strengthen decisions in the planning, implementation and evaluation of health policy related to immunisation against COVID-19. Funding This research was funded by the University of Cauca. Authorship Authors Omar Andrés Ramos Valencia, Yuliana Buitrón González and Jorge Sotelo Daza participated in the conceptualisation, research, methodological design, data curation, data analysis and writing of the manuscript. Authors Omar Andrés Ramos Valencia, Yuliana Buitrón González and Andrés Felipe Villaquiran participated in the revision and editing of the final manuscript. Acknowledgements Our thanks to the people who participated and made possible the development of this study. ==== Refs Referencias 1. OMS. Nuevo coronavirus 2019 [Internet]. Covid-19. 2019. Disponible en: https://www.who.int/es/emergencies/diseases/novel-coronavirus-2019 2. Scobie HM, Johnson AG, Suthar AB, Severson R, Alden NB, Balter S. Monitoring Incidence of COVID-19 Cases , Hospitalizations , and Deaths , by. Cdc Mmwr. 2021;70(37):1284–90. 3. Gupta S. Cantor J. Simon K.I. Bento A.I. Wing C. Whaley C.M. Vaccinations Against COVID-19 May Have Averted Up To 140,000 Deaths In The United States. 40 9 2021 1465 1472 10.1377/hlthaff202100619 Aug 18 4. Instituto Nacional de Salud. COVID-19 en Colombia [Internet]. Panorama. 2022. Disponible en: https://www.ins.gov.co/Noticias/paginas/coronavirus.aspx 5. Ministerio de Salud y Protección Social de Colombia. Plan Nacional de Vacunación contra el COVID-19 [Internet] Bogotá 2021 Disponible en: https://www.minsalud.gov.co/sites/rid/Lists/BibliotecaDigital/RIDE/VS/pnv-contra-covid-19.pdf 6. MacDonald N, SAGE Working Group on Vaccine Hesitancy Vaccine hesitancy: Definition, scope and determinants Vaccine. 33 34 2015 4161 4164 25896383 7. OMS, OPS. Principales amenazas a la salud en 2019 [Internet]. 2019. Disponible en: https://www3.paho.org/hq/index.php?option=com_content&view=article&id=14916:ten-threats-to-global-health-in-2019&Itemid=135&lang=es 8. Nossier S.A. Vaccine hesitancy: the greatest threat to COVID-19 vaccination programs J Egypt Public Health Assoc. 96 1 2021 18 20 34224031 9. MinSalud. Encuesta Pulso Social. Resultados – sexta ronda (Diciembre de 2020) [Internet]. 2021 Disponible en: https://ninezya.org/wp-content/uploads/2021/03/DANE-Pulso-social-diciembre-2020-extendida.pdf 10. Ruiz J.B. Bell R.A. Predictors of intention to vaccinate against COVID-19: Results of a nationwide survey Vaccine [Internet]. 39 7 2021 1080 1086 Disponible en: 10.1016/j.vaccine.2021.01.010 33461833 11. Viswanath K. Bekalu M. Dhawan D. Pinnamaneni R. Lang J. McLoud R. Individual and social determinants of COVID-19 vaccine uptake BMC Public Health. 21 1 2021 1 10 33388037 12. Detoc M. Bruel S. Frappe P. Tardy B. Botelho-Nevers E. Gagneux-Brunon A. Intention to participate in a COVID-19 vaccine clinical trial and to get vaccinated against COVID-19 in France during the pandemic Vaccine. 38 45 2020 7002 7006 32988688 13. Sherman S.M. Smith L.E. Sim J. Amlôt R. Cutts M. Dasch H. COVID-19 vaccination intention in the UK: results from the COVID-19 vaccination acceptability study (CoVAccS), a nationally representative cross-sectional survey Hum Vaccines Immunother. 17 6 2021 1612 1621 14. Nikolovskiid J. Koldijk M. Weverling G.J. Spertus J. Turakhia M. Saxon L. Factors indicating intention to vaccinate with a COVID-19 vaccine among older U.S. adults PLoS One [Internet]. 16 5 2021 1 14 Disponible en: 10.1371/journal.pone.0251963 15. Ministerio de Salud y Protección Social de Colombia. Municipios con cobertura de vacunación COVID-19 [Internet]. 2022. Disponible en: https://www.minsalud.gov.co/sites/rid/Lists/BibliotecaDigital/RIDE/VS/PP/referencia-municipios-vacunacion-70porciento-msps.zip 16. Ministerio de Salud y Protección Social de Colombia. Resolución No. 350 de 2022 [Internet]. Colombia; 2022 p. 11. Disponible en: https://www.minsalud.gov.co/sites/rid/Lists/BibliotecaDigital/RIDE/DE/DIJ/resolucion-350-de-2022.pdf 17. Urrunaga-Pastor D. Bendezu-Quispe G. Herrera-Añazco P. Uyen-Cateriano A. Toro-Huamanchumo C.J. Rodriguez-Morales A.J. Cross-sectional analysis of COVID-19 vaccine intention, perceptions and hesitancy across Latin America and the Caribbean Travel Med Infect Dis. 2021 41(April)
PMC010xxxxxx/PMC10288314.txt	==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 Published by Elsevier Ltd. S0264-410X(23)00755-7 10.1016/j.vaccine.2023.06.066 Article Modifiers of COVID-19 Vaccine Efficacy: Results from Four COVID-19 Prevention Network Efficacy Trials Turley Christine B. a⁎ Tables LaKesha b⁎ Fuller Trevon c⁎ Sanders Lisa J. d Scott Hyman e Moodley Amaran f Woodward Davis Amanda S. g Leav Brett h Miller Jacqueline h Schoemaker Kathryn i Vandebosch An j Sadoff Jerald j Woo Wayne k Cho Iksung k Dunkle Lisa M. k Li Sijia l van der Laan Lars l Gilbert Peter B. g Follmann Dean m Jaynes Holly g Kublin James G. g Baden Lindsey R. n Goepfert Paul o Kotloff Karen p Gay Cynthia L. q Falsey Ann R. r El Sahly Hana M. s Sobieszczyk Magdalena E. t Huang Yunda g Neuzil Kathleen M. p Corey Lawrence g Grinsztejn Beatriz u Gray Glenda v Rouphael Nadine w1 Luedtke Alex l1⁎ on behalf of the COVID-19 Prevention Network CoVPN 2 a Atrium Health Wake Forest School of Medicine, Charlotte, NC, United States b Morehouse School of Medicine, Atlanta, GA, United States c Infectious Diseases Department, Hospital Federal dos Servidores do Estado, Rio de Janeiro, RJ Brazil d University of South Florida Morsani College of Medicine, Tampa, FL, United States e San Francisco Department of Public Health, San Francisco, CA, United States f Division of Infectious Diseases, University of California San Diego and Rady Children’s Hospital, San Diego, CA, United States g Vaccine and Infectious Disease Division, Fred Hutchinson Cancer Center, Seattle, WA, United States h Moderna Inc., Cambridge, MA, United States i Biometrics, Vaccines & Immune Therapies, BioPharmaceuticals R&D, AstraZeneca, Gaithersburg, MD, United States j Janssen Vaccines and Prevention, Leiden, the Netherlands k Novavax, Gaithersburg, MD, United States l Department of Biostatistics, University of Washington, Seattle, WA, United States m Biostatistics Research Branch, National Institute of Allergy and Infectious Disease, National Institutes of Health, Bethesda, MD, United States n Brigham and Women's Hospital, Boston, MA, United States o Division of Infectious Diseases, Department of Medicine, University of Alabama at Birmingham, Birmingham, AL, United States p Division of Infectious Disease and Tropical Pediatrics, Department of Pediatrics, and the Center for Vaccine Development and Global Health, University of Maryland School of Medicine, Baltimore, MD, United States q Department of Medicine, Division of Infectious Diseases, UNC HIV Cure Center, University of North Carolina at Chapel Hill School of Medicine, Chapel Hill, NC, United States r Department of Medicine, Infectious Disease Division, University of Rochester, Rochester, NY, United States s Department of Molecular Virology and Microbiology and Section of Infectious Diseases, Department of Medicine, Baylor College of Medicine, Houston, TX, United States t Division of Infectious Diseases, Department of Medicine, Columbia University Irving Medical Center, New York, NY, United States u Evandro Chagas National Institute of Infectious Diseases-Fundação Oswaldo Cruz, Rio de Janeiro, Brazil v Perinatal HIV Research Unit, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa, South African Medical Research Council, Cape Town, South Africa w Hope Clinic, Emory University, Atlanta, GA, United States ⁎ Corresponding authors. 1 Members in each individual study/protocol team are listed in the supplementary materials. 2 Contributed equally 23 6 2023 23 6 2023 16 4 2023 20 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Questions remain regarding the effect of baseline host and exposure factors on vaccine efficacy (VE) across pathogens and vaccine platforms. We report placebo-controlled data from four Phase 3 COVID-19 trials during the early period of the pandemic. Cross-protocol analysis of four randomized, placebo-controlled efficacy trials (Moderna/mRNA1273, AstraZeneca/AZD1222, Janssen/Ad26.COV2.S, and Novavax/NVX-CoV2373) using a harmonized design. Trials were conducted in the United States and international sites in adults ≥ 18 years of age. VE was assessed for symptomatic and severe COVID-19. We analyzed 114,480 participants from both placebo and vaccine arms, enrolled July 2020 to February 2021, with follow up through July 2021. VE against symptomatic COVID-19 showed little heterogeneity across baseline socio-demographic, clinical or exposure characteristics, in either univariate or multivariate analysis, regardless of vaccine platform. Similarly, VE against severe COVID-19 in the single trial (Janssen) with sufficient endpoints for analysis showed little evidence of heterogeneity. COVID-19 VE is not influenced by baseline host or exposure characteristics across efficacy trials of different vaccine platforms and countries when well matched to circulating virus strains. This supports use of these vaccines, regardless of platform type, as effective tools in the near term for reducing symptomatic and severe COVID-19, particularly for older individuals and those with common co-morbidities during major variant shifts. Clinical trial registration numbers: NCT04470427, NCT04516746, NCT04505722, and NCT04611802. Keywords Comorbidity Effect Modifier, Epidemiologic Environmental exposure Occupational exposure SARS-CoV-2 Vaccine Efficacy ==== Body pmc1 Introduction Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in December 2019 causing a global pandemic that has resulted in nearly 762 million cases and 6·9 million deaths worldwide as of April 2023. Early in the pandemic, the National Institute of Allergy and Infectious Disease (NIAID) at the US National Institutes of Health (NIH) partnered with multiple pharmaceutical companies to expedite the development of an effective vaccine to prevent the spread of SARS-CoV-2. In addition to this unprecedented public-private collaboration, NIAID merged four existing clinical trial networks to form the COVID-19 Prevention Network (CoVPN). The CoVPN, along with the study sites affiliated with contract research organizations (CRO), facilitated the rapid enrollment of the tens of thousands of participants needed for the Phase 3 vaccine safety and efficacy trials [1]. Four SARS-CoV-2 vaccine candidates (mRNA-1273, AZD1222 ChAdOx1 nCoV-19, Ad26.COV2.S, and NVX-CoV2373) have been evaluated through double-blind, randomized, placebo-controlled Phase 3 clinical trials, and were reported to be safe and efficacious in adults ≥18 years of age, with early estimates of overall vaccine efficacy (VE) ranging from 56·3% to 94·1% [2], [3], [4], [5]. The data from these pivotal Phase 3 VE studies present a unique opportunity to comprehensively evaluate modifiers of COVID-19 VE in preventing symptomatic illness, as well as severe disease, through cross-protocol analysis. VE is influenced by factors related to the infectious agent (antigenic variants), the vaccine platform, the host (age, sex, genetics, presence of comorbid conditions, and immune function), and the environment (exposure and transmission rate) [6], [7], [8], [9]. For example, VE is lower among men and the elderly for influenza vaccines[7], among the immune suppressed for the hepatitis B vaccine[10], and among children living in low resource settings for rotavirus vaccines [11]. In the individual CoVPN trials, no significant differences in VE based on selected host factors were reported in the subgroup analyses [2], [3], [4], [5]. However, these univariate analyses varied between trials and no multivariate analysis was performed. Here we present a cross-protocol analysis designed with consistent methodology to assess the impact of these modifiers on VE. 2 Materials and methods 2.1 Study design We performed a participant-level data cross-protocol analysis of the Moderna, AstraZeneca, Janssen, and Novavax trials using [2], [3], [4], [5] data accrued through the blinded, pre-crossover phases from July 2020 through July 2021 (Appendix p. 17, Supplementary Table 1 ). National populations were excluded if they were enrolled during the circulation of the beta variant or Latin American sites that were not of ancestral/alpha or other lineage due to mismatch with vaccine target (Appendix p. 15, Supplementary Figure 1 ).Table 1 Symptomatic Covid-19 endpointsa by subgroup and randomization arm in the per-protocol cohort of each trial excluding South African participants and intersex participants (# endpoints / total # participants). Moderna AstraZeneca Janssen Novavax Placebo Vaccine Placebo Vaccine Placebo Vaccine Placebo Vaccine Overall 741/14164 55/14287 181/8528 134/17617 538/17113 173/17111 77/8385 17/17272 Socio-demographic Age (years) 18-29 88/1391 5/1417 29/1072 33/2209 96/1762 32/1774 26/1259 4/2735 30-39 138/2143 11/2169 31/1365 22/2756 71/1970 32/1992 14/1649 5/3338 40-49 180/2665 12/2597 42/1526 41/3281 112/3507 46/3526 14/1676 2/3601 50-59 170/2878 12/2969 39/1762 27/3658 136/3740 25/3680 14/1923 4/3757 60-69 109/3240 10/3339 33/1875 9/3818 90/4437 32/4360 7/1433 1/2946 ≥70 56/1847 5/1796 7/928 2/1895 33/1697 6/1779 2/445 1/895 Ethnicity Hispanic/Latino 176/2787 10/2831 59/2064 59/4032 250/8889 93/8767 18/1801 9/3707 Not Hispanic/ Latino 565/11377 45/11456 122/6464 75/13585 288/8224 80/8344 59/6584 8/13565 Race American Indian/Alaska Nativeb 5/113 0/109 20/372 26/747 65/1616 27/1641 6/522 1/1068 Asian 29/700 1/628 4/355 4/738 15/639 5/699 5/375 0/757 Black or African American 41/1352 4/1395 15/699 6/1401 30/1451 10/1416 8/947 1/1881 Other 30/422 1/464 3/164 2/338 24/621 6/589 1/54 0/146 Multiple 8/304 1/300 10/203 7/421 36/951 9/928 0/137 2/296 White 628/11273 48/11391 129/6735 89/13972 368/11835 116/11838 57/6350 13/13124 Sex Female 365/6670 25/6848 66/3714 56/7732 239/7629 73/7617 50/4158 10/8283 Male 376/7494 30/7439 115/4814 78/9885 299/9484 100/9494 27/4227 7/8989 Health Characteristics Placebo Vaccine Placebo Vaccine Placebo Vaccine Placebo Vaccine Body mass index (kg/m2) Healthy weight or Underweight (< 25) 150/3861 6/3970 41/2482 32/5275 163/5379 58/5501 29/2463 4/5220 Overweight (≥ 25, < 30) 267/4938 20/4857 62/3079 44/6406 245/7055 73/6901 14/2719 7/5610 Obese (≥ 30) 324/5365 29/5460 78/2967 58/5936 130/4679 42/4709 34/3203 6/6442 Class 3 Obese (≥ 40) 74/995 7/1015 12/473 10/976 17/601 7/643 7/572 3/1267 Cardiovascular Disease 196/4472 19/4468 49/2392 20/5057 116/4550 24/4422 18/2021 4/4059 Diabetes 76/1468 5/1484 19/873 10/1627 42/1587 17/1608 7/858 2/1622 HIV 5/87 0/93 6/134 1/278 7/274 3/240 1/49 0/135 Kidney Disease 4/74 0/73 2/45 0/131 3/113 0/106 1/56 0/125 Liver disease 5/101 1/113 9/160 1/308 6/168 2/169 0/62 1/134 Chronic Lung Disease 37/808 5/808 13/1055 14/2027 44/1150 16/1114 9/1264 1/2461 Risk Characteristics Placebo Vaccine Placebo Vaccine Placebo Vaccine Placebo Vaccine Workplace Risk of Exposurec Low 49/2666 36/5204 497/16355 159/16390 39/4951 7/9956 Medium 368/6735 30/6760 93/3765 61/7734 14/229 3/249 27/2670 6/5607 High 373/7429 25/7527 39/2097 37/4679 27/529 11/472 11/764 4/1709 Risk from Living Conditiond Low 113/2354 8/2322 78/4365 47/9044 232/8676 66/8739 57/6808 13/14002 Medium 564/10229 40/10348 33/1593 25/3227 189/4978 57/4923 16/1117 3/2274 High 40/1179 5/1165 29/1336 25/2744 97/2694 38/2683 2/339 0/733 Very High 24/402 2/452 41/1234 37/2602 20/765 12/766 2/121 1/263 Tobacco use 12/253 2/236 31/1705 22/3471 8/369 5/374 24/2609 6/5330 Geographic Location USA 741/14164 55/14287 145/7423 89/15389 331/9121 95/9156 73/7887 16/16261 Argentina 44/1414 20/1400 Brazil 43/3385 12/3394 Chile 10/670 9/1358 6/539 3/528 Colombia 86/1862 34/1856 Mexico 5/218 2/207 4/498 1/1011 Peru 26/435 36/870 23/574 7/570 Excluded variants were: Beta, Delta, Epsilon, Eta, Gamma, Iota, Lambda, Mu, Kappa, and Zeta. a All participants were right censored at time t0 regardless of whether they were observed to experience event after t0. Separate time points t0 were chosen for each trial such that about 10% participants are at risk in the vaccine arm. b Category is defined across all clinical sites. Indigenous people from South America were classified together with the American Indian or Alaska Native United States and Mexico demographic according to the FDA definition (American Indian or Alaska Native: A person having origins in any of the original peoples of North and South America (including Central America), and who maintains tribal affiliation or community attachment). In this analysis, the Moderna, AstraZeneca, Janssen and Novavax trials included 222, 1119, 3257, and 1590 participants, respectively, who identified as American Indian or Alaskan Native from North America. c Detailed derivation of exposure risk based on OSHA categories is provided in Supplemental Methods d Living condition encompasses housing type and household size, detailed derivation provided in Supplemental Methods. Figure 1 Estimates of vaccine efficacy against symptomatic Covid-19 within subgroups defined by categorical baseline covariates, with corresponding 95% confidence intervals. In the AstraZeneca an Novavax studies, the number of events among Black/African-American participants was too low to estimate VE. 2.2 Vaccines and adjuvants The Moderna vaccine was stored between -50°C an -15°C and was administered as two doses 28 days apart. The antigen administered was 100 microgram (μg) mRNA-1273/0.5 mL. The AstraZeneca vaccine was stored between 2°C and 8°C and was also administered as two doses 28 days apart. The antigen was AZD1222 (5×1010 viral particles)/0.5 mL. The Janssen vaccine was stored between 2°C and 8°C and consisted of a single dose of 5×1010 viral particle/0.5 mL. The NovaVax vaccine was also stored between 2°C and 8°C, was administered as two doses 21 days apart, and the antigen consisted of NVX-CoV2373 (5 μg of SARS-CoV-2 recombinant spike protein adjuvanted with 50 μg of Matrix-M)/0.5 mL. The route of administration was intramuscular in the deltoid muscle of the arm for all of the vaccines. Needle length varied based on the population served with 25-38 mm used for most adults. 2.3 Study outcomes Symptomatic COVID-19 for this analysis was defined as signs or symptoms consistent with COVID-19 and molecularly confirmed by PCR testing, which was harmonized across the studies with minor differences [1]. These were the primary endpoints from each trial except for ENSEMBLE, for which this corresponds to a secondary endpoint of mild, moderate, or severe COVID-19 (a companion paper lists the COVID-19 case definition used in each trial[1]). Severe COVID-19 was defined as additionally having shortness of breath at rest or respiratory distress, respiratory rate ≥30 per minute, heart rate ≥125 per minute, or oxygen saturation ≤93% on room air, organ failure, ICU admission, or death. 2.4 Predictors of study outcomes Potential predictors of the study outcomes included demographic characteristics (age, sex, race, ethnicity, and country); comorbid conditions including asthma, cardiovascular disease, hypertension, diabetes, smoking, obesity (BMI ≥30 kg/m2), lung disease, liver disease, kidney disease, and well-controlled HIV; SARS-CoV-2 exposure risk (Occupational Safety and Health Administration [OSHA] risk category); and living situation risk score (Appendix p. 14, Supplementary Methods). 2.5 Ethics approval Institutional review board approval was obtained for the four COVID-19 vaccine efficacy trials [2], [3], [4], [5] (Moderna/mRNA1273: NCT04470427, AstraZeneca/AZD1222: NCT04516746, Janssen/Ad26.COV2.S: NCT04505722, and Novavax/NVX-CoV2373: NCT04611802). Informed consent was obtained after the nature and possible consequences of the study had been fully explained to the subjects. 2.6 Statistical analysis The objectives of this study are to: 1) determine which baseline characteristics modify COVID-19 VE in each trial; 2) determine if combined baseline characteristics in each trial modify VE; 3) if there is evidence of heterogeneity of efficacy seen within a study based on baseline characteristics, rank the importance of baseline characteristics compared to the combined impact on VE; and 4) evaluate whether VE-modifying characteristics in three of the trials yield improved prediction of VE in the fourth trial. Analyses were prespecified in a statistical analysis plan (Appendix p. 2-13). Cumulative incidence VE was estimated for each vaccine through a fixed time point after enrollment in the per-protocol cohort. To ensure stable estimation, for each trial we selected the latest time point where the risk set consisted of at least 10% of participants in both arms (Appendix p. 18-19, Supplementary Tables 2-3 for a summary of the amount of follow-up in each trial). Inverse weighting was used to provide an interpretable and unbiased analysis of how VE varies within covariate subgroups. Weights were fitted via a proportional odds model, logistic regression, and stratified Kaplan-Meier estimator, respectively (details on Appendix p. 14, Supplementary Methods). Because each trial assessed a separate vaccine platform that may have its own heterogeneity profile, VE was assessed separately for each trial. Univariate analyses of VE were conducted via a nonparametric covariate-adjusted method. Multivariate analyses were conducted via ensemble methods that use cross-validation to combine predictions from a collection of candidate algorithms. To assess heterogeneity, participants within each trial were broken into tertile groups based on their covariate-stratified VE estimates: those predicted to have the lowest VE, those predicted to have the highest, and everyone else. VE within each of these three subgroups was assessed using the observed COVID-19 endpoints. We assessed the potential benefits of pooling data from the four trials using a leave-one-trial-out procedure. We pooled data from the other three vaccines to estimate their pooled efficacy conditional on baseline covariates. Then, we included this estimated efficacy as an additional putative efficacy modifier in a repetition of the multivariate analysis in the remaining trial. If the four vaccines have similar heterogeneity profiles, pooling the data should increase precision for assessing heterogeneity. Missing data in the covariates were minimal and were imputed by the median and mode for continuous and categorical variables, respectively. For a given trial, we report both uncorrected and Bonferroni-corrected 95% confidence intervals for univariate analyses of VE in supplementary tables, and only uncorrected intervals in figures. Analyses were performed in R version 4.2.1. 3 Results 3.1 Study population In total, 136,096 participants met the inclusion criteria and were randomized within the four trials and, of these, 114,477 were in the per-protocol cohort and not enrolled in South Africa (due to dominant circulating beta variant) or intersex/unknown sex, and therefore part of our analysis cohort (Appendix p. 16, Supplementary Figure 2 ). Across the four trials 77,747 (68%) participants were 18 to 59 years of age and 36,730 (32%) were 60 or older (Table 1). Women represented 52,651 (46%) of the participants. In terms of ethnicity, 34,878 (30%) of the participants were Hispanic. The racial composition of participants across the four trials was 6,188 (5%) American Indian/Alaska Native, 4,891 (4%) Asian, 10,542 (9%) Black or African American, and 86,518 (76%) White. Regarding clinical comorbidities, 42,790 (37%) of the participants had at least one of the following: diabetes, HIV, or cardiovascular, kidney, liver, or chronic lung disease; 38,761 (34%) of the participants were obese. Tobacco use was reported by 14,347(13%) of participants.Figure 2 Estimated vaccine efficacy by age and body mass index (BMI) across the four trials, with corresponding 95% confidence intervals. Estimates and intervals are derived according to a working model that enforced that the relative risk of a Covid-19 endpoint on vaccine versus on placebo must be log-linear. 3.2 Baseline covariates have little impact on VE against symptomatic COVID-19 in univariate and multivariate analyses In the analysis using harmonized case definitions, overall VE for preventing symptomatic COVID-19 was 93% (90-95%) in Moderna, 65% (54-76%) in AstraZeneca, 71% (64-77%) in Janssen, and 91% (87-96%) in Novavax (Appendix p. 20-21, Supplementary Table 4). While VE showed little heterogeneity for participants enrolled at sites within the US, VE ranged from 44-82% for participants in the Janssen trial across different countries (Appendix p. 20-21, Supplementary Table 4). No difference was observed in VE between older adults and younger populations or based on BMI, race, sex, underlying health conditions, and risk of exposure (Figure 1, Table 2). When participants with the lowest predicted VE were compared to participants with the highest predicted VE for each trial, there was little variability in estimates of VE across the subgroups defined by baseline covariates (Figure 3 ). When tertiles of VE were pooled from three trials, and then used to predict VE of the fourth, and no heterogeneity was found in the subgroups formed by baseline covariates (Appendix p. 22, Supplementary Table 5).Figure 3 Estimated tertiles of vaccine efficacy (VE) against Covid-19 endpoint as defined by all baseline covariates, with corresponding 95% confidence intervals (CIs). For each vaccine, vaccine efficacy was first predicted using all baseline covariates, and then participants were broken into three subgroups as defined by this prediction: the 33·33% with lowest predicted VE (Lowest VE), the 33·33% with the next lowest predicted VE (Middle VE), and the 33·33% with the highest predicted VE (Highest VE). Vaccine efficacy was then estimated in these subgroups using a cross-validation method. If needed, estimates were projected to satisfy the population-level constraint that efficacy should be nondecreasing when moving from the first to the third tertile of participants. 3.3 VE against severe COVID-19 remained high across all demographic subsets VE for severe COVID-19 was restricted to the Janssen trial due to very few endpoints in the vaccine arm of the other three trials (Appendix p. 23, Supplementary Table 6). VE without modeling the effect of baseline covariates was 85% (78-91%) and did not vary substantially by demographics, obesity, living conditions, or geographic region (Appendix p. 26, Supplementary Table 7). In multivariate analysis, there was also no evidence of heterogeneity across tertiles of VE (Appendix p. 27, Supplementary Table 8). 4 Discussion Understanding the factors that impact vaccine efficacy is crucial to ensure that vaccines are effective in a wide variety of hosts and settings. The ability to evaluate host and environmental factors that may modify VE is frequently limited by trial design, which often excludes vulnerable populations or may be too small to allow for subgroup analysis. Further, the lack of harmonization across different efficacy studies often prevents larger scale, cross-protocol analysis. The COVID-19 pandemic and subsequent coordinated worldwide engagement in the rapid development of vaccines created an opportunity to evaluate host characteristics and environmental factors in a novel cross-protocol analysis of pivotal studies conducted early in the pandemic. In these trials, populations known to be at high risk for acquisition of the disease or for developing severe disease were enrolled. Thorough baseline information was recorded for all participants, allowing a more nuanced view of VE in different subgroups during an extraordinarily well-characterized period of global risk. This multi-national harmonized collaboration offers an unprecedented chance to compare COVID-19 VE in a large, diverse adult sample using a consistent definition of disease, exposure risk, and statistical methodology across vaccine platforms. The COVID-19 pandemic highlighted the significant impact host and demographic factors can have on health outcomes. The dramatic disparities in COVID-19-related morbidity and mortality seen in elderly patients, populations of color, and those with various comorbidities further underscores the need to understand the potential impact of these factors on responses to available COVID-19 vaccines [12], [13], [14], [15], [16]. Our study revealed that these individual characteristics that raised the risk of COVID-morbidity and mortality, did not impact VE against symptomatic and severe COVID-19. Our findings confirm those of real-world studies [17], which have shown similar VE across populations and age groups. The aging process is accompanied by senescence of the immune system often characterized by dysregulated inflammatory responses and impairments in the processes of B and T cell differentiation [18]. The immune response to COVID-19 is a salient example. Early in the pandemic, age ≥65 years was recognized as a risk factor for severe health outcomes associated with COVID-19 [19]. Mechanisms hypothesized to explain differences in the clinical course of COVID-19 by age include lower levels of interferon gene expression and T cell diversity in older adults [20]. It therefore became important to evaluate VE in the elderly and to prioritize vaccinating them once an approved vaccine became available. Older adults are still the age group with the highest number of incident COVID-19 cases, and the highest rates of hospitalization and death and clinical treatment guidelines characterize age as the most important risk factor for progression to severe COVID-19. In this analysis of randomized controlled trials conducted early in the pandemic, we did not observe a difference in VE between older adults and younger populations. Similarly, observational studies conducted early in the pandemic, reported that the Moderna and Pfizer/BNT162b2 (Pfizer/BNT) vaccines were found to be highly effective (94%) in preventing hospitalization in fully vaccinated individuals ≥65 years of age [21] and a single dose of the Janssen vaccine was >84% effective at preventing severe disease in older adults [22]. Thus, these highly effective vaccines using the same dose as younger persons resulted in similar short-term protection. Whether response to boosting, alteration in the circulating virus transmission kinetics, or immune escape characteristics alter our observations is unknown and worthy of continued monitoring and evaluation. In addition, observational studies of the real-world effectiveness of the primary series of the Pfizer/BNT vaccine reported no significant differences in vaccine effectiveness based on individual characteristics like elevated body mass index, hypertension, or type 2 diabetes; however, the presence of three or more comorbidities was associated with slightly lower vaccine effectiveness [16]. A significant concern early in the COVID-19 pandemic was the risk to individuals being regularly exposed to the virus. Our findings showing that risk of exposure had little impact on VE in the short term is supported by multiple reviews of workers in high-risk occupations who were among the first to receive COVID-19 vaccinations. Several observational studies have explored the relationship between occupation and COVID-19 VE in health care personnel [23], [24], [25], [26], first responders, and other essential frontline workers [5], [27]. Prior to the emergence of variants, vaccine effectiveness against COVID-19 in fully vaccinated health care workers ranged from 85% for Pfizer/BNT in Israel to 100% for Moderna in France [23], [24]. All existing data confirm that personal or environmental variables do little to blunt the short-term effectiveness of a variety of formulations of COVID-19 vaccines in comparison to the VE observed in the trials, and that effectiveness is comparable over a range of vaccine recipients [1]. The number of severe endpoints across both arms was sufficient to evaluate modifiers of VE only for the ENSEMBLE trial. In this trial, which also was the only trial that evaluated a single dose vaccine regimen, VE against severe disease was 74·6% ≥28 days after vaccination [3], and did not vary substantially by age or comorbidities [28]. A recent review and network meta-analysis comparing the VE of COVID-19 vaccines found no statistical difference in risk of severe disease among the eight COVID-19 vaccine Phase 3 randomized controlled trials, suggesting that there is limited heterogeneity across vaccine platforms [29]. Overall, vaccines against COVID-19 are incredibly successful at preventing the most severe outcomes of disease, which is encouraging as new platforms and updated vaccines are rolled out as the virus continues to mutate at a rapid pace. Interrogation of VE within subgroups is a significant challenge for individual studies let alone across large, multinational trials. Our analysis offers several improvements over previous studies of the heterogeneity of COVID-19 VE [2], [3], [4], [5]. First, the trials analyzed here had a harmonized design utilizing very similar endpoints and sample sizes, and all were placebo-controlled. We also used a consistent method to analyze randomized trial datasets across the different vaccine regimens, ensuring that the endpoint and covariate definitions were harmonized. Additionally, our analysis made use of all available blinded, placebo-controlled follow-up data. To provide unbiased and generalizable findings, we used inverse probability weighting in our analysis. Furthermore, we conducted a multivariate analysis using a state-of-the-art machine learning method to evaluate the extent to which multiple baseline covariates can predict VE. Our study had several limitations. First, while we found that VE after the primary vaccination series was not influenced by baseline host or risk characteristics in the four Phase 3 vaccine efficacy trials, we have limited insight into whether the variables examined have a long-term effect on VE due to the relatively short follow up in the blinded/pre-crossover periods of these trials. Due to the high efficacy of vaccines, the period for accumulating cases for determination of VE was brief, limited to the interval in each study prior to crossover to active vaccine and termination of the placebo-controlled design and/or emergency use authorization and subsequent vaccine availability across populations. Second, the vaccines were each designed with the same target on SARS-CoV-2, the Spike protein, and had a good match with circulating strains at the time. This protein, while important for establishing an immune response in each of these vaccine platforms, has also been subject to rapid, persistent mutation, and VE of these vaccines against less-matched variants may reduce as virus variants emerge. Therefore, our findings may not apply to future data where longer follow up is observed in the context of emerging variants. Third, the Janssen vaccine trial data referenced here was a single dose series, while the other three studies used two-dose regimens, and none included a booster dose, which is the current standard of care, in the blinded/pre-crossover phase of the trials. Fourth, the trials enrolled in a variety of locations worldwide; while this enhanced the racial and ethnic diversity of the population within each trial, the spread of viral variants was occurring at different times in different locales, likely affecting enrollment of trial participants, and variability in VE estimates. Fifth, while we were able to analyze most factors commonly cited as affecting VE, because of small participant numbers we could not assess the effect of immunocompromising conditions like HIV or organ transplantation with suboptimal responses to COVID-19 vaccines. Finally, as the four trials had different start and end dates, differences in timing could have introduced uncontrolled confounders. We attempted to address this by controlling for the effect of variants in the statistical analysis. Our study supports that VE was high early in the pandemic independent of vaccine platform, host factors, and settings. The recruitment of racially and ethnically diverse participants, and the availability of placebo-controlled data enabled us to conclude that the vaccines were effective for all, including those most at risk for COVID-19. The responses to COVID-19 vaccines currently vary based on interval since last exposure to the virus, the immunologic background of the host, and the continued emergence of variants of concern such as omicron sublineages predominating in China, Europe, and the United States as the pandemic enters its third year. This analysis provides strong support for development and use of vaccines that are well-matched to circulating virus strains regardless of vaccine platform type, as effective tools in the near term for reducing symptomatic and severe COVID-19 infections, particularly for high-risk individuals during major variant shifts in the circulating SARS-CoV-2 strain [30]. This has broad implications for future COVID-19 vaccination strategy development and provides strong support that vaccination remains the best tool to prevent the most severe and debilitating forms of COVID-19. Data sharing Access to data underlying findings described in this manuscript may be allowed in accordance with the individual data sharing policies of the pharmaceutical companies contributing data to this analysis. As each of the clinical trials included in this meta-analysis are ongoing, data availability will begin after publication of the final study results in 2023 and 2024. Funding This work was supported by the Biomedical Advanced Research and Development Authority (BARDA) and National Institute of Allergy and Infectious Diseases (NIAID) (grants UM1 AI068614 and UM1 AI068635). The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. CRediT authorship contribution statement Christine B. Turley: Conceptualization, Investigation, Methodology, Project administration, Supervision, Validation, Writing – original draft. LaKesha Tables: Conceptualization, Methodology, Project administration, Supervision, Writing – original draft. Trevon Fuller: Investigation, Methodology, Project administration, Supervision, Validation, Visualization. Lisa J. Sanders: Conceptualization, Methodology, Writing – original draft. Hyman Scott: Conceptualization. Amaran Moodley: Conceptualization, Methodology, Writing – original draft. Amanda S. Woodward Davis: Methodology, Project administration. Brett Leav: . Jacqueline Miller: . Kathryn Schoemaker: . An Vandebosch: . Jerald Sadoff: . Wayne Woo: . Iksung Cho: . Lisa M. Dunkle: . Sijia Li: Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization. Lars van der Laan: Methodology, Software. Peter B. Gilbert: Conceptualization, Funding acquisition, Methodology. Dean Follmann: . Holly Jaynes: . James G. Kublin: . Lindsey R. Baden: Investigation. Paul Goepfert: Investigation. Karen Kotloff: Investigation. Cynthia L. Gay: Investigation. Ann R. Falsey: Investigation. Hana M. El Sahly: Investigation. Magdalena E. Sobieszczyk: Investigation. Yunda Huang: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation. Kathleen M. Neuzil: . Lawrence Corey: Funding acquisition. Beatriz Grinsztejn: Conceptualization, Investigation, Project administration. Glenda Gray: Conceptualization, Investigation, Project administration. Nadine Rouphael: Conceptualization, Project administration, Supervision. Alex Luedtke: Conceptualization, Formal analysis, Methodology, Project administration, Supervision, Visualization, Writing – original draft. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available on request. Acknowledgments The COVID-19 Prevention Network (CoVPN) was formed through a partnership of multiple existing NIAID-funded clinical trial networks, including the HIV Vaccine Trials Network (HVTN), Vaccine Treatment and Evaluation Units (VTEUs), the HIV Prevention Trials Network (HPTN), the AIDS Clinical Trials Group (ACTG), and the Infectious Diseases Clinical Research Consortium (IDCRC). We would also like to acknowledge and thank the trial participants, caregivers, investigators, health care providers, and research staff who contributed to the four trials. All authors attest they meet the ICMJE criteria for authorship. ==== Refs References 1 Mena Lora AJ, Long JE, Huang Y, Baden LR, El Sahly HM, Follmann D, et al. Rapid development of an integrated network infrastructure to conduct Phase 3 COVID-19 vaccine trials. JAMA Network Open. 2023;6:e2251974-e 2 Baden L.R. El Sahly H.M. Essink B. Kotloff K. Frey S. Novak R. Efficacy and safety of the mRNA-1273 SARS-CoV-2 vaccine New England Journal of Medicine. 384 2020 403 416 33378609 3 Sadoff J. Gray G. Vandebosch A. Cárdenas V. Shukarev G. Grinsztejn B. Safety and efficacy of single-dose Ad26.COV2.S vaccine against Covid-19 New England Journal of Medicine. 384 2021 2187 2201 33882225 4 Falsey A.R. Sobieszczyk M.E. Hirsch I. Sproule S. Robb M.L. Corey L. Phase 3 safety and efficacy of AZD1222 (ChAdOx1 nCoV-19) Covid-19 vaccine New England Journal of Medicine. 385 2021 2348 2360 34587382 5 Dunkle L.M. Kotloff K.L. Gay C.L. Áñez G. Adelglass J.M. Barrat Hernández A.Q. Efficacy and safety of NVX-CoV2373 in adults in the United States and Mexico New England Journal of Medicine. 386 2021 531 543 34910859 6 Falahi S. Kenarkoohi A. Host factors and vaccine efficacy: Implications for COVID-19 vaccines Journal of Medical virology. 94 2022 1330 1335 34845730 7 Dhakal S. Klein S.L. Host Factors Impact Vaccine Efficacy: Implications for Seasonal and Universal Influenza Vaccine Programs Journal of Virology. 93 2019 e00797 e00819 31391269 8 Kaslow D.C. Force of infection: a determinant of vaccine efficacy? npj Vaccines. 2021 9 Langwig K.E. Gomes M.G.M. Clark M.D. Kwitny M. Yamada S. Wargo A.R. Limited available evidence supports theoretical predictions of reduced vaccine efficacy at higher exposure dose Scientific Reports. 9 2019 3203 30824732 10 Lee G.-H. Lim S.-G. CpG-adjuvanted Hepatitis B vaccine (HEPLISAV-B®) update Expert Review of Vaccines. 20 2021 487 495 33783302 11 Carvalho M.F. Gill D. Rotavirus vaccine efficacy: current status and areas for improvement Human Vaccines & Immunotherapeutics. 15 2019 1237 1250 30215578 12 Magesh S, John D, Li WT, Li Y, Mattingly-app A, Jain S, et al. Disparities in COVID-19 outcomes by race, ethnicity, and socioeconomic status: a systematic review and meta-analysis. JAMA Network Open. 2021;4:e2134147-e 13 Butt A.A. Talisa V.B. Yan P. Shaikh O.S. Omer S.B. Mayr F.B. Real-world effectiveness of the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) mRNA vaccines in preventing confirmed infection in patients on chronic hemodialysis Clinical Infectious Diseases. 2022 14 Chodick G. Tene L. Rotem R.S. Patalon T. Gazit S. Ben-Tov A. The Effectiveness of the Two-Dose BNT162b2 Vaccine: Analysis of Real-World Data Clinical Infectious Diseases. 74 2022 472 478 33999127 15 Thomas S.J. Perez J.L. Lockhart S.P. Hariharan S. Kitchin N. Bailey R. Efficacy and safety of the BNT162b2 mRNA COVID-19 vaccine in participants with a history of cancer: subgroup analysis of a global phase 3 randomized clinical trial Vaccine. 40 2022 1483 1492 35131133 16 Dagan N. Barda N. Kepten E. Miron O. Perchik S. Katz M.A. BNT162b2 mRNA Covid-19 vaccine in a nationwide mass vaccination setting New England Journal of Medicine. 384 2021 1412 1423 33626250 17 Zheng C. Shao W. Chen X. Zhang B. Wang G. Zhang W. Real-world effectiveness of COVID-19 vaccines: a literature review and meta-analysis International Journal of Infectious Diseases. 114 2022 252 260 34800687 18 Pinti M. Appay V. Campisi J. Frasca D. Fülöp T. Sauce D. Aging of the immune system: Focus on inflammation and vaccination European Journal of Immunology. 46 2016 2286 2301 27595500 19 Chen Y. Klein S.L. Garibaldi B.T. Li H. Wu C. Osevala N.M. Aging in COVID-19: vulnerability, immunity and intervention Ageing Research Reviews. 65 2021 101205 20 Yoshida M. Worlock K.B. Huang N. Lindeboom R.G.H. Butler C.R. Kumasaka N. Local and systemic responses to SARS-CoV-2 infection in children and adults Nature. 602 2022 321 327 34937051 21 Tenforde M.W. Olson S.M. Self W.H. Talbot H.K. Lindsell C.J. Steingrub J.S. Effectiveness of Pfizer-BioNTech and Moderna vaccines against COVID-19 among hospitalized adults aged ≥65 Years - United States, January-March 2021 Morbidity and Mortality Weekly Report. 70 2021 674 679 33956782 22 Moline H.L. Whitaker M. Deng L. Rhodes J.C. Milucky J. Pham H. Effectiveness of COVID-19 vaccines in preventing hospitalization among adults aged ≥65 Years - COVID-NET, 13 states, February-April 2021 Morbidity and Mortality Weekly Report. 70 2021 1088 1093 34383730 23 Angel Y. Spitzer A. Henig O. Saiag E. Sprecher E. Padova H. Association between vaccination with BNT162b2 and incidence of symptomatic and asymptomatic SARS-CoV-2 infections among health care workers JAMA. 325 2021 2457 2465 33956048 24 Paris C, Perrin S, Hamonic S, Bourget B, Roué C, Brassard O, et al. Effectiveness of mRNA-BNT162b2, mRNA-1273, and ChAdOx1 nCoV-19 vaccines against COVID-19 in healthcare workers: an observational study using surveillance data. Clinical Microbiology and Infection. 2021;27:1699.e5-.e8 25 Pilishvili T, Gierke R, Fleming-Dutra KE, Farrar JL, Mohr NM, Talan DA, et al. Effectiveness of mRNA Covid-19 vaccine among U.S. health care personnel. The New England Journal of Medicine. 2021;385:e90 26 Hallal P.C. Hartwig F.P. Horta B.L. Silveira M.F. Struchiner C.J. Vidaletti L.P. SARS-CoV-2 antibody prevalence in Brazil: results from two successive nationwide serological household surveys The Lancet Global Health. 8 2020 e1390 e1398 32979314 27 Ho R.J.Y. Warp-Speed Covid-19 vaccine development: beneficiaries of maturation in biopharmaceutical technologies and public-private partnerships Journal of Pharmaceutical Sciences. 110 2021 615 618 33212162 28 Hardt K. Vandebosch A. Sadoff J. Le Gars M. Truyers C. Lowson D. Efficacy, safety, and immunogenicity of a booster regimen of Ad26.COV2.S vaccine against COVID-19 (ENSEMBLE2): results of a randomised, double-blind, placebo-controlled, phase 3 trial The Lancet Infectious Diseases. 22 2022 1703 1715 36113538 29 Rotshild V. Hirsh-Raccah B. Miskin I. Muszkat M. Matok I. Comparing the clinical efficacy of COVID-19 vaccines: a systematic review and network meta-analysis Sci Rep. 11 2021 22777 34815503 30 Acevedo M.L. Gaete-Argel A. Alonso-Palomares L. de Oca M.M. Bustamante A. Gaggero A. Differential neutralizing antibody responses elicited by CoronaVac and BNT162b2 against SARS-CoV-2 Lambda in Chile Nature Microbiology. 7 2022 524 529
PMC010xxxxxx/PMC10288315.txt	==== Front Neurol Clin Neurol Clin Neurologic Clinics 0733-8619 1557-9875 Elsevier Inc. S0733-8619(23)00058-0 10.1016/j.ncl.2023.06.006 Article Era of COVID-19 in Multiple Sclerosis Care Krett Jonathan D. MD a Salter Amber PhD b Newsome Scott D. DO a∗ a Division of Neuroimmunology and Neurological Infections, Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA b Section on Statistical Planning & Analysis, Department of Neurology, University of Texas Southwestern Medical Center, Dallas, Texas, USA ∗ Corresponding author: Division of Neuroimmunology and Neurological Infections, Johns Hopkins Hospital , 600 N Wolfe St, Pathology 627, Baltimore, Maryland 21287, USA. 23 6 2023 23 6 2023 © 2023 Elsevier Inc. All rights reserved. 2023 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Synopsis The unprecedented scope of the coronavirus disease 2019 (COVID-19) pandemic resulted in numerous disruptions to daily life, including for people with multiple sclerosis (PwMS). In this article, we review how disruptions in MS care prompted innovations in delivery of care (e.g., via telemedicine) and mobilized the global MS community to rapidly adopt safe and effective practices. We discuss how our understanding of the risks of COVID-19 in PwMS has evolved along with recommendations pertaining to disease-modifying therapies (DMTs) and vaccines. With lessons learned during the COVID-19 pandemic, we examine potential questions for future research in this new era of MS care. Key Words COVID-19 multiple sclerosis telemedicine registries disease-modifying therapy vaccination ==== Body pmcKey Points • People with MS (PwMS) experienced disruptions in their care and everyday lives during the COVID-19 pandemic. • Innovations such as telemedicine helped preserve access to clinicians, while its optimal application to future MS care remains a topic of debate. • Data from large MS registries proved to be informative regarding risks associated with COVID-19 and interactions with MS disease-modifying therapies. • Many of the risk factors for poor outcomes in COVID-19 for PwMS are similar to those in the general population (e.g., older age, black race); among PwMS, greater disability and B cell depleting therapies are associated with increased risk. • Vaccines against COVID-19 are safe and effective for PwMS, although humoral responses to vaccination are blunted by certain disease-modifying therapies. Clinics care points • Until more data are available, use of telemedicine for MS care should be based on the preferences of people with MS and providers along with local regulations. • Considering currently available safety data, disease-modifying therapies can be started and sequenced similarly post-pandemic compared to the pre-pandemic era, assuming risks and benefits are discussed in detail with each person with MS. • Anti-CD20 monoclonal antibody therapies remain first-line options for some, and people on these therapies should be counseled about increased infection risk along with the possibility of impaired vaccine responses. Extended interval dosing requires further investigation, should be considered in select cases, and has relevance beyond the scope of COVID-19. • COVID-19 vaccines are recommended for people with MS and do not appear to be associated with an increased risk of relapse. SDN: received consultant fees for scientific advisory boards from Biogen, Genentech, Bristol Myers Squibb, EMD Serono, Greenwich Biosciences, Horizon Therapeutics, Novartis, TG Therapeutics; study lead PI for a Roche clinical trial program; advisor to Autobahn; received research funding (paid directly to institution) from Biogen, Lundbeck, Roche, Genentech, National MS Society, The Stiff Person Syndrome Research Foundation, Department of Defense, and Patient Centered Outcomes Research Institute. Introduction As of April 2023, there have been over 760 million confirmed cases and 6.8 million deaths worldwide due to coronavirus disease 2019 (COVID-19).1 In the United States, there have been more than 100 million cases and 1 million deaths.2 The pandemic resulted in profound disruptions to society and healthcare systems globally. For people with multiple sclerosis (PwMS) and their clinicians, the COVID-19 pandemic presented significant challenges. It not only affected the psychosocial well-being of PwMS but also caused major interruptions in routine MS care.3 For example, missed clinical, laboratory, and imaging appointments related to the pandemic made it more difficult for clinicians to monitor disease activity and quality-of-life issues in PwMS.4 Uncertainty surrounding the safety of MS disease-modifying therapies (DMTs) due to their varied effects on the immune system was also a major concern. In this review, we will discuss the broad impact of the COVID-19 pandemic on MS care. We will highlight lessons learned by the MS community regarding delivery of care, COVID-19 risks, DMT selection, and strategies to optimize the efficacy of vaccinations against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). We will conclude by examining implications for future care as we transition from the COVID-19 global health emergency to a phase of endemic and seasonal infection.5 Disruptions along the continuum of MS care due to COVID-19 The COVID-19 pandemic caused significant disruption for PwMS and the public. Lockdowns and physical distancing measures which were implemented for public safety made it difficult to access routine care for chronic conditions like MS. Naturally, there was uncertainty about which activities outside the home could be done safely and, in some jurisdictions, PwMS may not have been permitted to leave home, with rare exceptions. Interruptions along the continuum of MS care were common during the pandemic. A cross-sectional survey of more than 1000 PwMS conducted in April 2020 found that 22% cancelled a visit with their neurologist, 11% cancelled an MRI, 21% cancelled a laboratory test, and 10% altered the administration schedule of their DMT.6 Another study of more than 4000 individuals with autoimmune disorders, of whom more than 800 were PwMS showed that nearly half experienced an interruption in healthcare services.7 Delays in infusions and lost rehabilitation visits were frequent sources of disruption.8 Surveys of MS care providers confirmed that postponements in usual care were common and that providers were concerned about the risk COVID-19 posed to PwMS and themselves.9 , 10 Along with consternation about being able to safely monitor PwMS, many providers expressed misgivings about the risk-benefit ratio of using higher efficacy DMTs in the context of the pandemic.9 , 10 Whether disruptions in MS care resulted in any long-lasting consequences at an individual level is currently unknown. Several studies also showed that changes to daily activities including work and socialization as a result of the pandemic were common among PwMS (i.e., remaining at home, using virtual methods), similar to the general population.8 , 11 Despite the heterogeneity of PwMS regarding health-related quality of life and disability, it was suggested that substantial psychosocial and occupational change might have a greater impact for PwMS, particularly those with pre-existing activity limitations.11 , 12 In one such study, women with MS were more likely than men with MS to experience job termination or furlough during the pandemic and expressed greater concern about the risk posed by COVID-19 to their health.6 Moreover, psychological distress amongst PwMS pertaining to COVID-19 risk adversely affected their well-being, particularly when few effective treatments and no vaccines were available during the early pandemic.13 Altogether, public health measures put in place to protect us from COVID-19 no doubt had unintended impacts on MS care and required compensatory strategies to counterbalance them (Figure 1 ). The next section will explore innovative care delivery methods that were accelerated during the pandemic to keep PwMS connected with their providers.Figure 1 Disruptions and compensatory strategies in the COVID-19 era of MS care. Examples provided depict how compensatory strategies and innovations by the MS community helped counterbalance pandemic-related disruptions to care. Bringing care to people with MS using telemedicine Telemedicine, or telehealth, leverages the ability for individuals and their providers to connect despite not being physically present in the same location. Telemedicine was already being used for chronic and acute medical care prior to the COVID-19 pandemic and can take a variety of forms. These include synchronous contact over an audiovisual platform using the internet, or asynchronous methods such as pre-recorded or other electronic communication.14 Studies prior to the COVID-19 pandemic demonstrated that assessment of PwMS, including disability measures, is feasible in a virtual format.15 , 16 In response to pandemic-related disruptions, many MS centers started conducting most visits using telemedicine.17 , 18 Several studies highlighted benefits of telemedicine for PwMS, including improved access for those who live far from an MS center and those with mobility issues.17 , 19 Some PwMS have greater comfort being in their home environment and having additional carers present virtually who may not otherwise be present at in-person visits.20 Additionally, those with higher disability may benefit from more frequent clinical touchpoints by supplementing in-person visits with telemedicine appointments. Surveys suggest that most neurologists and PwMS who used telemedicine during the COVID-19 pandemic were satisfied with its use, but we don’t yet know whether this translates to better MS disease outcomes.17 , 21 Of great concern is that subtle changes on neurological exam could be missed when using a virtual platform, resulting in inappropriately maintaining therapies that have suboptimal effectiveness. There are other concerns that telemedicine could make accessing care harder for certain PwMS. For example, persons of lower socioeconomic status (SES), health literacy, and skill with technology may have more difficulty using telemedicine.3 In at least one study, this concern did not bear out given that providers were able to conduct follow-up visits using mainly smartphones (and these are globally available at relatively low cost).15 , 20 It is notable that most surveys examining satisfaction with telemedicine sampled individuals who were of higher SES and may not be generalizable to marginalized populations.3 , 12 Arguably, qualitative evaluation of telemedicine during the exceptional circumstance of the COVID-19 pandemic may have been more focused on its feasibility rather than what might be optimal in the long-term care of PwMS. We can be reassured that most studies viewed the use of telemedicine positively and suggested that no major short-term complications arose as a result of its widespread use.17 , 20 , 22 , 23 It is worth carefully considering how best to integrate this knowledge into future clinical practice. For example, it is unclear whether exclusive use of telemedicine for both new and follow-up visits is ideal as opposed to the use of telemedicine only for follow-ups. One study from Norway suggested that some clinicians were dissatisfied completing new patient visits using telemedicine and looked upon telemedicine more favorably for follow-up visits.19 Furthermore, ensuring universal access remains a concern in jurisdictions with multiple healthcare payers such as the United States.14 Lingering concerns also exist pertaining to cybersecurity and privacy due to use of internet-based platforms.14 Overall, the rapid uptake of telemedicine during the COVID-19 pandemic may result in lasting changes in MS practice. The use of telemedicine is fluid and ever-changing based on regulations around its use and uncertainty on how best to apply it. Future research is needed to elucidate the full range of implications associated with short- and long-term use of telemedicine. The power of large registries for assessing COVID-19 risk in people with MS During the early part of the COVID-19 pandemic, MS researchers recognized the need for large studies to answer key questions. The Global Data Sharing Initiative developed a data sharing process to study the effects of COVID-19 in PwMS across small and large efforts globally.24 This provided harmonized data across multiple countries and helped determine potential risk factors using large registries. Registry data was ideal for this purpose because large populations could be studied while adjusting for confounders and examining for rare outcomes. Table 1 summarizes key contributions from several databases.Table 1 Selected registries and key contributions to COVID-19 risk assessment in MS Registry Country of origin Largest MS cohort Key contributions Multiple Sclerosis International Federation (MSIF) Global Data Sharing Initiative Global (HQ in UK) 5648 Harmonized data collection through multiple pooled registries24,26 North American Research Committee on Multiple Sclerosis (NARCOMS) United States 4955 Survey showed high rates (84.1%) of SARS-CoV-2 vaccination and determined reasons for vaccine hesitancy in PwMS96 COVID-19 Infections in MS & Related Diseases (COViMS) United States 1626 Risk factors for poor outcome with COVID-19 include increased disability, older age, hypertension, diabetes, obesity, black race, anti-CD20 DMT, and recent corticosteroid treatment25; pregnancy outcomes are no worse in PwMS35 Multiple Sclerosis and COVID-19 (MuSC-19) Italy 1362 Increased risk of poor COVID-19 outcome in PwMS with EDSS >3 or at least 1 comorbidity28 and use of anti-CD20 DMT32 Covisep France 347 Age, EDSS 6 or higher, and obesity were found to be independent risk factors for hospitalization with COVID-1927 German MS Register Germany Hundreds Contributed to the Global Data Sharing Initiative, and provided evidence that SARS-CoV-2 vaccination does not increase risk of relapse83 UK MS Register UK 404 Multiple studies, including those that described common symptoms of self-diagnosed COVID-19 in PwMS38 and determined that PwMS commonly experienced amplification of MS symptoms during COVID-19 infection; an effect that was attenuated by DMT46 Swedish MS Registry Sweden 17692 Corroborated increased risk of hospitalization with COVID-19 in PwMS on anti-CD20 DMT but suggested that this risk may be more modest than the risk associated with older age, increased disability, and medical comorbidities34 MSBase COVID-19 Substudy Australia Thousands Most notable publications relating to COVID-19 were under the auspices of the Global Data Sharing Initiative above Abbreviations: HQ – headquartered; PwMS – people with MS; DMT – disease-modifying therapy. A crucial question was whether PwMS possessed any unique risk factors for poor outcomes with COVID-19. Several registry-based MS studies found that many of the risk factors were similar to individuals in the general population, including older age and the presence of specific medical comorbidities (e.g., diabetes).25, 26, 27 Data from the North American COVID-19 Infections in MS and related diseases (COViMS) registry showed a 30% increase in the risk of hospitalization and intensive care unit (ICU) admission and/or need for artificial ventilation for every 10-year increase in age, with a 76.5% increased risk of death for every 10-year age epoch.25 Hypertension, diabetes, and morbid obesity also increased the risk of poor outcomes. Greater levels of ambulatory disability (e.g., Expanded Disability Status Scale (EDSS) >6 or requiring any assistance to walk) more than double the odds of more severe COVID-19 outcomes.25 , 27 , 28 Being non-ambulatory was associated with a 25-fold increased odds of death compared with fully ambulatory PwMS and black race was associated with >40% increased odds of being hospitalized (but not with an increased risk of death).25 Determining the risk of poor COVID-19 outcomes related to MS DMTs was also an intense focus of investigation. Smaller registries, such as those with hundreds rather than thousands of patients, initially found no effect of DMT exposure.27 , 29 , 30 Revisiting this using larger datasets revealed that anti-CD20 monoclonal antibody therapies were associated with increased risk.25 , 26 , 31 , 32 Rituximab was associated with 4.5-fold increased odds of hospitalization with COVID-19.25 Pooled international data confirmed that ocrelizumab and rituximab (compared to other DMTs) increased the odds of hospitalization and ICU admission but not death.31 This is consistent with data from other patients with immune-mediated disorders included in the Global Rheumatology Alliance registry, which found that rituximab was associated with 4-fold increased odds of death compared to methotrexate-treated individuals.33 DMT-treated status in isolation is not sufficient to risk stratify PwMS. Investigators in Italy found that COVID-19 risk was confined to PwMS in a ‘higher risk’ group, defined as those with EDSS >3 or with at least 1 comorbidity.28 Conversely, PwMS with EDSS 3 or less and no comorbidities had a risk of severe COVID-19 outcome similar to age- and sex-matched controls.28 Of note, untreated PwMS appear to have a higher risk of poor outcome31 that is variably present in different studies following adjustment for factors such as age and MS phenotype.27 , 32 This may reflect that this group is comprised of both individuals who are untreated due to having milder MS and those with more severe disability or progressive course who do not benefit from DMT (and may be at higher baseline risk of severe COVID-19 due to ambulatory status). A study of >17,000 PwMS from Sweden further supported the notion that pre-morbid disability and progressive MS course were likely more predictive of poor COVID-19 outcome compared to DMT type; an increased risk in rituximab-treated PwMS was again seen (albeit to a lesser degree than in smaller studies).34 Pregnant and post-partum PwMS and children with MS do not appear to be at higher risk of poor COVID-19 outcomes, however conclusions are limited by small sample size and under-representation of pregnant/young PwMS with high levels of ambulatory disability/comorbidities.35 , 36 In general, pregnant women who develop COVID-19 may have a higher risk of preterm birth,37 so individualized counseling remains important. Data from the United Kingdom (UK) MS Register suggested that the likelihood of developing COVID-19 is not influenced by DMT-treated status or premorbid disability; however, these conclusions are limited by patient self-reporting.38 COVID-19 risk also relates to treatments used for MS relapses. Glucocorticoid use in the 2 months preceding infection was associated with a doubling of the odds of hospitalization and quadrupling the risk of death among PwMS with COVID-19.25 Intravenous (IV) methylprednisolone use in the month preceding COVID-19 increases the risk, but it is unknown whether lower doses typically used as premedication interacts with the increased risk observed with anti-CD-20 monoclonal antibodies.32 The reasons for this observation are not well understood, given that dexamethasone is beneficial in severe acute COVID-19 respiratory infection.39 The timing, dose, and duration of corticosteroid administration relative to pathogen exposure may determine the net immunomodulatory and therefore clinical effects. In summary, pooled data from large registries has proven instrumental for informing the MS community about risk factors for poor outcomes secondary to COVID-19. Harmonized data using variables which were readily collected at the point of care enabled conclusions about clinically relevant risk factors for PwMS that appear consistent across studies. Possible limitations relate to voluntary reporting of data, use of variables only included a priori in the register, and lack of potentially relevant details such as DMT dose/frequency and MS disease activity. Symptomatic manifestations of COVID-19 in people with MS Studies based on patient self-report found that COVID-19 symptoms experienced by PwMS (e.g., ageusia, hyposmia, upper respiratory tract symptoms) are no different than individuals in the general population.40, 41, 42 In a study assessing post-acute symptoms in PwMS, nearly 30% of 8000 respondents reported COVID-19 symptoms lasting more than 1 month.43 The risk was higher in those with severe pre-existing neurological disability and mental health comorbidities.43 Many persistent manifestations such as lower respiratory tract symptoms (e.g., cough) and nondescript muscle aches were not consistent with MS, however new or worsened fatigue had a prevalence of nearly 70% among those with post-acute COVID-19 sequelae.43 Since symptoms such as fatigue and cognitive impairment are common to both MS and post-acute COVID-19, any pathogenetic interaction between the two disorders remains speculative currently.44 Evidence is sparse regarding whether COVID-19 produces durable changes in inflammatory disease activity in MS. It is common for PwMS to experience neurological symptoms during acute COVID-19, however in these studies, self-reported data limits the ability to conclusively distinguish pseudoexacerbations from new focal CNS inflammation due to MS.45, 46, 47 Impact of COVID-19 on disease-modifying therapy selection The expanding landscape of DMTs for PwMS has provided hope in terms of controlling the macroscopic neuroinflammatory component of MS (relapses and new lesions on magnetic resonance imaging). Since many DMTs modulate, suppress, or reconstitute components of the immune system, concerns about safety came up early in the COVID-19 pandemic. Practical recommendations were needed to balance infectious safety concerns with preserving efficacy for people with active MS. We will focus on key lessons about DMT selection in the context of COVID-19. For an in-depth review of SARS-CoV-2 pathogenesis and its relationship to DMT mechanisms, we direct the reader to other published literature.48, 49, 50, 51, 52 So far, no MS DMT has proven to be protective against COVID-19. There was initially optimism that beta interferons could counteract SARS-CoV-2 through antiviral effects and possibly by dampening pro-inflammatory host responses.53 Following one study which showed a non-statistically significant trend towards lower rates of hospitalization for PwMS on beta interferon,25 further studies found no significant beneficial effect.31 , 53 , 54 Hypotheses that other DMTs could attenuate COVID-19 severity through immunomodulatory mechanisms,55 particularly in the case of fingolimod and natalizumab, have not borne out in large datasets.26 , 31 , 48 , 56 , 57 During the pandemic, recommendations for DMT prescribing and risk assessments were largely based on expert consensus or experience with other infectious diseases.51 , 58 Changes in DMT use by PwMS were considered for some patients based on survey data and presumably related to concern about COVID-19.6 , 8, 9, 10 Although, many MS providers remained comfortable with new DMT starts during the pandemic if appropriate based on MS severity, and only a small minority (8%) recommended postponing all DMT administrations.59 This comfort level may have been enhanced by the larger COVID-19 MS registries highlighting that the majority of DMTs did not appear to increase the risk of contracting SARS-CoV-2 or experience more severe COVID-19 outcomes.26 , 31 , 32 Some PwMS may have had new MS disease activity and/or progression due to a change or discontinuation in therapy, however, evidence is lacking in this regard. Prescribing patterns for DMTs were altered during the COVID-19 pandemic. A study from the UK showed a steady trend to increasing monoclonal antibody DMT prescriptions from 2016–2019 that was reversed in 2020 at the start of the pandemic (except for natalizumab) with a 16.7% reduction in new starts.60 A similar trend was observed in Spain with decreases in prescriptions for anti-CD20 therapies and an increase in new natalizumab prescriptions, ostensibly due to lesser peripheral immunosuppression with natalizumab.61 A study of 670 PwMS prescribed DMT in the United States showed a 10% reduction in intravenous infusion DMT prescriptions with an increase in oral DMT prescriptions (+7%) which persisted from the pre-vaccine to the post-vaccine period of the pandemic.62 Delays in infused DMT (mostly B cell therapies) were more common than switches in DMT class or type. Prescribing patterns of self-injected therapies remained stable, although the study overlapped with a time where ofatumumab (a self-injected anti-CD20 DMT) was becoming more widely prescribed.62 Although the global health emergency has been declared over, endemic COVID-19 risk will continue to factor into therapeutic shared decision-making along with the perceived benefits of treatment. This will require similar ‘risk calculus’ as was used by experienced MS clinicians in the past. Prior to the COVID-19 pandemic, a qualitative study reinforced the idea that providers should engage PwMS in personalized discussions about risk tolerance when prescribing a DMT.63 In this study, many PwMS would accept a risk of non-life-threatening infection in order to better control their MS and preserve function.63 We direct the reader to other excellent resources for a more detailed discussion about starting or sequencing DMTs in the era of COVID-19.52 , 58 , 64, 65, 66 Some general recommendations from the National MS Society apply including: 1) PwMS currently on a DMT should not stop the treatment unless instructed to do so; 2) PwMS with COVID-19 symptoms or with a positive COVID-19 test should speak with their provider(s) (primary care and neurology clinicians); and 3) individualized decisions should be made regarding initiating or switching DMTs.64 Practical recommendations for use of approved DMTs, including when to consider interrupting treatment can be found in Table 2 .Table 2 Summary of recommendations for commercially available MS disease-modifying therapies in the COVID-19 era DMT Class Brief mechanism of action Mode of administration Effect on COVID-19 outcome Recommendation in SARS-CoV-2 infected PwMS Beta interferon Inhibits pro-inflammatory cytokines SQ or IM No increased risk. Protective effect has not been proven53 Continue treatment Glatiramer acetate Modulates T cell cytokine profile to increased Th2; promotes Treg cells SQ No increased risk26 Continue treatment Teriflunomide Inhibits reactive lymphocyte proliferation Oral No increased risk26 Continue treatment Fumarates Enhances Nrf2 pathway, improves oxidative stress response, and limits survival/activation of T cell subsets Oral No increased risk, except rarely if lymphopenic (ALC <800 cells/mm3)104 Continue; consider suspending if severe infection or ALC <800 cells/mm3 S1P receptor modulators Inhibits lymphocyte trafficking out of peripheral lymph nodes Oral No evidence of direct increased risk26 but may impair response to vaccines92 Continue; consider suspending if ALC <200 cells/mm3 Natalizumab Prevents leukocyte trafficking across the blood-CNS barrier by targeting alpha-4 integrins IV No evidence of increased risk.26 Extended interval dosing may reduce risk of hospital exposure71 Continue; may consider delaying infusion if critically ill Anti-CD20 monoclonal antibodies Lyses and depletes B lymphocytes by targeting CD20 on their surface IV (e.g., ocrelizumab) or SQ (ofatumumab) Increased risk of hospitalization estimated in the range of 2–5-fold.25,26,67 May impair response to vaccines86,94 Suspend/delay dosing Cladribine Inhibits DNA synthesis and depletes B > T lymphocytes Oral – two cycles separated by 1 year Likely only at increased risk when severely lymphopenic26,51 Suspend/delay dosing Alemtuzumab Lyses and depletes mature T and B lymphocytes and several innate immune cells by targeting CD52 on their surface IV– two cycles separated by 1 year Likely only at increased risk when severely lymphopenic/leukopenic26,51 Suspend/delay dosing Abbreviations: DMT – disease-modifying therapy; PwMS – people with MS; Th2 – T helper-2 cytokine profile; Treg – T regulatory cells; S1P – sphingosine-1 phosphate; Nrf2 – nuclear factor erythroid 2-related factor 2; IM – intramuscular; SQ – subcutaneous; IV – intravenous; ALC – absolute lymphocyte count. Most of the COVID-19-specific concern surrounds cell-depleting DMTs such as anti-CD20 monoclonal antibodies. Theoretical concern surrounds the induction phase of alemtuzumab and cladribine treatment due to their respective mechanisms, but they were not shown to be associated with increased risk in studies which included patients in the post-induction phase.27 , 31 , 32 , 67 Interferons, glatiramer acetate, fumarates, teriflunomide, sphingospine-1-phosphate (S1P) receptor modulators and natalizumab were not shown to be associated with an increased risk of COVID-19 severity.27 , 31 , 32 , 67 Lymphopenia with absolute lymphocyte counts (ALC) <800 cells/mm3 can occur rarely with fumarates and with cladribine or alemtuzumab treatment and should be monitored since severe lymphopenia may increase risk.48 , 50 , 51 , 68 Lymphocytes are peripherally sequestered by S1P modulators (e.g., fingolimod), so low lymphocyte counts likely represent a functional lymphopenia and may not increase risk unless the measured ALC is severely reduced (<200 cells/mm3).69 , 70 Even though the anti-CD20 infusions are dosed intermittently, this results in maintenance of a B cell-depleted state and may result in hypogammaglobulinemia.50 Immune reconstitution therapies (alemtuzumab and cladribine) possibly carry a greater peak risk during induction but may be appealing options for some PwMS who prefer their risk to be frontloaded.31 , 48 Although large observational studies did not show statistically significant increases in COVID-19 risk with immune reconstitution therapies, this does not necessarily indicate superior safety, especially since the number of patients on these therapies was small in comparison to the other DMTs.26 Autologous hematopoietic stem cell transplant remains an option for treatment refractory PwMS.64 COVID-19 risk is likely elevated for several months following the immunoablative phase of treatment but this has not been well studied.64 Despite the increased risk of COVID-19 severity with some of the MS DMTs, available data along with expert opinion suggests that some PwMS should continue to use these therapies in the proper clinical context.64 PwMS should also adhere to routine precautions to reduce risk of infection and exposure to COVID-19.64 These therapies remain an effective option for reducing disease activity and undertreating MS may be a greater risk than COVID-19. During the pandemic, some MS providers preferentially used natalizumab in John Cunningham virus (JCV) seronegative patients who had active disease given lower COVID-19 risks and less impact on vaccination response.30 , 69 The exposure risk of frequent visits to the infusion center can likely be mitigated by extended interval dosing.50 , 71 Extended interval dosing for ocrelizumab and rituximab (i.e., delaying maintenance redosing by four or more weeks or personalized redosing based on CD19 cell count) was found not to be associated with increased MS disease activity in the short-term.72, 73, 74, 75 However, a single cohort study of extended interval ocrelizumab dosing raised concern about increased MRI activity without an increase in confirmed disability progression,76 emphasizing that the long-term safety and efficacy of this off-label approach need to be confirmed with prospective studies. When PwMS are diagnosed with COVID-19, several Food and Drug Administration (FDA)-approved antiviral treatments may be available to them depending on their baseline risk factors and severity of illness.77 Of note, Evusheld™ and REGEN-COV are not authorized by the FDA anymore since they no longer cover dominant circulating SARS-CoV-2 variants.77 Vaccination against SARS-CoV-2 for people with MS Since their advent in August 2021, large population-based data shows that SARS-CoV-2 mRNA vaccines are safe and effective in PwMS, as in the general population.78, 79, 80, 81, 82 Adverse effects are typically mild and self-limited (injection site reaction, malaise, headache, fever) and are not associated with increased short-term risk of MS relapse.80 , 81 , 83 Early on, it was not known whether PwMS treated with certain DMT classes (e.g., cell-depleting therapies) would mount adequate immune responses, including memory B cell responses to SARS-CoV-2 vaccination. Multiple studies confirmed that those on cell-depleting therapies (anti-CD20, alemtuzumab) and those on S1P receptor modulators have diminished humoral responses to SARS-CoV-2 vaccination.84, 85, 86, 87, 88 However, T cell-mediated adaptive responses are preserved in many patients on anti-CD20 therapy and may be more robust, suggesting that other immune mechanisms may compensate for a blunted humoral response.86 , 89, 90, 91 These effects persist in the long-term, and affect memory B cells.84 It has become evident that timing of vaccine and DMT administration (see Table 3 ) influences the magnitude of serologic immune response observed.92, 93, 94 However, it is unknown whether these suspected compensatory immune mechanisms or DMT timing result in preventing SARS-CoV-2 infections and/or severe COVID-19 outcomes.Table 3 Summary of recommendations for COVID-19 vaccination in people with MS by treatment status MS Treatment Negative effect on SARS-CoV-2 vaccine response? Timing of vaccine before starting treatment Timing of vaccine after treatment started* No DMT None: having MS in isolation does not affect ability to mount a humoral response to available COVID-19 vaccines78 SARS-CoV-2 vaccines are safe and effective for PwMS and recommended for all patients81 N/A Beta interferon None78,93 Do not delay starting for vaccine No adjustments required; vaccinate when able Glatiramer acetate None78,93 Do not delay starting for vaccine No adjustments required; vaccinate when able Teriflunomide No data to suggest impaired response78,93 Do not delay starting for vaccine No adjustments required; vaccinate when able Fumarates No reduction in humoral or T cell-dependent responses for dimethyl fumarate; unknown for other fumarate DMTs93,105 Do not delay starting for vaccine No adjustments required; vaccinate when able S1P receptor modulators Impairment in humoral response in PwMS on fingolimod92 Vaccinate at least 2 weeks before starting Continue taking the medication as prescribed and vaccinate when able Natalizumab No data to suggest impaired responses for SARS-CoV-2 vaccines93,105 Do not delay starting for vaccine No adjustments required; vaccinate when able Anti-CD20 monoclonal antibodies Impairment in humoral response was observed in several studies, however certain T cell responses were found to be more robust78,86,87,90,106 Vaccinate at least 2 weeks before starting Ideal timing to vaccinate is 4 weeks before next infusion or 4 weeks after last dose of ofatumumab Cladribine Empiric data suggest no impairment in humoral antibody responses; theoretical concern during lymphodepletive treatment phase84 Vaccinate at least 2 weeks before starting No adjustments required; vaccinate when able Alemtuzumab Empiric data suggest no impairment in humoral antibody responses; theoretical concern during lymphodepletive treatment phase93 Vaccinate at least 4 weeks before starting Consider vaccination 24 weeks or more after the last infusion High-dose corticosteroids Not demonstrated with empirical data but theoretical concern exists given mechanism of action Vaccinate 3-5 days after last dose Vaccinate 3-5 days after the last dose *Note that ideal timing of therapy with vaccination may not be possible. Abbreviations: DMT – disease-modifying therapy; S1P – sphingosine-1 phosphate; PwMS – people with MS. It is recommended that all PwMS receive vaccination against SARS-CoV-2 unless there is a contraindication (e.g., allergic reaction) and keep their vaccinations up to date when they are able to receive boosters.82 , 95 Table 3 summarizes recommendations regarding the optimal timing of SARS-CoV-2 vaccinations with various DMTs. Addressing concerns PwMS may have about vaccination is paramount. Surveys suggest that most PwMS are willing to receive SARS-CoV-2 vaccines (>75-90% depending on the study).96 , 97 The most prevalent concerns relate to long-term safety and efficacy.96 , 98 , 99 In one large patient survey (n=∼5000), factors that increased the likelihood of receiving the SARS-CoV-2 vaccine included obtaining influenza vaccine, older age (≥65 years), higher SES, physical activity, and use of DMT.96 People with MS were less likely to be vaccine hesitant when explicitly counseled by their MS provider about risks and benefits.100 Overall, providers should continue to individualize counseling to promote vaccination wherever possible to reduce long-term COVID-19 risk as the pandemic shifts to a more endemic/seasonal pattern globally. Discussion With the end of the COVID-19 global health emergency, it remains unknown whether changes to MS care will persist. A shift to endemic patterns of infection likely means that a degree of normalcy will return. However, the field has demonstrated that MS care can be adapted to meet PwMS in their home environments with telemedicine. Prospective investigation will be required to clarify the extent telemedicine should be used in routine practice based on patient care outcomes. As efficacious DMTs continue to be developed, the MS community can incorporate lessons learned during the COVID-19 pandemic to develop contingency plans in the event of future viral pandemics. DMT and vaccination guidelines will need to discuss suggested actions in the event of future viral outbreaks. PwMS will require information about risk while on DMTs and how vaccine responses may be impacted. Finally, the long-term effects of widespread COVID-19 on PwMS are not yet fully understood. The degree to which long-COVID might interact with MS fatigue and other symptomatology has been explored101 , 102 but relationships have not been conclusively demonstrated. It will also take time to study sustained effects of pandemic-related occupational and psychosocial disruption, which may have disproportionately affected certain PwMS,11 , 13 , 68 , 103 particularly those who were already vulnerable socioeconomically. Observational studies should continue to monitor the long-term impact of the COVID-19 pandemic on MS incidence and disease activity; although preliminary data suggests that its short-term impacts are no different than other respiratory viruses.40 , 102 As we enter the post-pandemic COVID-19 era, the future of person-centered MS care looks as promising as ever. Summary Despite multiple COVID-19 pandemic-related disruptions along the continuum of MS care, providers and PwMS collaborated to develop innovative solutions. There were unprecedented efforts across the MS community to work together to collect as much information as possible to make informed decisions in the care of PwMS. We reviewed considerations for DMT prescription and vaccination for PwMS in the era of COVID-19 and speculate on questions for future research in this area. Disclosures JDK: receives fellowship funding (paid directly to institution) provided by a University of Calgary Medical Group Helios Advanced Training Award. AS: receives research funding (paid directly to institution) from Multiple Sclerosis Canada, National Multiple Sclerosis Society, Consortium of Multiple Sclerosis Centers and the Department of Defense Congressionally Directed Medical Research Program and is a member of the editorial board for Neurology. She serves as a consultant for Gryphon Bio, LLC. She is a member of the Data and Safety Monitoring Board for Premature Infants Receiving Milking or Delayed Cord Clamping (PREMOD2), Central Vein Sign: A Diagnostic Biomarker in Multiple Sclerosis (CAVS-MS), and Methotrexate treatment of Arthritis caused by Chikungunya virus (MARCH). She is supported (in part) by a Biostatistics/Informatics Junior Faculty Award (BI-2105-37656) from the National Multiple Sclerosis Society. She holds the Kenney Marie Dixon‐Pickens Distinguished Professorship in Multiple Sclerosis Research. ==== Refs References 1 World Health Organization Coronavirus Dashboard [Internet]. Internet. Updated 2023 Feb 8. Accessed 2023 Feb 8, https://covid19.who.int/ 2 Centers for Disease Control and Prevention, COVID Data Tracker [Internet]. Internet. Accessed 2023 Feb 8, 2023 Feb 8. https://covid.cdc.gov/covid-data-tracker/#datatracker-home 3 Chen M.H. Goverover Y. Botticello A. DeLuca J. Genova H.M. Healthcare Disruptions and Use of Telehealth Services Among People With Multiple Sclerosis During the COVID-19 Pandemic Arch Phys Med Rehabil 103 7 Jul 2022 1379 1386 10.1016/j.apmr.2021.12.028 35093328 4 Zhang Y. Staker E. Cutter G. Krieger S. Miller A.E. Perceptions of risk and adherence to care in MS patients during the COVID-19 pandemic: A cross-sectional study Mult Scler Relat Disord 50 May 2021 102856 10.1016/j.msard.2021.102856 5 Harris E. WHO Declares End of COVID-19 Global Health Emergency JAMA May 17 2023 10.1001/jama.2023.8656 6 Vogel A.C. Schmidt H. Loud S. McBurney R. Mateen F.J. Impact of the COVID-19 pandemic on the health care of >1,000 People living with multiple sclerosis: A cross-sectional study Mult Scler Relat Disord. Nov 46 2020 102512 10.1016/j.msard.2020.102512 7 Fitzgerald KC, Mecoli CA, Douglas M, et al. Risk Factors for Infection and Health Impacts of the Coronavirus Disease 2019 (COVID-19) Pandemic in People With Autoimmune Diseases. Clin Infect Dis. Feb 11 2022;74(3):427-436. doi:10.1093/cid/ciab407 8 Moss B.P. Mahajan K.R. Bermel R.A. Multiple sclerosis management during the COVID-19 pandemic Mult Scler 26 10 Sep 2020 1163 1171 10.1177/1352458520948231 32772807 9 Mateen FJ, Rezaei S, Alakel N, Gazdag B, Kumar AR, Vogel A. Impact of COVID-19 on U.S. and Canadian neurologists' therapeutic approach to multiple sclerosis: a survey of knowledge, attitudes, and practices. J Neurol. Dec 2020;267(12):3467-3475. doi:10.1007/s00415-020-10045-9 10 Morrison E.H. Michtich K. Hersh C.M. How the COVID-19 Pandemic has changed multiple sclerosis clinical practice: Results of a nationwide provider survey Mult Scler Relat Disord 51 Jun 2021 102913 10.1016/j.msard.2021.102913 11 Goverover Y, Chen MH, Botticello A, et al. Relationships between changes in daily occupations and health-related quality of life in persons with multiple sclerosis during the COVID-19 pandemic. Mult Scler Relat Disord. Jan 2022;57:103339. doi:10.1016/j.msard.2021.103339 12 Parkinson A, Drew J, Hall Dykgraaf S, et al. 'They're getting a taste of our world': A qualitative study of people with multiple sclerosis' experiences of accessing health care during the COVID-19 pandemic in the Australian Capital Territory. Health Expect. Oct 2021;24(5):1607-1617. doi:10.1111/hex.13284 13 Alschuler K.N. Roberts M.K. Herring T.E. Ehde D.M. Distress and risk perception in people living with multiple sclerosis during the early phase of the COVID-19 pandemic Mult Scler Relat Disord 47 Jan 2021 102618 10.1016/j.msard.2020.102618 14 Hatcher-Martin J.M. Busis N.A. Cohen B.H. American Academy of Neurology Telehealth Position Statement Neurology 97 7 Aug 17 2021 334 339 10.1212/WNL.0000000000012185 33986141 15 Bove R. Bevan C. Crabtree E. Toward a low-cost, in-home, telemedicine-enabled assessment of disability in multiple sclerosis Mult Scler 25 11 Oct 2019 1526 1534 10.1177/1352458518793527 30141729 16 Bove R. Garcha P. Bevan C.J. Crabtree-Hartman E. Green A.J. Gelfand J.M. Clinic to in-home telemedicine reduces barriers to care for patients with MS or other neuroimmunologic conditions Neurol Neuroimmunol Neuroinflamm. Nov 5 6 2018 e505 10.1212/NXI.0000000000000505 17 Abbatemarco JR, Hartman J, McGinley M, et al. Providing Person-Centered Care via Telemedicine in the Era of COVID-19 in Multiple Sclerosis. J Patient Exp. 2021;8:2374373520981474. doi:10.1177/2374373520981474 18 McGinley M.P. Ontaneda D. Wang Z. Teleneurology as a Solution for Outpatient Care During the COVID-19 Pandemic Telemed J E Health 26 12 Dec 2020 1537 1539 10.1089/tmj.2020.0137 32552509 19 Kristoffersen E.S. Sandset E.C. Winsvold B.S. Faiz K.W. Storstein A.M. Experiences of telemedicine in neurological out-patient clinics during the COVID-19 pandemic Ann Clin Transl Neurol. Feb 8 2 2021 440 447 10.1002/acn3.51293 20 Corea F, Ciotti S, Cometa A, et al. Telemedicine during the Coronavirus Disease (COVID-19) Pandemic: A Multiple Sclerosis (MS) Outpatients Service Perspective. Neurol Int. Jan 18 2021;13(1):25-31. doi:10.3390/neurolint13010003 21 McGinley M.P. Gales S. Rowles W. Expanded access to multiple sclerosis teleneurology care following the COVID-19 pandemic Mult Scler J Exp Transl Clin. Jan-Mar 7 1 2021 2055217321997467 10.1177/2055217321997467 22 Keszler P. Maloni H. Miles Z. Jin S. Wallin M. Telemedicine and Multiple Sclerosis: A Survey of Health Care Providers Before and During the COVID-19 Pandemic Int J MS Care 24 6 Nov-Dec 2022 266 270 10.7224/1537-2073.2021-103 36545646 23 Li V. Roos I. Monif M. Impact of telehealth on health care in a multiple sclerosis outpatient clinic during the COVID-19 pandemic Mult Scler Relat Disord 63 Jul 2022 103913 10.1016/j.msard.2022.103913 24 Peeters L.M. Parciak T. Walton C. COVID-19 in people with multiple sclerosis: A global data sharing initiative Mult Scler. Sep 26 10 2020 1157 1162 10.1177/1352458520941485 25 Salter A. Fox R.J. Newsome S.D. Outcomes and Risk Factors Associated With SARS-CoV-2 Infection in a North American Registry of Patients With Multiple Sclerosis JAMA Neurol 78 6 Jun 1 2021 699 708 10.1001/jamaneurol.2021.0688 33739362 26 Simpson-Yap S. Pirmani A. Kalincik T. Updated Results of the COVID-19 in MS Global Data Sharing Initiative: Anti-CD20 and Other Risk Factors Associated With COVID-19 Severity. Neurol Neuroimmunol Neuroinflamm Nov 9 6 2022 10.1212/NXI.0000000000200021 27 Louapre C. Collongues N. Stankoff B. Clinical Characteristics and Outcomes in Patients With Coronavirus Disease 2019 and Multiple Sclerosis JAMA Neurol 77 9 Sep 1 2020 1079 1088 10.1001/jamaneurol.2020.2581 32589189 28 Sormani M.P. Schiavetti I. Carmisciano L. COVID-19 Severity in Multiple Sclerosis: Putting Data Into Context Neurol Neuroimmunol Neuroinflamm 9 1 Jan 2022 10.1212/NXI.0000000000001105 29 Bsteh G. Assar H. Hegen H. COVID-19 severity and mortality in multiple sclerosis are not associated with immunotherapy: Insights from a nation-wide Austrian registry PLoS One 16 7 2021 e0255316 10.1371/journal.pone.0255316 30 Arrambide G. Llaneza-Gonzalez M.A. Costa-Frossard Franca L. SARS-CoV-2 Infection in Multiple Sclerosis: Results of the Spanish Neurology Society Registry Neurol Neuroimmunol Neuroinflamm 8 5 Jul 2021 10.1212/NXI.0000000000001024 31 Simpson-Yap S, De Brouwer E, Kalincik T, et al. Associations of Disease-Modifying Therapies With COVID-19 Severity in Multiple Sclerosis. Neurology. Nov 9 2021;97(19):e1870-e1885. doi:10.1212/WNL.0000000000012753 32 Sormani M.P. Salvetti M. Labauge P. DMTs and Covid-19 severity in MS: a pooled analysis from Italy and France Ann Clin Transl Neurol 8 8 Aug 2021 1738 1744 10.1002/acn3.51408 34240579 33 Strangfeld A. Schafer M. Gianfrancesco M.A. Factors associated with COVID-19-related death in people with rheumatic diseases: results from the COVID-19 Global Rheumatology Alliance physician-reported registry Ann Rheum Dis. Jul 80 7 2021 930 942 10.1136/annrheumdis-2020-219498 34 Longinetti E. Bower H. McKay K.A. COVID-19 clinical outcomes and DMT of MS patients and population-based controls Ann Clin Transl Neurol. Sep 9 9 2022 1449 1458 10.1002/acn3.51646 35 Salter A. Cross A.H. Cutter G.R. COVID-19 in the pregnant or postpartum MS patient: Symptoms and outcomes Mult Scler Relat Disord. Sep 65 2022 104028 10.1016/j.msard.2022.104028 36 Oncel I. Alici N. Solmaz I. Oge D.D. Ozsurekci Y. Anlar B. The Outcome of COVID-19 in Pediatric-Onset Multiple Sclerosis Patients Pediatr Neurol 134 Sep 2022 7 10 10.1016/j.pediatrneurol.2022.06.004 35772229 37 Yam C. Jokubaitis V. Hellwig K. Dobson R. MS, pregnancy and COVID-19 Mult Scler 26 10 Sep 2020 1137 1146 10.1177/1352458520949152 32924838 38 Evangelou N. Garjani A. dasNair R. Self-diagnosed COVID-19 in people with multiple sclerosis: a community-based cohort of the UK MS Register J Neurol Neurosurg Psychiatry 92 1 Aug 27 2020 107 109 10.1136/jnnp-2020-324449 32855290 39 Group R.C. Horby P. Lim W.S. Dexamethasone in Hospitalized Patients with Covid-19 N Engl J Med 384 8 Feb 25 2021 693 704 10.1056/NEJMoa2021436 32678530 40 Sepulveda M. Llufriu S. Martinez-Hernandez E. Incidence and Impact of COVID-19 in MS: A Survey From a Barcelona MS Unit Neurol Neuroimmunol Neuroinflamm 2 Mar 4 2021 8 10.1212/NXI.0000000000000954 41 Schiavetti I. Carmisciano L. Ponzano M. Signs and symptoms of COVID-19 in patients with multiple sclerosis Eur J Neurol 29 12 Dec 2022 3728 3736 10.1111/ene.15554 36086905 42 Parrotta E. Kister I. Charvet L. COVID-19 outcomes in MS: Observational study of early experience from NYU Multiple Sclerosis Comprehensive Care Center Neurol Neuroimmunol Neuroinflamm. Sep 7 5 2020 10.1212/NXI.0000000000000835 43 Garjani A. Middleton R.M. Nicholas R. Evangelou N. Recovery From COVID-19 in Multiple Sclerosis: A Prospective and Longitudinal Cohort Study of the United Kingdom Multiple Sclerosis Register Neurol Neuroimmunol Neuroinflamm. Jan 9 1 2022 10.1212/NXI.0000000000001118 44 Bellucci G. Rinaldi V. Buscarinu M.C. Multiple Sclerosis and SARS-CoV-2: Has the Interplay Started? Front Immunol 12 2021 755333 10.3389/fimmu.2021.755333 45 Etemadifar M, Sedaghat N, Aghababaee A, et al. COVID-19 and the Risk of Relapse in Multiple Sclerosis Patients: A Fight with No Bystander Effect? Mult Scler Relat Disord. Jun 2021;51:102915. doi:10.1016/j.msard.2021.102915 46 Garjani A. Middleton R.M. Hunter R. COVID-19 is associated with new symptoms of multiple sclerosis that are prevented by disease modifying therapies Mult Scler Relat Disord 52 Jul 2021 102939 10.1016/j.msard.2021.102939 47 Kataria S. Tandon M. Melnic V. Sriwastava S. A case series and literature review of multiple sclerosis and COVID-19: Clinical characteristics, outcomes and a brief review of immunotherapies eNeurologicalSci. Dec 21 2020 100287 10.1016/j.ensci.2020.100287 48 Cabreira V. Abreu P. Soares-Dos-Reis R. Guimaraes J. Sa M.J. Multiple Sclerosis, Disease-Modifying Therapies and COVID-19: A Systematic Review on Immune Response and Vaccination Recommendations Vaccines (Basel) 9 7 Jul 11 2021 10.3390/vaccines9070773 49 Chaudhry F. Jageka C. Levy P.D. Cerghet M. Lisak R.P. Review of the COVID-19 Risk in Multiple Sclerosis J Cell Immunol 3 2 2021 68 77 10.33696/immunology.3.080 33959727 50 Zheng C. Kar I. Chen C.K. Multiple Sclerosis Disease-Modifying Therapy and the COVID-19 Pandemic: Implications on the Risk of Infection and Future Vaccination CNS Drugs. Sep 34 9 2020 879 896 10.1007/s40263-020-00756-y 51 Giovannoni G. Hawkes C. Lechner-Scott J. Levy M. Waubant E. Gold J. The COVID-19 pandemic and the use of MS disease-modifying therapies Mult Scler Relat Disord. Apr 39 2020 102073 10.1016/j.msard.2020.102073 52 Hollen C. Bernard J. Multiple Sclerosis Management During the COVID-19 Pandemic Curr Neurol Neurosci Rep. Aug 22 8 2022 537 543 10.1007/s11910-022-01211-9 53 Simpson-Yap S. Pirmani A. De Brouwer E. Severity of COVID19 infection among patients with multiple sclerosis treated with interferon-beta Mult Scler Relat Disord 66 Oct 2022 104072 10.1016/j.msard.2022.104072 54 Freedman M.S. Jack D. Murgasova Z. Todorovic M. Seitzinger A. Outcomes of COVID-19 among patients treated with subcutaneous interferon beta-1a for multiple sclerosis Mult Scler Relat Disord. Nov 56 2021 103283 10.1016/j.msard.2021.103283 55 Mohn N. Konen F.F. Pul R. Experience in Multiple Sclerosis Patients with COVID-19 and Disease-Modifying Therapies: A Review of 873 Published Cases J Clin Med 12 Dec 16 2020 9 10.3390/jcm9124067 33374507 56 Tasat D.R. Yakisich J.S. Rationale for the use of sphingosine analogues in COVID-19 patients Clin Med (Lond). Jan 21 1 2021 e84 e87 10.7861/clinmed.2020-0309 57 Teymouri S. Pourbayram Kaleybar S. Hejazian S.S. The effect of Fingolimod on patients with moderate to severe COVID-19 Pharmacol Res Perspect. Feb 11 1 2023 e01039 10.1002/prp2.1039 58 Ciotti J.R. Grebenciucova E. Moss B.P. Newsome S.D. Multiple Sclerosis Disease-Modifying Therapies in the COVID-19 Era Ann Neurol 88 6 Dec 2020 1062 1064 10.1002/ana.25907 32951235 59 Portaccio E. Fonderico M. Hemmer B. Impact of COVID-19 on multiple sclerosis care and management: Results from the European Committee for Treatment and Research in Multiple Sclerosis survey Mult Scler. Jan 28 1 2022 132 138 10.1177/13524585211005339 60 Williams T, Mishra R, Bharkhada B, et al. Impact of the COVID-19 pandemic on the prescription of disease-modifying therapy for multiple sclerosis in England: a nationwide study. J Neurol Neurosurg Psychiatry. Mar 10 2022;93(11):1229-1230. doi:10.1136/jnnp-2021-328340 61 Cobo-Calvo A. Zabalza A. Rio J. Impact of COVID-19 pandemic on frequency of clinical visits, performance of MRI studies, and therapeutic choices in a multiple sclerosis referral centre J Neurol. Apr 269 4 2022 1764 1772 10.1007/s00415-021-10958-z 62 Zaheer R. Amin R. Riddick L. Impact of COVID-19 on prescribing patterns and treatment selection of disease modifying therapies in multiple sclerosis Mult Scler Relat Disord. Mar 71 2023 104575 10.1016/j.msard.2023.104575 63 Fox R.J. Cosenza C. Cripps L. A survey of risk tolerance to multiple sclerosis therapies Neurology 92 14 Apr 2 2019 e1634 e1642 10.1212/WNL.0000000000007245 30867272 64 National MS Society, Disease Modifying Therapy Guidance During COVID-19 [Internet]. Updated 2023. https://www.nationalmssociety.org/coronavirus-covid-19-information/ms-treatment-guidelines-during-coronavirus 65 MS International Federation, MS, COVID-19 and Vaccines - Updated Global Advice [Internet]. Updated May 24, 2022. https://www.msif.org/news/2020/02/10/the-coronavirus-and-ms-what-you-need-to-know/ 66 Pugliatti M. Berger T. Hartung H.P. Oreja-Guevara C. Bar-Or A. Multiple sclerosis in the era of COVID-19: disease course, DMTs and SARS-CoV2 vaccinations Curr Opin Neurol 35 3 Jun 1 2022 319 327 10.1097/WCO.0000000000001066 35674075 67 Sormani M.P. De Rossi N. Schiavetti I. Disease-Modifying Therapies and Coronavirus Disease 2019 Severity in Multiple Sclerosis Ann Neurol 89 4 Apr 2021 780 789 10.1002/ana.26028 33480077 68 Abbadessa G. Lavorgna L. Trojsi F. Coppola C. Bonavita S. Understanding and managing the impact of the Covid-19 pandemic and lockdown on patients with multiple sclerosis Expert Rev Neurother. Jul 21 7 2021 731 743 10.1080/14737175.2021.1957673 69 Luna G, Alping P, Burman J, et al. Infection Risks Among Patients With Multiple Sclerosis Treated With Fingolimod, Natalizumab, Rituximab, and Injectable Therapies. JAMA Neurol. Feb 1 2020;77(2):184-191. doi:10.1001/jamaneurol.2019.3365 70 Sullivan R. Kilaru A. Hemmer B. COVID-19 Infection in Fingolimod- or Siponimod-Treated Patients: Case Series Neurol Neuroimmunol Neuroinflamm. Nov 9 1 2021 10.1212/NXI.0000000000001092 71 Foley J.F. Defer G. Ryerson L.Z. Comparison of switching to 6-week dosing of natalizumab versus continuing with 4-week dosing in patients with relapsing-remitting multiple sclerosis (NOVA): a randomised, controlled, open-label, phase 3b trial Lancet Neurol. Jul 21 7 2022 608 619 10.1016/S1474-4422(22)00143-0 72 Maarouf A. Rico A. Boutiere C. Extending rituximab dosing intervals in patients with MS during the COVID-19 pandemic and beyond? Neurol Neuroimmunol Neuroinflamm 7 5 Sep 2020 10.1212/NXI.0000000000000825 73 Rolfes L. Pawlitzki M. Pfeuffer S. Ocrelizumab Extended Interval Dosing in Multiple Sclerosis in Times of COVID-19 Neurol Neuroimmunol Neuroinflamm 8 5 Sep 2021 10.1212/NXI.0000000000001035 74 Sahi N.K. Abidi S.M.A. Salim O. Abraham R. Kalra S. Al-Araji A. Clinical impact of Ocrelizumab extended interval dosing during the COVID-19 pandemic and associations with CD19(+)B-cell repopulation Mult Scler Relat Disord. Nov 56 2021 103287 10.1016/j.msard.2021.103287 75 van Lierop Z.Y. Toorop A.A. van Ballegoij W.J. Personalized B-cell tailored dosing of ocrelizumab in patients with multiple sclerosis during the COVID-19 pandemic Mult Scler. Jun 28 7 2022 1121 1125 10.1177/13524585211028833 76 Zanghi A. Avolio C. Signoriello E. Is It Time for Ocrelizumab Extended Interval Dosing in Relapsing Remitting MS? Evidence from An Italian Multicenter Experience During the COVID-19 Pandemic Neurotherapeutics. Sep 19 5 2022 1535 1545 10.1007/s13311-022-01289-6 77 National MS Society, Medicines to Prevent and Treat COVID-19 [Internet]. Updated January 27, 2023. https://www.nationalmssociety.org/coronavirus-covid-19-information/suspected-covid-19-and-ms 78 Achiron A. Dolev M. Menascu S. COVID-19 vaccination in patients with multiple sclerosis: What we have learnt by February 2021 Mult Scler 27 6 May 2021 864 870 10.1177/13524585211003476 33856242 79 Garjani A. Patel S. Bharkhada D. Impact of mass vaccination on SARS-CoV-2 infections among multiple sclerosis patients taking immunomodulatory disease-modifying therapies in England Mult Scler Relat Disord. Jan 57 2022 103458 10.1016/j.msard.2021.103458 80 Lotan I. Wilf-Yarkoni A. Friedman Y. Stiebel-Kalish H. Steiner I. Hellmann M.A. Safety of the BNT162b2 COVID-19 vaccine in multiple sclerosis (MS): Early experience from a tertiary MS center in Israel Eur J Neurol. Nov 28 11 2021 3742 3748 10.1111/ene.15028 81 Stefanou M.I. Palaiodimou L. Theodorou A. Safety of COVID-19 vaccines in multiple sclerosis: A systematic review and meta-analysis Mult Scler. Apr 29 4-5 2023 585 594 10.1177/13524585221150881 82 Coyle P.K. Gocke A. Vignos M. Newsome S.D. Vaccine Considerations for Multiple Sclerosis in the COVID-19 Era Adv Ther. Jul 38 7 2021 3550 3588 10.1007/s12325-021-01761-3 83 Frahm N, Fneish F, Ellenberger D, et al. Frequency and Predictors of Relapses following SARS-CoV-2 Vaccination in Patients with Multiple Sclerosis: Interim Results from a Longitudinal Observational Study. J Clin Med. May 24 2023;12(11)doi:10.3390/jcm12113640 84 Disanto G. Galante A. Cantu M. Longitudinal Postvaccine SARS-CoV-2 Immunoglobulin G Titers, Memory B-Cell Responses, and Risk of COVID-19 in Multiple Sclerosis Over 1 Year Neurol Neuroimmunol Neuroinflamm 1 Jan 2023 10 10.1212/NXI.0000000000200043 85 Wu X. Wang L. Shen L. Tang K. Response of COVID-19 vaccination in multiple sclerosis patients following disease-modifying therapies: A meta-analysis EBioMedicine. Jul 81 2022 104102 10.1016/j.ebiom.2022.104102 86 Brill L. Rechtman A. Zveik O. Humoral and T-Cell Response to SARS-CoV-2 Vaccination in Patients With Multiple Sclerosis Treated With Ocrelizumab JAMA Neurol 78 12 Dec 1 2021 1510 1514 10.1001/jamaneurol.2021.3599 34554197 87 Kornek B. Leutmezer F. Rommer P.S. B Cell Depletion and SARS-CoV-2 Vaccine Responses in Neuroimmunologic Patients Ann Neurol. Mar 91 3 2022 342 352 10.1002/ana.26309 88 Cohen J.A. Bermel R.A. Grossman C.I. Immunoglobulin G immune response to SARS-CoV-2 vaccination in people living with multiple sclerosis within Multiple Sclerosis Partners Advancing Technology and Health Solutions Mult Scler. Jun 28 7 2022 1131 1137 10.1177/13524585211061343 89 Ali A. Dwyer D. Wu Q. Characterization of humoral response to COVID mRNA vaccines in multiple sclerosis patients on disease modifying therapies Vaccine 39 41 Oct 1 2021 6111 6116 10.1016/j.vaccine.2021.08.078 34483021 90 Apostolidis S.A. Kakara M. Painter M.M. Cellular and humoral immune responses following SARS-CoV-2 mRNA vaccination in patients with multiple sclerosis on anti-CD20 therapy Nat Med. Nov 27 11 2021 1990 2001 10.1038/s41591-021-01507-2 91 Gadani S.P. Reyes-Mantilla M. Jank L. Discordant humoral and T cell immune responses to SARS-CoV-2 vaccination in people with multiple sclerosis on anti-CD20 therapy EBioMedicine. Nov 73 2021 103636 10.1016/j.ebiom.2021.103636 92 Achiron A. Mandel M. Gurevich M. Immune response to the third COVID-19 vaccine dose is related to lymphocyte count in multiple sclerosis patients treated with fingolimod J Neurol. May 269 5 2022 2286 2292 10.1007/s00415-022-11030-0 93 Tallantyre E.C. Vickaryous N. Anderson V. COVID-19 Vaccine Response in People with Multiple Sclerosis Ann Neurol. Jan 91 1 2022 89 100 10.1002/ana.26251 94 Disanto G. Sacco R. Bernasconi E. Association of Disease-Modifying Treatment and Anti-CD20 Infusion Timing With Humoral Response to 2 SARS-CoV-2 Vaccines in Patients With Multiple Sclerosis JAMA Neurol 78 12 Dec 1 2021 1529 1531 10.1001/jamaneurol.2021.3609 34554185 95 National MS Society, COVID-19 Vaccine Guidance for People Living with MS [Internet]. Updated February 23, 2023. https://www.nationalmssociety.org/coronavirus-covid-19-information/covid-19-vaccine-guidance 96 Marrie RA, Dolovich C, Cutter GR, Fox RJ, Salter A. Attitudes toward coronavirus disease 2019 vaccination in people with multiple sclerosis. Mult Scler J Exp Transl Clin. Apr-Jun 2022;8(2):20552173221102067. doi:10.1177/20552173221102067 97 Ciotti JR, Perantie DC, Moss BP, et al. Perspectives and experiences with COVID-19 vaccines in people with MS. Mult Scler J Exp Transl Clin. Jan-Mar 2022;8(1):20552173221085242. doi:10.1177/20552173221085242 98 Ehde D.M. Roberts M.K. Humbert A.T. Herring T.E. Alschuler K.N. COVID-19 vaccine hesitancy in adults with multiple sclerosis in the United States: A follow up survey during the initial vaccine rollout in 2021 Mult Scler Relat Disord 54 Sep 2021 103163 10.1016/j.msard.2021.103163 99 Uhr L. Mateen F.J. COVID-19 vaccine hesitancy in multiple sclerosis: A cross-sectional survey Mult Scler. Jun 28 7 2022 1072 1080 10.1177/13524585211030647 100 Yap S.M. Al Hinai M. Gaughan M. Vaccine hesitancy among people with multiple sclerosis Mult Scler Relat Disord. Nov 56 2021 103236 10.1016/j.msard.2021.103236 101 Barzegar M. Mirmosayyeb O. Gajarzadeh M. COVID-19 Among Patients With Multiple Sclerosis: A Systematic Review Neurol Neuroimmunol Neuroinflamm. Jul 8 4 2021 10.1212/NXI.0000000000001001 102 Bsteh G. Assar H. Gradl C. Long-term outcome after COVID-19 infection in multiple sclerosis: a nation-wide multicenter matched-control study Eur J Neurol Jun 25 2022 10.1111/ene.15477 103 Chiaravalloti N.D. Amato M.P. Brichetto G. The emotional impact of the COVID-19 pandemic on individuals with progressive multiple sclerosis J Neurol 268 5 May 2021 1598 1607 10.1007/s00415-020-10160-7 32813051 104 Mehta D. Miller C. Arnold D.L. Effect of dimethyl fumarate on lymphocytes in RRMS: Implications for clinical practice Neurology 92 15 Apr 9 2019 e1724 e1738 10.1212/WNL.0000000000007262 30918100 105 Levit E. Longbrake E.E. Stoll S.S. Seroconversion after COVID-19 vaccination for multiple sclerosis patients on high efficacy disease modifying medications Mult Scler Relat Disord. Apr 60 2022 103719 10.1016/j.msard.2022.103719 106 Bar-Or A. Calkwood J.C. Chognot C. Effect of ocrelizumab on vaccine responses in patients with multiple sclerosis: The VELOCE study Neurology 95 14 Oct 6 2020 e1999 e2008 10.1212/WNL.0000000000010380 32727835
PMC010xxxxxx/PMC10288316.txt	==== Front Energy Policy Energy Policy Energy Policy 0301-4215 0301-4215 Elsevier Ltd. S0301-4215(23)00263-X 10.1016/j.enpol.2023.113678 113678 Article The concern about main crises such as the Covid-19 pandemic, the war in Ukraine, and climate change's impact on energy-saving behavior Liobikienė Genovaitė a∗ Matiiuk Yuliia a Krikštolaitis Ričardas bc a Department of Environmental Sciences, Vytautas Magnus University, Vileikos st. 8, LT-44404, Kaunas, Lithuania b Department of Mathematics and Statistics, Vytautas Magnus University, Universiteto str. 10, Akademija, LT, 53361, Kaunas Dist, Lithuania c Lithuanian Energy Institute, Breslaujos str. 3, LT-44403, Kaunas, Lithuania ∗ Corresponding author. 23 6 2023 23 6 2023 11367824 3 2023 1 6 2023 11 6 2023 © 2023 Elsevier Ltd. All rights reserved. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The number of crises experienced around the world forces people to reconsider and reassess various aspects of their lives. The energy crisis caused by the war in Ukraine and uncontrolled climate change revealed the importance of energy-saving behavior. Thus, the aim of this paper is to analyze the concerns about current crises such as the Covid-19 pandemic, the war in Ukraine, and climate change's impact on energy-saving behavior and changes in environmental concern. Referring to the survey conducted in Lithuania in 2022, where 1000 respondents participated, the results revealed that the war in Ukraine was the most concerning problem. The level of climate change concern was slightly lower. Meanwhile, the Covid-19 pandemic was the least important problem in Lithuania in 2022. Furthermore, respondents stated that the Covid-19 pandemic contributed to the changes in environmental concern and energy-saving actions more than the war in Ukraine did. Meanwhile, the Generalized Linear Model results revealed that only the war in Ukraine positively and significantly influenced energy-saving behavior. The Covid-19 pandemic concern negatively affected energy-saving behavior, while the climate change concern factor affected it indirectly, as the interaction of attitudes toward energy consumption. Thus, this study revealed the main aspect of and how to encourage energy-saving behavior in the context of the main current crises. Keywords Energy-saving behavior Climate change The Covid-19 pandemic The war in Ukraine Attitudes Pro-environmental behavior ==== Body pmc1 Introduction In recent years, the world has faced more crises than ever before. The war in Ukraine caused big insecurity in all European Union countries, particularly in the Baltic countries. The energy crisis and increasing prices of this resource imbalance the EU countries’ economies (Kuzemko et al., 2022; Zakeri et al., 2022). Lithuania is the country where the prices of energy increased the most among the EU countries. Thus, the promotion of renewable energy production and energy saving has become one of the most important tools to solve the energy crisis (Trypolska and Rosner, 2022; Creutzig, 2022). The encouragement of energy-saving behavior is the easiest tool which does not require a lot of investment. Thus, almost all of the EU policymakers encouraged people to save energy not only on the institutional level but in their households as well. The war in Ukraine is a potential opportunity to begin to change energy consumption behavior and mitigate not only the energy crisis but climate change as well. However, how the concern about the war in Ukraine has contributed to energy-saving behavior lacks scientific research. Another crisis which touched all of the world is the Covid-19 pandemic. The pandemic disrupted not only the health system but also the safety of economy and ecology around the world (Erokhin and Gao, 2020; Bouarar et al., 2020; Klenert et al., 2020). In literature, scientists analyzed not only how the Covid-19 pandemic is related to environmental pollution (Al Huraimel et al., 2020; Siddique et al., 2021; Maatoug et al., 2021), but also how the pandemic influenced resource-saving behavior (Pop et al., 2022; Mustapa et al., 2021; Ueno, 2022; Jiang et al., 2021). However, little attention has been paid to how this pandemic influenced people's behavior post-Covid-19 – when the virus still exists, but strict management tools such as quarantine are withdrawn (Ahmad et al., 2022). Authors declared that the pandemic can successfully change behavior in a short period of time (Zakeri et al., 2022; Mi et al., 2021; Lucarelli et al., 2020), but whether the behaviors are reverted after lockdown is also important to analyze. Climate change is the last crisis which remains long-term (Boto-García and Bucciol, 2020). Throughout this century, the governments have been discussing and searching for tools to mitigate climate change. Energy consumption reduction is the main way we can reduce greenhouse gas emissions even also considering the household sector (Trotta, 2018; Huang et al., 2022). Thus, authors analyzed the main determinants of energy-saving behavior vastly (Park and Kwon, 2017; Carrus et al., 2021; Shrestha et al., 2021). The concern about climate change is one of the main drivers which influences energy-saving behavior (Bouman et al., 2020; Ohler and Billger, 2014; Nauges and Wheeler, 2017; Boto-García and Bucciol, 2020; Gregersen et al., 2021). However, to the best of our knowledge, none of the other authors analyzed how the climate change crisis impacts such behavior in context of other crises such as the war in Ukraine or the Covid-19 pandemic. Thus, the aim of this paper is to analyze how concerns about present crises, such as war in Ukraine, Covid-19 pandemic, and climate change impact the energy-saving behavior and environmental concern. Thus, the aim of this paper is to analyze the concerns about current crises (the war in Ukraine, the Covid-19 pandemic, and climate change) impact on energy-saving behavior. Considering that the encouragement of energy-saving behavior in recent years is escalated by media and policymakers, it is vital to reveal the main determinants of this behavior. 2 Literature review 2.1 Historical precedence of different crises The current crises due to the globalization concerned all the world. However, the rather similar crises occurred in all centuries and different regions. For example, cholera pandemic (1961-ongoing) began in the Southeast Asian archipelago and rapidly spread around the rest of the globe (Kaper et al., 1995). Thus, served as a stark reminder for society to recognize the delicate nature of biodiversity and the intricate links between human health and climate change (Colwell and Huq, 1994). It is transmitted through untreated water and food, and occasionally from an infected individual to a healthy one (Clemens et al., 2017). The emergence of diseases can be attributed to a range of human activities, while various factors such as climatic, environmental, technological, social, economic, and political can influence the development of diseases and their progression into pandemics within populations (Colwell, 1997). Several studies have led scientists to believe that the cholera pandemic may be directly related to climate change-influenced variables (Craig, 2018). This, in turn, increases the risk of a cholera pandemic. Many authors also indicated that climatic factors, such as soil moisture, land surface, chlorophyll-a concentration, sea surface salinity, sea level, and rainfalls can have an effect on cholera pandemic outbreaks (Montilla et al., 1996; Lobitz et al., 2000; Lipp et al., 2002; Pascual et al., 2002; Chowdhury et al., 2017; Campbell et al., 2020). Koelle et al. (2005) revealed that El Tor variant of cholera takes advantage when monsoon rainfall becomes more extreme. Therefore, these findings suggest that Vibrio cholerae pandemic outbreaks can be linked to climate change. As energy crisis we can compare with Oil Crisis (1973) had significant implications on the world economy and for the restructuring of energy policies (Ross, 2013). The precedent for this crisis was set when the United States and other Western countries backed Israel during the Yom Kippur War (Ross, 2013). However, it made Governments and policymakers of countries, affected by the embargo to re-access their energy strategies, which resulted in a concentration on minimizing the use of energy supplies, energy efficiency, and the development of new energy sources (green energy) (Council Resolution of 17 December 1974; COUNCIL RESOLUTION of 9 June, 1980). 2.2 The impact of war in Ukraine on energy-saving behavior The outbreak of the Russian war against Ukraine left Europe and the rest of the world in a profound state of shock and exposed its dependence on Russian energy supplies. To counter Russian aggression, the European Union and its allied countries implemented sanctions, which included the suspension of oil, coal, and gas imports (Kuzemko et al., 2022). While the energy crisis had its roots in the Covid-19 period, it was significantly exacerbated by the Russian war against Ukraine, leading to a substantial increase in energy costs in global markets and a worsening of the resource imbalance in the European Union's economy, which impacted consumers (Chomać-Pierzecka et al., 2022). In response to the energy crisis and the rising energy costs for households, several EU countries have announced new policies on their local levels to help consumers. However, these measures only offer short-term relief (Kuzemko et al., 2022). Zakeri et al. (2022) suggested transforming energy demand and consumption towards responsible, sustainable, and sufficient ways. Trypolska and Rosner (2022) prioritized environmentally friendly appliances such as solar panels and water heaters to combat the energy shortage. Therefore, the crisis has accelerated the trend toward energy efficiency in Europe and led to a greater focus on reducing energy consumption and promoting sustainable energy sources. The encouragement of energy-saving behavior is the easiest tool to tackle the energy crisis. Policymakers also intensively highlighted this behavior as one of the alternatives to reduce the expenditure of energy resources. However, there is a lack of research on how the concern about the war in Ukraine contributed to energy-saving behavior, which mainly caused energy crisis. 2.3 The impact of the Covid-19 pandemic on energy-saving behavior The Covid-19 pandemic has profoundly affected global energy consumption patterns, as widespread lockdowns and social distancing measures have resulted in a significant decrease in energy demand across multiple sectors. As a result of the pandemic, people started to spend more time at home, caused by social distancing rules or the simple fear of being infected, shortages of working places, and an increase in working from home. Movement regulations consequently led to reduced energy consumption in offices and commercial buildings but an increase in residential energy consumption. Ueno (2022) discovered increasing energy consumption in households caused by Covid-19 pandemic. The pandemic increased not only the concern about health (Bae and Chang, 2021; Ahorsu et al., 2020; Schimmenti et al., 2020), but also concern about environmental problems. Gupta et al. (2021) and Lawler et al. (2021) link the Covid-19 pandemic to the climate change issue, leading individuals to re-evaluate their lifestyles. The research conducted by Lucarelli et al. (2020) revealed that individuals with a higher awareness of the interconnections between Covid-19 and climate change display increased intention and strengthened pro-environmental behaviors. According to Zebardast and Radaei (2022), the pandemic positively influenced pro-environmental attitudes and the relationship between pro-environmental intentions and pro-environmental actions. Mi et al. (2021) analyzed whether public knowledge of Covid-19 can predict pro-environmental behavior intentions. The results revealed that not only does Covid-19 serve as a mediator between emergency cognition and pro-environmental behavioral intentions, but positive environmental affective reactions also have a significant positive effect on the household sphere of pro-environmental behavioral intentions. Considering the energy-saving behavior, Jiang et al. (2021) showed the substantial impact of this pandemic on energy consumption. Ahmad et al. (2022) investigated the impact of cognitive and emotional factors on intentions to save energy and the energy-saving behavior with perceived Covid-19 disruptiveness as a moderator. The results indicate that the perceived disruptiveness of Covid-19 positively moderates this relationship. According to the findings of Pop et al. (2022), it was observed that consumers exhibited heightened awareness regarding energy consumption during the Covid-19 pandemic, resulting in increased energy-saving practices. Meanwhile, Matiiuk et al. (2023) revealed that tools assigned to stop the Covid-19 pandemic did not influence the resource-saving behavior. To the best of our knowledge, it was not analyzed how the pandemic contributes to energy-saving behavior during the post-Covid-19 period. 2.4 The impact of climate change on energy-saving behavior In recent years, increasing focus and concern have been directed toward climate change. More and more people are becoming aware of the importance of reducing emissions of greenhouse gases to mitigate climate change. Climate change concern is one of the main determinants of energy-saving behavior, which refers to the level of worry or anxiety that individuals or societies feel about the potential impacts of climate change on the environment, economy, and human well-being (Whitmarsh and Capstick, 2018; Lacroix and Gifford, 2018; Fierros-Gonzalez and Lopez-Feldman, 2021). Carrus et al. (2021) found that climate change beliefs have a positive to moderate association with energy conservation behaviors. The study of Whitmarsh and O'Neill (2010) discovered that importance of climate change is a significant factor of political actions and all energy/water conservation actions. Gregersen et al. (2021) and Matiiuk et al. (2023) found concern about climate change was a strong predictor of people engaging in energy-saving behaviors. The results, obtained by Boto-García and Bucciol (2020) suggest that increases in climate change concern positively impact energy-saving behavior among European countries‘ citizens as well. However, according to Ohler and Billger (2014), environmental concern has little effect on annual electricity usage. 2.5 The impact of attitudes toward energy consumption impact on energy-saving behavior The attitude toward energy consumption as an internal factor is also an important factor for energy saving (conservation) behavior Brounen et al. (2013), Chen et al. (2017) and Boomsma et al. (2019) found that this factor directly influenced pro-environmental behavior. However, Paҫo and Lavrador (2017), Vringer et al. (2007) and Huebner et al. (2015) revealed that attitudinal factor weakly influenced energy consumption. Meanwhile, Jakučionytė-Skodienė et al. (2020) found an insignificant relationship between electricity consumption and attitude towards energy consumption, when environmental awareness does not give a positive impact on pro-environmental behavior due to various barriers such as lack of knowledge, and social pressure. Moreover, other authors declared that attitudes such as catalytic effect can influence pro-environmental behaviors (Paҫo and Lavrador in 2017, Barber et al., 2009; Flamm, 2009; Polonsky et al., 2012; Liobikiene et al., 2016). Thus, in this paper, both affects (direct and indirect) were analyzed, which reveals whether this attitude (via climate change concern) influenced the energy-saving behavior. 3 Methodology 3.1 The characteristics of the participants of the survey The data of the survey conducted in April 2022 was referred to evaluate how the concern about the main crises (the war in Ukraine, the Covid-19 pandemic, and climate change) contributed to attitudes and energy-saving behavior in Lithuania. The respondents were randomly selected considering the main societies’ age, gender and living space proportion. The quota sampling method was applied. Respondents were questioned via telephone. 1000 respondents participated in this survey. According to the socio-demographic characteristics provided in Table 1 , 45.5% of the respondents were men, while the remaining respondents were women. The average age of respondents was 44.5. Half of the respondents have higher education and live in old-built (50-20 years, blocked with central heating) non-renovated apartments. Notably, the share of income levels displayed a relatively equitable pattern.Table 1 Characteristics of respondents. Table 1 Respondent characteristics N = 1000 Number Percentage Gender Men 455 45.5 Women 545 54.5 Age average 44.5 Education (%) Primary 3 0.3 Lower Secondary 38 3.8 Vocational 61 6.1 Secondary 227 22.7 Non-university higher 194 19.4 Higher 477 47.7 Income per family member (EUR/month) <300 EUR 61 6.17 301-500 EUR 140 14.0 501-700 EUR 203 20.3 701-1000 EUR 215 21.5 1001> EUR 208 20.8 Difficult to say 173 17.3 Type of house (%) Old-built non-renovated apartments 476 47.6 Old-built renovated apartments 139 13.9 New-built apartments 51 5.1 Detached house 294 29.4 Cottage 40 4.0 3.2 The methods of scales’ construction In order to reveal how the concern about the main crises influenced energy-saving behavior the survey data was referred to. The impact of concern about climate change, the Covid-19 pandemic, and the war in Ukraine was analyzed. The concern levels of all these crises were measured referring to the Likert scale (1- “this problem is not serious” - 10- “this problem is extremely serious”). The same method was used in the Eurobarometer surveys when the concern level related to climate change was considered (Jakučionytė-Skodienė and Liobikienė, 2022). The other scales were constructed using the statistical factor analysis. There were scales constructed as changes in environmental concern caused by the war in Ukraine and the Covid-19 pandemic, attitude towards energy consumption, and energy-saving behavior. In this factor analysis, the rotated components matrix was applied in order to guarantee that the constructs would be unrelated and uncorrelated. The value of Kaisеr-Mеyer-Olkin (KMO) indicator (0.959) showed that the used constructs are suitable for factor analysis (Chia et al., 2016; Field, 2013). The Bartlеtt's tеst result was statistically significant (p < .001) and confirmed the suitability of factor analysis as well (Kaiser, 1974). The factor loading coefficient, which exceeded the 0.5 boundary also revealed that factor analysis and construction of the scales were satisfactory (Table 2 ). The scales of changes in environmental concern caused by the war in Ukraine and the Covid-19 pandemic were constructed considering how individuals evaluated the changes in environmental concern due to the existing crises. The items for this construct were adopted using Lin and Syrgabayeva's (2016) scale. The scales of energy-saving behavior and attitudes toward energy consumption were prepared according to Paço and Lavrador (2017) study and adjusted taking into account the situation in Lithuania. All these scales were measured using a five-point Likert scale (from 1 “I strongly disagree” to 5 “I strongly agree”). Evaluating the reliability, the values of Cronbach's alpha coefficients were used and the values of 0.70–0.98 revealed highly reliable scales (DeVellis, 2003; Bland and Altman, 1997; Hair et al., 2010) (Table 2). Therefore, the constructs of analyzed variables were well-constructed and stable.Table 2 Rotated component matrix of analyzed constructs, reliability statistics, and mean score. Table 2 Loading coefficients Variance Explanation (%) Cronbach alpha Mean Standard deviation Changes in environmental concern caused by war in Ukraine War in Ukraine adds to my concern about the environment 0.850 52.024 0.979 2.42 1.127 War in Ukraine adds to my concern about environmental problems 0.850 2.46 1.136 War in Ukraine adds to my concern about climate change 0.851 2.40 1.140 War in Ukraine adds to my concern about waste problems 0.852 2.39 1.136 War in Ukraine adds to my concern about water pollution 0.864 2.41 1.135 War in Ukraine adds to my concern about air pollution 0.850 2.42 1.136 War in Ukraine adds to my concern about biodiversity loss 0.858 2.36 1.095 Changes in environmental concern caused by Covid-19 pandemic Covid-19 adds to my concern about the environment 0.821 9.798 0.966 2.74 1.093 Covid-19 adds to my concern about environmental problems 0.824 2.65 1.068 Covid-19 adds to my concern about climate change 0.827 2.63 1.105 Covid-19 adds to my concern about waste problems 0.825 2.64 1.090 Covid-19 adds to my concern about water pollution 0.806 2.62 1.089 Covid-19 adds to my concern about air pollution 0.831 2.65 1.085 Covid-19 adds to my concern about biodiversity loss 0.791 2.59 1.051 Attitudes toward energy consumption My energy consumption influences the general energy consumption level 0.764 8.970 0.746 3.20 1.041 I can influence government actions related to solutions 0.724 2.37 1.040 I can influence the actions of various companies related to energy issues 0.725 2.38 1.040 My personal energy consumption contributes to climate change problems 0.720 3.30 0.986 Energy saving behavior I turn off the light when I do not need it 0.751 6.737 0.691 4.44 0.832 I use power saving bulbs (fluorescent lamps or light-emitting diodes (LEDs) 0.818 4.19 0.904 I purchase and use energy efficiency class of the electrical appliances (A+, A++) 0.747 3.86 0.994 3.3 The statistical methods and the model used In this paper, the differences between concern about climate change, the Covid-19 pandemic, and the war in Ukraine were assessed using the non-parametric Friedman test. Differences in changes in environmental concern caused by the war in Ukraine and by the Covid-19 pandemic as well as differences of separate actions related to energy-saving caused by the war in Ukraine and the Covid-19 pandemic were assessed using the non-parametric Wilcoxon Signed Ranks Test. The p-value coefficients of these tests lower than 0.05 were considered to be significant. The Generalized Linear Model (GML) was applied to evaluate the main determinants of energy-saving behavior. As independent factors, the attitudes toward energy consumption, climate change concern, the Covid-19 pandemic concern, the war in Ukraine concern, and the interaction between attitudes toward energy consumption and climate change concern were assessed (Fig. 1 ). The GML analysis reveals which variable influenced the energy saving behavior the most. The probability plots and variance inflation factor (VIF) analyzes were used to check the regression analysis assumptions such as normality of residuals and collinearity respectively. The Beta (B) coefficients, Wald chi-square indicators, and significance p-values were evaluated to reveal what factors determined the energy-saving behavior.Fig. 1 The principal scheme of the study. Fig. 1 4 Results and discussions 4.1 Descriptive analysis In Lithuania, individuals are demonstrating inclinations towards energy conservation within their own households. Considering individual actions, the majority of the respondents declared that they always turn off the light when they do not need it. Furthermore, people often used power-saving bulbs. Meanwhile, fewer individuals are linked to buying and using energy efficient appliances, which are more expensive (Table 2). According to data from the Lithuanian Department of Statistics, electricity consumption increased not only during the Covid-19 pandemic, when people spent more time at home, but also in the post-pandemic period. Two factors could be responsible for this. First, it may be due to the persistence of habits formed during the pandemic, when people used to spend a considerable amount of time at home. Second, it may be affected by the rebound effect, that usually occurs after installation of energy-efficient appliances and light bulbs, individuals tend to consume more energy overall. In terms of the attitude toward energy consumption, half of the respondents agreed that their personal energy consumption affects overall energy consumption that contributes to climate change on a broader scale. These tendencies are not very good, because it reveals that not all people understand their importance in terms of energy consumption, or people see themselves as only small parts of larger societies. Furthermore, respondents do not tend to think that they can influence the government and companies related to energy issues (Table 2). The first steps of liberalization of electricity suppliers show that Lithuanian people are not very active when it comes to choosing an electricity supplier. Only half of the Lithuanian citizens participate in this program, other people do not trust or do not want to change their habits and they trust the authorities and the universal electricity supplier. Thus, this passive participation in the market of electricity suppliers shows these tendencies. After living in a non-democratic system for a long period of time, it takes more than one generation to change the inactivity in society. When analyzing the changes in environmental concern caused by the Covid-19 pandemic, the results presented in Table 2 showed that respondents do not tend to agree that the pandemic strengthens their concern about environmental problems. However, the most individuals agree that the pandemic increased their concern about the environment in general. Meanwhile, considering separate environmental problems people tend to not agree that the Covid-19 pandemic strengthens the concern about these problems. People declared that the pandemic increased their concern about biodiversity the least. For people it is very difficult to evaluate the changes in the importance of separate environmental problems. Moreover, the decrease in biodiversity was the least important environmental problem before the pandemic as well (Liobikiene et al., 2021). These results showed the lack of knowledge about the decline of biodiversity and how important it is to save it. Furthermore, the Covid-19 illness is dangerous not only to humans but also to other species. Thus, the decline of biodiversity has become a very serious problem. Considering the changes in environmental concern caused by the war in Ukraine, respondents do not agree much that it changed their concern. Individuals mostly stated that the war increased their concern about environmental problems in general. Meanwhile, according to the respondents' opinion, the concern about waste problems and biodiversity got strengthened the least due to the war in Ukraine (Table 2). In this case, as in previous results, the decline of biodiversity is not a very important problem. Furthermore, people do not relate war to increasing waste generation. Meanwhile, air and water pollution are the problems that were diffused. A lot of information was provided by the media not only about the course of the war but also about eco-crisis and air and water pollution. However, Lithuania is rather far from Ukraine, and it does not affect Lithuanians’ concern related to these separate environmental problems. 4.2 The differences in concern about current crises, changes in environmental concern and actions related to energy-saving caused by the war in Ukraine and the Covid-19 pandemic After evaluating the concerns about the main crises, we realized that people mostly cared about the war in Ukraine. Although Lithuania is not geographically adjacent to Ukraine and it seems that the ongoing war is quite far, it shares a border with Russia. In the mass media and on social networks this war is defined as Poland and Baltic States’ war. Lithuania has a long history of being occupied by Russians, and in general Russia is perceived as a threat by Lithuanian citizens (Jakniūnaitė, 2015). Therefore, considering the trends of Lithuanian mass and social media, the opinion that if Russia will win the war against Ukraine, Lithuania and the other Baltic States will be next is quite popular. Thus, this concern of the Lithuanian people is very strong, especially because that this war is a new issue. The level of concern about climate change differed from the concern about the war in Ukraine. However, respondents declared that climate change is a very serious problem as well. The provision of a lot of information in media about extreme weather events in other countries increased this concern. Meanwhile, extreme weather events were rather rare in Lithuania during the last year. Thus, globalization and increasing information flows contribute to concern about climate change. The Covid-19 pandemic was the least important to Lithuanian respondents (Fig. 2 ). Only a minor part of the respondents declared that they are very concerned about the pandemic. As per the findings of the study conducted by Matiiuk et al. (2023), the concern about the Covid-19 pandemic decreased significantly when comparing pre- and post-Covid-19 periods. Thus, at the beginning of the pandemic people were very concerned about the illness, but later on the concern decreased due to successful vaccination, and other successfully implemented management tools, and the fact that people now know more about this illness.Fig. 2 The differences among concerns about climate change, the Covid-19 pandemic, and the war in Ukraine. Fig. 2 Comparing how the Covid-19 pandemic and the war in Ukraine caused the changes in concern about environmental problems (Fig. 3 ) we see that the Covid-19 pandemic contributed to the concern more compared to the war in Ukraine, and that this difference is statistically significant (Z = −9.897, p = .000) (Fig. 3). It could be related to the fact that people understand that the environment and health issues are related. Meanwhile, the war in Ukraine is more of a political aspect and is mainly related to security and fear of war. Despite the large amount of information about ecocide, the genocide issue is more important to the society.Fig. 3 The differences in changes in environmental concern caused by the Covid-19 pandemic and the war in Ukraine. Fig. 3 When analyzing the differences in actions related to energy saving caused by the Covid-19 pandemic and the war in Ukraine, we see in Fig. 4 that the pandemic encouraged energy-saving behavior more than the war in Ukraine (‘I saved the environmental resource more’ Z = −6.017, p = .000; ‘I saved electricity more’ - Z = −5.827, p = .000; ‘I purchased energy efficient appliances more’- Z = −8.348, p = .000). In all cases, particularly considering electricity saving, people stated that the pandemic encourages them to do it more. It could be related to the fact that during the pandemic people spent a lot of time at home and their habits remain the same even in the post-Covid-19 period, thus citizens declared that they save electricity and other resources more because of the pandemic. Meanwhile, considering the war in Ukraine, this survey was conducted two months after Russian invasion of Ukraine. The energy crisis was not yet escalating, and the prices had increased negligibly. The purchase level of energy efficiency appliances was primarily driven by the pandemic of Covid-19, rather by war in Ukraine. These results could be related to the fact that during the Covid-19 pandemic citizens could not spend their money on holidays and trips, thus they can spend the saved-up money on energy efficient appliances, which are more expensive. Meanwhile, the war in Ukraine imbalanced the security level, and people began to save money for unforeseen situations.Fig. 4 The differences in actions related to energy-saving behavior caused by the Covid-19 pandemic and the war in Ukraine. Fig. 4 4.3 The assumption and results of regression analysis The correlation analysis was performed to reveal the tolerable level of discriminant validity and multicollinearity assumptions. The correlation matrix presented in Table 3 shows that all independent variables’ values were below 0.5. Thus, the discriminant validity and multicollinearity assumptions were satisfied (Bryman and Cramer, 2001). All determinants included in the model (attitudes towards energy consumption, climate change, the Covid-19 pandemic, and the war in Ukraine concerns) were dissimilar. To confirm the multicollinearity assumption, tolerance and VIF statistic values were assessed and the values greater than 0.6 for tolerance and lower than 2 for VIF (Table 4 ) showed that there is no problem with multicollinearity (Hair et al., 1998).Table 3 Correlation matrix. Table 3 Attitudes towards energy consumption Climate change concern Covid-19 pandemic concern Climate change concern 0.369* Covid-19 pandemic concern 0.228* 0.362* War in Ukraine concern 0.251* 0.475* 0.390* p < .01. Table 4 Test of multicollinearity. Table 4 Tolerance VIF Attitude towards energy consumption 0.850 1.177 Climate change concern 0.683 1.465 Covid-19 pandemic concern 0.801 1.249 War in Ukraine concern 0.717 1.395 The results of GLM analyzes were presented in Table 5 and they showed that considering all crises only the concern about the war in Ukraine significantly and positively influenced energy-saving behavior. Consequently, a stronger level of care regarding the war in Ukraine was associated with a higher likelihood of individuals declaring their engagement in energy-saving practices within their households. There could be several reasons. Firstly, increased insecurity could cause this behavior. Second, the war in Ukraine triggered the energy crisis, which began in the first days of the war and continued until this winter. The increasing prices of energy resources, particularly gases motivate people to save energy at home. The fact that Russian attacks on critical Ukrainian infrastructure frequently result in blackouts of heat and power may also contribute to the widespread support and solidarity with Ukrainians (60 Minutes of Darkness), that could lead to energy-saving behaviors.Table 5 Regression results of energy-saving behavior. Table 5Coefficients Estimate Std. error Wald Chi-Square p. (Intercept) 4.372 0.2024 466.616 <0.001 Attitudes toward energy consumption −0.385 0.0801 23.162 <0.001 Climate change concern −0.026 0.0303 0.720 0.396 Covid-19 pandemic concern −0.029 0.0087 11.570 0.001 War in Ukraine concern 0.044 0.0098 20.344 <0.001 Attitudes toward energy consumption Climate change concern 0.044 0.0109 16.320 <0.001 R2 = 0.118 Dependent variable energy-saving behavior. On the contrary, The Covid-19 pandemic significantly but negatively determined energy-saving behavior (Table 5). However, when examining people's opinions about how the war in Ukraine and the Covid-19 pandemic affected their behavior and environmental concerns, the results portrayed a different situation. It could be due to the gap between peoples' attitude and real behavior. Our results revealed that the more people are concerned about the Covid-19 pandemic, the less they declare that they exhibit energy-saving behavior. Thus, the pandemic influenced the behavior only short-term as was revealed by other authors (Zakeri et al., 2022; Mi et al., 2021; Lucarelli et al., 2020). However, taking into account the long-term situation, the concern about the pandemic did not motivate citizens to save energy. People return to their usual daily routine and the concern about the Covid-19 pandemic even increases this behavior. Thus, energy-saving behavior is not as important as health issues. The climate change concern negatively but insignificantly influenced energy-saving behavior. Thus, this crisis directly had no effect on energy-saving behavior. However, this factor indirectly influenced the demonstration of this behavior as the interaction of concern about climate change and attitude toward energy consumption (Table 5). Therefore, respondents who were concerned about climate change and had positive attitudes about the importance of energy consumption were motivated to save energy in households. Fang et al. (2021), also indicated that concern for energy and environmental problems are important variables narrowing the gap between actions and statements of individuals. The attitudes toward energy consumption significantly but negatively influenced energy-saving behavior (Table 5). Despite that people declared that they understand their contribution to general energy consumption and climate change and that they can influence industry and government, however, they did not link to saving energy in households. This result contradicts Brounen et al. (2013), Chen et al. (2017) and Boomsma et al. (2019) findings. However, referring to Jakučionytė-Skodienė et al. (2020) results, environmental awareness does not give a positive impact on pro-environmental behavior due to various barriers such as lack of knowledge and social pressure. Furthermore, this factor as a catalyst via climate change concern positively and statistically significantly influenced this behavior (Paҫo and Lavrador, 2017; Barber et al., 2009; Flamm, 2009; Polonsky et al., 2012; Liobikiene et al., 2016). 5 Conclusions and policy implication The War in Ukraine, the Covid-19 pandemic and climate change are the main crises in recent years concerning the world. Like never before, the policymakers highlight the importance of energy-saving which is required not only to control the energy crisis caused by the war in Ukraine but also is essential to mitigate climate change. Thus, the aim of this paper is to analyze the concerns about current crises – the war in Ukraine, the Covid-19 pandemic, and climate change's impact on energy-saving behavior and changes in environmental concern. The war in Ukraine was the most serious problem in Lithuania. The level of climate change concern was a little bit lower compared to concern about war but was not very high. Thus, the education and environmental information provision about climate change, its consequences, and the drivers are particularly important. The problem could be that climate change is a long-term issue and that only severe and extreme weather events remind people of it. However, the long-term perspective of climate change often fails to receive the attention it deserves. Thus, the concern about climate change should increase and contribute to behavioral change as well. The Covid-19 pandemic was the least important problem, which shows that by implementing successful management tools, the pandemic can be tackled. According to the respondents’ opinion, the Covid-19 pandemic contributed to the changes in environmental concern and energy-saving actions more than the war in Ukraine. However, the level of contribution was not very high. Despite that the war in Ukraine and the Covid-19 pandemic are not directly and visibly related to environmental issues, however, the impact of these crises on the environment is undoubted. Thus, providing information about the pandemic and war, it is also important to highlight the connections to the environmental impact. The better quality of the environment can contribute to better health. Moreover, by supporting Ukraine financially we can not only help stop the war but also in turn relieve the environment from the extra pressure caused by the war. Referring to the regression results, only the war in Ukraine positively and significantly influenced energy-saving behavior. Considering that this problem was rather new in the Lithuania, it could influenced this strong relationship. However, in order to check the impact of war on energy-saving behavior long-term analyses are needed. The concern about the Covid-19 pandemic negatively and the climate change concern factor indirectly as the interaction of attitudes toward energy consumption affected this behavior. Therefore, the policymakers should continue (even after the energy crises) to highlight the importance of energy-saving behavior, which is also vital to climate change mitigation. The exploitation of current crises could help change pro-environmental behavior. However, taking into account the long-term situation, the concern about the pandemic did not motivate citizens to save energy. Therefore, it is important to educate people that energy-saving behavior not only is important to mitigate climate change, but also can contribute to a better quality of the environment, which is also important for our health. Uncited references COUNCIL RESOLUTION, 1974; COUNCIL RESOLUTION, 1986; Islam et al., 1990 CRediT authorship contribution statement Genovaitė Liobikienė: Conceptualization, Investigation, Writing – original draft, preparation, Writing – review & editing. Yuliia Matiiuk: Formal analysis, Writing – original draft, preparation. Ričardas Krikštolaitis: Data curation, Methodology, Software, Validation, All authors have read and agreed to the published version of the manuscript. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available on request. ==== Refs References Ahmad B. Irfan M. Salem S. Asif M.H. Energy efficiency in the post-COVID-19 era: exploring the determinants of energy-saving intentions and behaviors Front. Energy Res. 9 2022 Ahorsu D.K. Lin C.Y. Imani V. Saffari M. Griffiths M.D. Pakpour A.H. The fear of COVID-19 scale: development and initial validation Int. J. Ment. Health Addiction 2020 1 9 Al Huraimel K. Alhosani M. Kunhabdulla S. Stietiya M.H. SARS-CoV-2 in the environment: modes of transmission, early detection and potential role of pollution Sci. Total Environ. 744 2020 140946 Bae S.Y. Chang P.J. The effect of coronavirus disease-19 (COVID-19) risk perception on behavioural intention towards ‘untact’tourism in South Korea during the first wave of the pandemic (March 2020) Curr. Issues Tourism 24 7 2021 1017 1035 Barber N. Taylor D.C. Strick S. Environmental Knowledge and Attitudes: Influencing the Purchase Decisions of Wine Consumers 2009 Bland J.M. Altman D.G. Statistics notes: cronbach's alpha Bmj 314 7080 1997 572 9055718 Boomsma C. Jones R.V. Pahl S. Fuertes A. Do psychological factors relate to energy saving behaviours in inefficient and damp homes? A study among English social housing residents Energy Res. Social Sci. 47 2019 146 155 Boto-García D. Bucciol A. Climate change: personal responsibility and energy saving Ecol. Econ. 169 2020 106530 Bouarar A.C. Mouloudj K. Mouloudj S. The Impact of Coronavirus on Tourism Sector-An Analytical Study 2020 Bouman T. Verschoor M. Albers C.J. Böhm G. Fisher S.D. Poortinga W. When worry about climate change leads to climate action: how values, worry and personal responsibility relate to various climate actions Global Environ. Change 62 2020 102061 Brounen D. Kok N. Quigley J.M. Energy literacy, awareness, and conservation behavior of residential households Energy Econ. 38 2013 42 50 Bryman A. Cramer D. Quantitative Data Analysis with SPSS Release 10 for Windows 2001 Routledge London 2001 Campbell A.M. Racault M.F. Goult S. Laurenson A. Cholera risk: a machine learning approach applied to essential climate variables Int. J. Environ. Res. Publ. Health 17 24 2020 9378 Carrus G. Tiberio L. Mastandrea S. Chokrai P. Fritsche I. Klöckner C.A. Psychological predictors of energy saving behavior: a meta-analytic approach Front. Psychol. 12 2021 648221 Chen C.F. Xu X. Day J.K. Thermal comfort or money saving? Exploring intentions to conserve energy among low-income households in the United States Energy Res. Social Sci. 26 2017 61 71 Chia J. Harun A. Kassim A.W.M. Martin D. Kepal N. Understanding factors that influence house purchase intention among consumers in Kota Kinabalu: an application of buyer behavior model theory J. Technol. Manag. Bus. 3 2 2016 Chomać-Pierzecka E. Sobczak A. Urbańczyk E. RES market development and public awareness of the economic and environmental dimension of the energy transformation in Poland and Lithuania Energies 15 15 2022 5461 Chowdhury F.R. Nur Z. Hassan N. von Seidlein L. Dunachie S. Pandemics, pathogenicity and changing molecular epidemiology of cholera in the era of global warming Ann. Clin. Microbiol. Antimicrob. 16 1 2017 1 6 28061852 Clemens J.D. Nair G.B. Ahmed T. Qadri F. Holmgren J. Cholera Lancet 390 2017 1539 1549 10.1016/S0140-6736(17)30559-7 28302312 Colwell Rita Global Climate and Infectious Disease: the Cholera Paradigm vol. 274 1997 Science New York, N.Y. 2025 2031 10.1126/science.274.5295.2025 Colwell R.R. Huq A. Environmental reservoir of Vibrio cholerae the causative agent of cholera a Ann. N. Y. Acad. Sci. 740 1 1994 44 54 7840478 COUNCIL RESOLUTION Concerning Community Energy Policy Objectives for 1985 17 December 1974 COUNCIL RESOLUTION Concerning New Community Energy Policy Objectives for 1995 and Convergence of the Policies of the Member States (86/C 241/01) 16 September 1986 COUNCIL RESOLUTION of 9 June Concerning Community Energy Policy Objectives for 1990 and Convergence of the Policies of the Member States 1980 Craig R.K. Cholera and climate change: pursuing public health adaptation strategies in the face of scientific debate Hous. J. Health L. & Pol'y 18 2018 29 Creutzig F. Fuel crisis: slash demand in three sectors to protect economies and climate Nature 606 7914 2022 460 462 (Read later) 35764814 DeVellis R.F. Scale Development: Theory and Applications 2003 SAGE publications Thousand Okas, London, New Delhi Erokhin V. Gao T. Impacts of COVID-19 on trade and economic aspects of food security: evidence from 45 developing countries Int. J. Environ. Res. Publ. Health 17 16 2020 5775 Fang X. Wang L. Sun C. Zheng X. Wei J. Gap between words and actions: empirical study on consistency of residents supporting renewable energy development in China Energy Pol. 148 2021 111945 Field A. Discovering Statistics Using IBM SPSS Statistics 2013 sage Fierros-Gonzalez I. Lopez-Feldman A. Farmers' perception of climate change: a review of the literature for Latin America Front. Environ. Sci. 9 2021 205 Flamm B. The impacts of environmental knowledge and attitudes on vehicle ownership and use Transport. Res. Transport Environ. 14 4 2009 272 279 Gregersen T. Doran R. Böhm G. Poortinga W. Outcome expectancies moderate the association between worry about climate change and personal energy-saving behaviors PLoS One 16 5 2021 e0252105 Gupta S. Rouse B.T. Sarangi P.P. Did climate change influence the emergence, transmission, and expression of the COVID-19 pandemic? Front. Med. 2021 2549 Hair J.F. Anderson R.E. Tatham R.L. Black W.C. Multivariate Data Analysis fifth ed. 1998 Pearson Education Upper Saddle River, NJ Hair J.F. Black W.C. Babin B.J. Anderson R.E. Multivariate Data Analysis Global edition 2010 Huang H. Wang H. Hu Y.J. Li C. Wang X. Optimal plan for energy conservation and CO2 emissions reduction of public buildings considering users' behavior: case of China Energy 261 2022 125037 Huebner G.M. Hamilton I. Chalabi Z. Shipworth D. Oreszczyn T. Explaining domestic energy consumption–the comparative contribution of building factors, socio-demographics, behaviours and attitudes Appl. Energy 159 2015 589 600 Islam M.S. Drasar B.S. Bradley D.J. Long-term persistence of toxigenic Vibrio cholerae 01 in the mucilaginous sheath of a blue-green alga, Anabaena variabilis J. Trop. Med. Hyg. 93 2 1990 133 139 2109096 Jakniūnaitė D. A small state in the asymmetrical bilateral relations: Lithuania in Lithuanian-Russian relations since 2004 Baltic J. Polit. Sci. 4 2015 70 93 Jakučionytė-Skodienė M. Liobikienė G. The changes in climate change concern, responsibility assumption and impact on climate-friendly behaviour in EU from the paris agreement until 2019 Environ. Manag. 69 1 2022 1 16 Jakučionytė-Skodienė M. Dagiliūtė R. Liobikienė G. Do general pro-environmental behaviour, attitude, and knowledge contribute to energy savings and climate change mitigation in the residential sector? Energy 193 2020 116784 Jiang P. Van Fan Y. Klemeš J.J. Impacts of COVID-19 on energy demand and consumption: challenges, lessons and emerging opportunities Appl. Energy 285 2021 116441 Kaiser H.F. An index of factorial simplicity Psychometrika 39 1 1974 31 36 Kaper J.B. Morris J.G. Jr. Levine M.M. Cholera. Clin Microbiol Rev. 8 1 1995 48 86 10.1128/CMR.8.1.48 Erratum in: Clin Microbiol Rev 1995 Apr;8(2):316. PMID: 7704895; PMCID: PMC172849 7704895 Klenert D. Funke F. Mattauch L. O'Callaghan B. Five lessons from COVID-19 for advancing climate change mitigation Environ. Resour. Econ. 76 4 2020 751 778 Koelle K. Pascual M. Yunus M. Pathogen adaptation to seasonal forcing and climate change Proc. Biol. Sci. 272 1566 2005 971 977 16024354 Kuzemko C. Blondeel M. Dupont C. Brisbois M.C. Russia's war on Ukraine, European energy policy responses & implications for sustainable transformations Energy Res. Social Sci. 93 2022 102842 Lacroix K. Gifford R. Psychological barriers to energy conservation behavior: the role of worldviews and climate change risk perception Environ. Behav. 50 7 2018 749 780 Lawler O.K. Allan H.L. Baxter P.W. Castagnino R. Tor M.C. Dann L.E. The COVID-19 pandemic is intricately linked to biodiversity loss and ecosystem health Lancet Planet. Health 5 11 2021 e840 e850 34774124 Lin C.Y. Syrgabayeva D. Mechanism of environmental concern on intention to pay more for renewable energy: application to a developing country Asia Pac. Manag. Rev. 21 3 2016 125 134 Liobikiene G. Niaura A. Mandravickaite J. Vabuolas Ž. Does religiosity influence environmental attitude and behaviour? The case of young Lithuanians Eur. J. Sci. Theol 12 1 2016 81 96 Liobikiene G. Krikštolaitis R. Miceikienė A. Implementation of Sustainable Bioeconomy in European Union Countries: the Decomposition Analysis Approach 2021 Lipp E.K. Huq A. Colwell R.R. Effects of global climate on infectious disease: the cholera model Clin. Microbiol. Rev. 15 4 2002 757 770 12364378 Lobitz B. Beck L. Huq A. Wood B. Fuchs G. Faruque A.S.G. Colwell R. Climate and infectious disease: use of remote sensing for detection of Vibrio cholerae by indirect measurement Proc. Natl. Acad. Sci. USA 97 4 2000 1438 1443 10677480 Lucarelli C. Mazzoli C. Severini S. Applying the theory of planned behavior to examine pro-environmental behavior: the moderating effect of COVID-19 beliefs Sustainability 12 24 2020 10556 Maatoug A. Triki M.B. Fazel H. How do air pollution and meteorological parameters contribute to the spread of COVID-19 in Saudi Arabia? Environ. Sci. Pollut. Control Ser. 28 32 2021 44132 44139 Matiiuk Y. Krikštolaitis R. Liobikienė G. The Covid-19 pandemic in context of climate change perception and resource-saving behavior in the European Union countries J. Clean. Prod. 2023 136433 Mi L. Zhao J. Xu T. Yang H. Lv T. Shang K. How does COVID-19 emergency cognition influence public pro-environmental behavioral intentions? An affective event perspective Resour. Conserv. Recycl. 168 2021 105467 Montilla R. Chowdhury M.A.R. Huq A. Xu B. Colwell R.R. Serogroup conversion of Vibrio cholerae non-O1 to Vibrio cholerae O1: effect of growth state of cells, temperature, and salinity Can. J. Microbiol. 42 1 1996 87 93 8595602 Mustapa S.I. Rasiah R. Jaaffar A.H. Bakar A.A. Kaman Z.K. Implications of COVID-19 pandemic for energy-use and energy saving household electrical appliances consumption behaviour in Malaysia Energy Strategy Rev. 38 2021 100765 Nauges C. Wheeler S.A. The complex relationship between households' climate change concerns and their water and energy mitigation behaviour Ecol. Econ. 141 2017 87 94 Ohler A.M. Billger S.M. Does environmental concern change the tragedy of the commons? Factors affecting energy saving behaviors and electricity usage Ecol. Econ. 107 2014 1 12 Paço A. Lavrador T. Environmental knowledge and attitudes and behaviours towards energy consumption J. Environ. Manag. 197 2017 384 392 Park E. Kwon S.J. What motivations drive sustainable energy-saving behavior?: an examination in South Korea Renew. Sustain. Energy Rev. 79 2017 494 502 Pascual M. Bouma M.J. Dobson A.P. Cholera and climate: revisiting the quantitative evidence Microb. Infect. 4 2 2002 237 245 Polonsky M.J. Vocino A. Grau S.L. Garma R. Ferdous A.S. The impact of general and carbon-related environmental knowledge on attitudes and behaviour of US consumers J. Market. Manag. 28 3–4 2012 238 263 Pop R.A. Dabija D.C. Pelău C. Dinu V. Usage intentions, attitudes, and behaviors towards energy-efficient applications during the COVID-19 pandemic J. Bus. Econ. Manag. 23 3 2022 668 Ross M.L. How the 1973 oil embargo saved the planet Foreign Aff. 15 2013 16 Schimmenti A. Billieux J. Starcevic V. The four horsemen of fear: an integrated model of understanding fear experiences during the COVID-19 pandemic Clin. Neuropsychiatr. 17 2 2020 41 Shrestha B. Tiwari S.R. Bajracharya S.B. Keitsch M.M. Rijal H.B. Review on the importance of gender perspective in household energy-saving behavior and energy transition for sustainability Energies 14 22 2021 7571 Siddique A. Shahzad A. Lawler J. Mahmoud K.A. Lee D.S. Ali N. Unprecedented environmental and energy impacts and challenges of COVID-19 pandemic Environ. Res. 193 2021 110443 Trotta G. Factors affecting energy-saving behaviours and energy efficiency investments in British households Energy Pol. 114 2018 529 539 Trypolska G. Rosner A. The use of solar energy by households and energy cooperatives in post-war Ukraine: lessons learned from Austria Energies 15 20 2022 7610 Ueno T. Capturing changes in residential occupant behavior due to work from home in Japan as a consequence of the COVID-19 pandemic Sustainability 14 4 2022 2180 Vringer K. Aalbers T. Blok K. Household energy requirement and value patterns Energy Pol. 35 1 2007 553 566 Whitmarsh L. Capstick S. Perceptions of climate change Psychology and Climate Change 2018 Academic Press 13 33 Whitmarsh L. O'Neill S. Green identity, green living? The role of pro-environmental self-identity in determining consistency across diverse pro-environmental behaviours J. Environ. Psychol. 30 3 2010 305 314 Zakeri B. Paulavets K. Barreto-Gomez L. Echeverri L.G. Pachauri S. Boza-Kiss B. Pandemic, War, and Global Energy 2022 Zebardast L. Radaei M. The influence of global crises on reshaping pro-environmental behavior, case study: the COVID-19 pandemic Sci. Total Environ. 811 2022 151436
PMC010xxxxxx/PMC10288317.txt	==== Front J Stroke Cerebrovasc Dis J Stroke Cerebrovasc Dis Journal of Stroke and Cerebrovascular Diseases 1052-3057 1532-8511 Elsevier Inc. S1052-3057(23)00256-2 10.1016/j.jstrokecerebrovasdis.2023.107233 107233 Article Hospital Discharge and Readmissions Before and During the COVID-19 Pandemic for California Acute Stroke Inpatients Albert George P. BS 1# McHugh Daryl C. MD, MPH 2# Roberts Debra E. MD, PhD 2 Kelly Adam G. MD 2 Okwechime Remi MD, MPH 2 Holloway Robert G. MD, MPH 2 George Benjamin P. MD, MPH 2⁎ 1 State University of New York, Downstate College of Medicine, Brooklyn, NY 2 University of Rochester Medical Center, Department of Neurology, Rochester, NY ⁎ Correspondence: Benjamin P George, MD, MPH, University of Rochester Medical Center, Department of Neurology, 601 Elmwood Avenue, Box 673, Rochester, NY 14642. 585-275-9238. # Equal contribution 23 6 2023 23 6 2023 10723321 12 2022 19 6 2023 21 6 2023 © 2023 Elsevier Inc. All rights reserved. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Acute stroke therapy and rehabilitation declined during the COVID-19 pandemic. We characterized changes in acute stroke disposition and readmissions during the pandemic. Methods We used the California State Inpatient Database in this retrospective observational study of ischemic and hemorrhagic stroke. We compared discharge disposition across a pre-pandemic period (January 2019 to February 2020) to a pandemic period (March to December 2020) using cumulative incidence functions (CIF), and re-admission rates using chi-squared. Results There were 63,120 and 40,003 stroke hospitalizations in the pre-pandemic and pandemic periods, respectively. Pre-pandemic, the most common disposition was home [46%], followed by skilled nursing facility (SNF) [23%], and acute rehabilitation [13%]. During the pandemic, there were more home discharges [51%, subdistribution hazard ratio 1.17, 95% CI 1.15-1.19], decreased SNF discharges [17%, subdistribution hazard ratio 0.70, 95% CI 0.68-0.72], and acute rehabilitation discharges were unchanged [CIF, p<0.001]. Home discharges increased with increasing age, with an increase of 8.2% for those ≥85 years. SNF discharges decreased in a similar distribution by age. Thirty-day readmission rates were 12.7 per 100 hospitalizations pre-pandemic compared to 11.6 per 100 hospitalizations during the pandemic [p<0.001]. Home discharge readmission rates were unchanged between periods. Readmission rates for discharges to SNF (18.4 vs. 16.7 per 100 hospitalizations, p=0.003) and acute rehabilitation decreased (11.3 vs. 10.1 per 100 hospitalizations, p=0.034). Conclusions During the pandemic a greater proportion of patients were discharged home, with no change in readmission rates. Research is needed to evaluate the impact on quality and financing of post-hospital stroke care. Keywords Acute stroke ischemic stroke subarachnoid hemorrhage intracerebral hemorrhage COVID-19 pandemic hospital discharge readmissions rehabilitation ==== Body pmcIntroduction Stroke is a leading cause of death and disability in the United States, with approximately 795,000 new strokes occurring annually, including acute ischemic stroke (AIS), intracerebral hemorrhage (ICH), and subarachnoid hemorrhage (SAH).1 The Coronavirus Disease 2019 (COVID-19) pandemic has impacted all levels of care for stroke patients, such as delays in initial presentation, reduction in acute therapies, limitations of in-patient resources, delays or lack of initiation of secondary stroke prevention therapy, and limitations in rehabilitation services after hospital discharge. Published data regarding stroke presentations and outcomes during the COVID-19 pandemic have been varied.2, 3, 4, 5, 6 We sought to characterize differences in acute stroke inpatient admissions and disposition during the COVID-19 pandemic compared to the time period preceding the pandemic by performing a retrospective review of acute stroke admissions in the state of California. Our hypothesis was that an increasing frequency of home discharges took place during the pandemic, with no change in readmission rates among these patients. Methods We used the California State Inpatient Database (SID) from the Agency for Healthcare Research and Quality (AHRQ), Healthcare Cost and Utilization Project to perform a retrospective longitudinal observational study examining the impact of the COVID-19 pandemic on acute inpatient care. The University of Rochester Medical Center Research Subjects Review Board approved the study. Data Source The California SID includes a complete enumeration of all-payer administrative claims data on hospital discharges from all non-federal acute care hospitals within the state of California in each year. The California SID is capable of tracking patients longitudinally, across hospitals admissions to understand interhospital transfers and hospital readmissions. The California SID also has the capability of linking American Hospital Association data to further understand hospital characteristics. Other available data include patient demographics (age, sex, race, insurance payor, median household income for ZIP), primary and secondary diagnoses (ICD-10-CM codes), and detailed disposition. Race and ethnicity in California SID are directly reported by HCUP partner organizations and consolidated by HCUP to uniform values which combined race and ethnicity into a single variable. In HCUP methodology, ethnicity took precedence over race. For example, if a patient was identified as Black and Hispanic, they were assigned to Hispanic. Additionally, HCUP consolidates some race categories (i.e., Asian and Native Hawaiian or Pacific Islander). We used up to 36 secondary diagnoses to calculate the Elixhauser comorbidity index calculated from up to 31 categories of disease for each admission with the v2021.1 AHRQ Elixhauser Comorbidity Software.7 Comorbidities relevant to stroke pathology (hypertension, diabetes mellitus, obesity, heart failure, and atrial fibrillation) were separately delineated. Timeframe The California SID contains data for hospital discharges which occurred between January 1st and December 31st of a calendar year. Admission and discharge month and year are known for all patients contained within the database. The specific date of admission and discharge is not known. The timeframe for the study was January 2019 to December 2020. Individuals admitted within the timeframe of the study but discharged after December 2020 are not contained within these data. For the purpose of this study, the beginning of the COVID-19 pandemic was considered to be March 2020. Given the first confirmed case of community transmission in California was identified on February 26, 2020,8 followed by a revision of US Centers for Disease Control and Prevention criteria for testing patients suspected of having COVID-19 infection on February 28, 2020,9 which led to increased use of testing in California. Furthermore, a State of Emergency was declared in California on March 4, 2020.10 Since our dataset is limited to granularity by admission month, we used March 1, 2020 as the beginning of the pandemic period. Assignment to the pre-pandemic and pandemic period were performed using the variable for admission month. Inclusion and Exclusion Criteria We selected acute care hospitalizations for adult (age ≥18 years) patients admitted from January 1, 2019 to December 31, 2020 with acute ischemic stroke, intracerebral hemorrhage, or subarachnoid hemorrhage in California using ICD-10-CM primary diagnosis codes I63, I61, and I60, respectively. Observations with missing data for key variables (e.g., sex, race and ethnicity, and disposition) were excluded. Due to the need to identify single patients, this study required an encrypted patient identifier, which allows for the tracking of patients across multiple hospitalizations while adhering to privacy regulations. Observations with missing encrypted patient identifiers were excluded [Supplement Figure S1]. For stroke hospitalizations, there were no observations with missing data for other covariates analyzed within the study. Outcome Measures We defined the pre-pandemic period as January 1, 2019 to February 29, 2020, and the pandemic period as March 1, 2020 to December 31, 2020. We classified discharge disposition into home, skilled nursing facility (SNF), acute rehabilitation including long-term acute care hospital (LTCH), interhospital transfer, death, and other. We examined disposition during the pre-pandemic and pandemic periods, stratified by age, race, and sex descriptively. We additionally examined readmission rates during pre-pandemic and pandemic periods, stratified by discharge disposition. We defined a readmission as any admission within 30 days of index hospitalization discharge.11 Primary etiology of readmissions was examined using AHRQ Clinical Classifications Software Refined (CCSR) diagnostic categories. We further identified admissions related to COVID-19 using CCSR diagnosis category INF012. Statistical Analysis We used a cumulative incidence function (CIF) for each discharge disposition to investigate disposition differences between the pre-pandemic and pandemic periods, where alternative dispositions were treated as competing risks. Fine-Gray models were used to calculate subdistribution hazard ratios adjusted for age, race and ethnicity, stroke type, Elixhauser comorbidity index, insurance payer, and median income for household ZIP.12, 13, 14. To evaluate differences in cumulative incidence of dispositions between pre-pandemic and pandemic periods, we used the K-sample test statistic.15 We compared categorical and continuous variables before and during the pandemic period using chi-squared and Wilcoxon Rank Sum tests, respectively. We performed multivariable logistic regression analysis predicting readmission stratified by initial hospital disposition (i.e., Home, SNF, Acute Rehab/LTCH, Transfer) controlling for age, race and ethnicity, stroke type, Elixhauser comorbidity index, insurance payer, median income for household ZIP, and length of stay of the index hospitalization in days. Individuals who died during the initial hospitalization were no longer at risk for readmission and therefore excluded from readmission analyses. CIFs and Fine-Gray subdistribution hazard ratios were computed using the cmprsk v2.2-11 library in R version 4.1.2 (Vienna, Austria). All other analyses were performed using Stata version 16.0 (College Station, TX). Statistical significance was two-sided and set a priori at p<0.05. Results There were 121,119 adult acute care stroke hospitalizations admitted across 347 hospitals before and during the pandemic in California. After applying our inclusion/exclusion criteria, there were 103,123 acute stroke admissions in our primary analysis and 11,697 readmissions within 30 days included in our readmissions analysis [Supplement Figure S1]. There were 63,120 stroke hospitalizations occurring in the pre-pandemic period and 40,003 stroke hospitalizations during the pandemic period. There were 51,064 (81%) ischemic stroke, 9,243 (15%) ICH, and 2,813 (4%) SAH during the pre-pandemic and 32,522 (81%) ischemic stroke, 5,839 (15%) ICH, and 1,642 (4%) SAH during the pandemic. Age, sex, race and ethnicity, insurance, Elixhauser Comorbidity Index, and median household income were similar in the pre-pandemic and pandemic periods [Table 1 ]. Notable differences included age ≥85 years (18.5% pre-pandemic vs. 17.6% pandemic, p<0.05), Elixhauser comorbidity index ≥5 (42.4% pre-pandemic vs 44.6% pandemic, p<0.05), and Medicare insurance payer (63.4% pre-pandemic vs. 62.6% pandemic, p<0.05).Table 1 Acute stroke hospitalization characteristics before and during the COVID-19 pandemic Table 1Hospitalization Characteristics Pre-Pandemic* During Pandemic,† Age in yrs, n (%) 18-54 9,130 (14.5) 5,877 (14.7) 55-64 12,165 (19.3) 8,013 (20.0) 65-74 15,227 (24.1) 9,690 (24.2) 75-84 14,898 (23.6) 9,400 (23.5) ≥85 11,700 (18.5) 7,023 (17.6) Race and Ethnicity, n (%) Asian or Pacific Islander 8,206 (13.0) 5,064 (12.7) Black 6,499 (10.3) 4,043 (10.1) Hispanic 14,507 (23.0) 9,257 (23.1) White 30,804 (48.8) 19,415 (48.5) Other‡ 3,104 (4.9) 2,224 (5.6) Sex, n (%) Female 30,341 (48.1) 19,049 (47.6) Male 32,779 (51.9) 20,954 (52.4) Stroke Type, n (%) Acute Ischemic 51,064 (80.9) 32,522 (81.3) ICH 9,243 (14.6) 5,839 (14.6) SAH 2,813 (4.5) 1,642 (4.1) Elixhauser Comorbidity Index§, n (%) 0-1 4,358 (6.9) 2,570 (6.4) 2 8,295 (13.1) 4,843 (12.1) 3 11,838 (18.8) 7,251 (18.1) 4 11,850 (18.8) 7,506 (18.8) ≥5 26,779 (42.4) 17,833 (44.6) Comorbidities, n (%) Hypertension 53,950 (85.5) 34,111 (85.3) Diabetes mellitus 24,883 (39.4) 15,916 (39.8) Obesity 8,800 (13.9) 6,035 (15.1) Heart failure 11,569 (18.3) 7,522 (18.8) Atrial fibrillation 16,124 (25.5) 10,206 (25.5) Insurance Payer, n (%) Medicare 40,032 (63.4) 25,050 (62.6) Private 11,123 (17.6) 7,071 (17.7) Medicaid 9,762 (15.5) 6,473 (16.2) Other\|\| 2,195 (3.5) 1,405 (3.5) Household Income#, n (%) 1st Quartile 16,814 (27.4) 10,333 (25.8) 2nd Quartile 15,384 (25.0) 10,056 (25.1) 3rd Quartile 15,675 (25.5) 9,652 (24.1) 4th Quartile 13,571 (22.1) 8,573 (21.4) Length of Stay, mean (SD) 5.44 (7.54) 5.22 (6.89) Abbreviations: ICH – intracerebral hemorrhage; SAH – subarachnoid hemorrhage. ⁎ The pre-pandemic period is from January 1, 2019 to February 29, 2020 and the pandemic period is March 1, 2020 to December 31, 2020. † P-values were excluded from the table as all comparisons using the chi-squared test were statistically significant with p<0.001. ‡ “Other” race and ethnicity includes individuals not categorized by the California State Inpatient Database, which include those identified as multiple races, other race not classified, or unknown. Individuals identified as Native American and Alaskan Native are included within this group for confidentiality reasons due to fewer than 10 records within the sample. § Elixhauser comorbidity index is a measure of comorbidity for use with large administrative datasets with higher numbers representing the presence of more comorbidities, accounting for up to 31 categories of disease. \|\| “Other” insurance payer includes self-pay, no charge, Worker's Compensation, Civilian Health and Medical Program of the Uniformed Services or Veterans Affairs, Title V, and other government programs. # Household income quartiles were assigned based on the median income of the patient's ZIP Code, where the first quartile is the lowest income and fourth quartile is the highest income. Disposition In the pre-pandemic period, most admissions resulted in discharge to home (46%), followed by SNF (23%), and acute rehabilitation or LTCH (13%) [Figure 1 A]. During the pandemic, the distribution of dispositions shifted with more discharges home (51%) and less to SNF (17%), while discharges to acute rehabilitation or LTCH remained the same.Figure 1 Acute stroke discharge disposition before and during the COVID-19 pandemic. Abbreviations: LTCH – long-term acute care hospital Percentages of discharges to each disposition among discharges within the period specified for (A) all stroke patients and (B) by age group. The pre-pandemic period is from January 1, 2019 to February 29, 2020 and the pandemic period is March 1, 2020 to December 31, 2020. Figure 1 The increase in discharges to home, and decrease in discharge to SNF, were driven primarily by older age groups [Table 2 ]. For example, 59% of 18–54 year old admissions resulted in discharge home prior to the pandemic and 61% were discharged home during the pandemic, compared to 31% of ≥85-year-old admissions resulting in discharge home before the pandemic and 39% during the pandemic [Figure 1 B]. There were no differences in sex and race and ethnicity for discharges before and during the pandemic [Table 2].Table 2 Changes in acute stroke discharge disposition during the COVID-19 pandemic by age, race, and sex Table 2 Home SNF Transfer Acute Rehab/LTCH Died Other Percent Change Disposition Within Group,†Pandemic vs. Pre-Pandemic‡ Overall +5.1 -5.8 -0.6 +0.02NS -0.2NS +1.5 Age, yrs 18-54 +1.7 -2.5 -0.8NS -0.4NS +0.7NS +1.2 55-64 +2.6 -2.9 -0.5NS +0.2NS +0.1NS +0.6 65-74 +4.3 -4.5 -0.7 +0.1NS -0.2NS +1.0 75-84 +7.0 -7.9 -0.8 -0.1NS -0.3NS +2.0 ≥85 +8.2 -10.1 -0.3NS -0.1NS -1.1 +3.4 Race and Ethnicity Asian or Pacific Islander +5.3 -7.6 -0.4NS -0.2NS +0.9NS +2.0 Black +5.3 -5.5 -0.7NS -0.9NS +0.4NS +1.4 Hispanic +6.3 -6.5 -0.8 -0.1NS -0.2NS +1.3 White +4.7 -5.3 -0.4 +0.5NS -0.7 +1.4 Other§ +3.6 -3.5 -1.0NS -0.6NS -0.3NS +1.8 Sex Female +6.5 -7.0 -0.6 -0.6 -0.2NS +1.9 Male +3.9 -4.7 -0.6 +0.5NS -0.3NS +1.1 Abbreviations: SNF – skilled nursing facility; LTCH – long-term acute care hospital ns = not significant ⁎ “Other” disposition category includes those discharged to hospice care or those who left against medical advice. † Each cell represents the absolute percent change in the share of discharge disposition for each detailed disposition option (e.g., home, SNF, transfer). For example, among patients 18-54 years of age there were 58.8% discharged home in the pre-pandemic period compared to 60.5% discharged home in the pandemic period, or +1.7% absolute percent change. All absolute percent changes are statistically significant p<0.05 unless indicated. ‡ The pre-pandemic period is from January 1, 2019 to February 29, 2020 and the pandemic period is March 1, 2020 to December 31, 2020. § “Other” race and ethnicity includes individuals not categorized by the California State Inpatient Database, which include those identified as multiple races, other race not classified, or unknown. Individuals identified as Native American and Alaskan Native are included within this group for confidentiality reasons due to fewer than 10 records within the sample. To account for competing risks in discharge disposition, we demonstrate a cumulative incidence function and calculate subdistribution hazard ratios [SHR] adjusted for covariates showing an increase in discharge to home (SHR 1.17, 95% CI 1.15-1.19) and decline in discharge to SNF (SHR 0.70, 95% CI 0.68-0.72) and interhospital transfer (SHR 0.88, 95% CI 0.83-0.93) (CIF, p<0.001 for all) [Figure 2 and Supplement Table S1].Figure 2 Cumulative incidence for acute stroke disposition before and during the COVID-19 pandemic. Abbreviations: AR/LTCH – Acute rehabilitation/long-term acute care hospital; SNF – skilled nursing facility Cumulative incidence functions (CIF) representing incidence of disposition as a function of length of stay, accounting for each possible disposition as a competing risk. Length of stay was truncated at 30 days for visualization purposes only. The pre-pandemic period is from January 1, 2019 to February 29, 2020 and the pandemic period is March 1, 2020 to December 31, 2020. CIF is displayed for (A) all stroke, and by individual disposition: (B) home, (C) SNF, (D) transfer, (E) AR/LTCH, and (F) Died Figure 2 Readmissions The overall 30-day hospital readmission rate for acute stroke before the pandemic was 12.7 per 100 hospitalizations compared to 11.6 per 100 hospitalizations during the pandemic. The primary diagnosis for 30-day hospital readmissions following acute stroke before and during the pandemic are shown in Supplement Figure S2. The most common reasons for readmission following acute stroke included ischemic stroke, sepsis, hemorrhagic stroke, and renal disorders. Sepsis was the only primary diagnosis for readmission which increased in the pandemic period compared to pre-pandemic (p=0.01). Among all 30-day hospital readmission, ∼2% were readmitted with a primary diagnosis of COVID-19. COVID-19 readmission was most common among discharges to SNF. The readmission rate for those discharged to home (11.0 vs. 10.9 per 100 hospitalizations, adjusted Odds Ratio [aOR] 0.95, 95% CI 0.90-1.01) and interhospital transfer (11.8 vs. 11.7 per 100 hospitalizations, aOR 0.99, 95% CI 0.82-1.18) were unchanged [Figure 3 ]. However, the readmission rate for those discharged to SNF (18.4 vs. 16.7 per 100 hospitalizations, aOR 0.88, 95% CI 0.81-0.95) and acute rehabilitation or LTCH (11.3 vs. 10.1 per 100 hospitalizations, aOR 0.87, 95% CI 0.78-0.97) decreased during the pandemic compared to before the pandemic [Figure 3]. In an unadjusted subgroup analysis, 30-day readmission rates by age, race and ethnicity, and sex were unchanged with few exceptions [Table 3 ].Figure 3 Multivariate logistic regression predicting odds of readmission within 30 days of index hospitalization discharge. Abbreviations: SNF – skilled nursing facility; LTCH – long-term acute care hospital; ICH – intracerebral hemorrhage; SAH – subarachnoid hemorrhage Home, SNF, Transfer, Acute Rehab/LTCH indicate the discharge disposition for the index acute stroke hospitalization. For example, acute stroke patients discharged to SNF during the pandemic period compared to pre-pandemic had lower odds of being readmitted within 30 days following discharge, while those discharged to home during the pandemic compared to pre-pandemic had no difference in odds of 30-day readmission. The pre-pandemic period is from January 1, 2019 to February 29, 2020 and the pandemic period is March 1, 2020 to December 31, 2020. Figure 3 Table 3 Changes in acute stroke readmission rates during the COVID-19 pandemic by age, race, and sex Table 3 Home SNF Transfer Acute Rehab/LTCH Other* Pre-Pandemic Pandemic† Pre-Pandemic Pandemic† Pre-Pandemic Pandemic† Pre-Pandemic Pandemic† Pre-Pandemic Pandemic† Readmission Rate Per 100 Hospitalizations‡ Age (years) 18-54 10.0 9.4 18.3 15.5 11.9 12.1 9.3 9.3 19.6 18.7 55-64 10.2 10.2 17.8 18.7 12.5 14.0 10.2 8.1 16.0 16.1 65-74 10.8 10.8 19.4 17.1S 11.7 12.9 10.2 10.8 9.6 7.0 75-84 11.9 12.0 20.2 18.0S 11.0 9.3 13.4 11.9 4.4 4.3 ≥85 12.8 12.3 16.0 13.9 12.3 8.5 13.4 10.0S 1.7 1.2 Race and Ethnicity Asian or Pacific Islander 9.6 10.2 18.8 16.5 12.4 9.6 9.9 8.4 4.1 3.4 Black 12.9 12.6 21.2 19.5 14.6 10.9 10.6 10.9 15.3 12.4 Hispanic 10.7 10.5 19.1 17.3 12.7 14.3 12.1 9.9 7.1 9.0 White 10.9 10.8 17.4 15.8 10.7 11.2 11.8 10.6 6.0 4.8 Other§ 12.1 12.0 18.1 17.3 9.6 12.5 8.9 9.8 8.4 8.1 Sex Female 11.2 11.2 16.6 15.2 11.7 10.5 11.0 10.1 4.7 5.1 Male 10.8 10.6 20.5 18.2S 12.0 12.6 11.5 10.1 9.6 7.9 Abbreviations: SNF – skilled nursing facility; LTCH – long-term acute care hospital; s = significant to p<0.05 ⁎ “Other” disposition category includes those discharged to hospice care or those who left against medical advice. † The pre-pandemic period is from January 1, 2019 to February 29, 2020 and the pandemic period is March 1, 2020 to December 31, 2020. ‡ Each cell represents the 30-day readmission rate for index acute stroke hospitalizations occurring in the pre-pandemic or pandemic period. Significant change from pre-pandemic to pandemic period is marked by s for p<0.05. For example, male readmission rate following discharge to SNF was 18.2 per 100 hospitalizations during the pandemic, which was less than 20.5 per 100 hospitalizations in the pre-pandemic period. § “Other” race and ethnicity includes individuals not categorized by the California State Inpatient Database, which include those identified as multiple races, other race not classified, or unknown. Individuals identified as Native American and Alaskan Native are included within this group for confidentiality reasons due to fewer than 10 records within the sample. Discussion We performed a retrospective observational analysis examining the impact of the COVID-19 pandemic on disposition following an inpatient admission for acute stroke. We found an overall decrease in acute stroke admissions during the pandemic, and an increase in home discharges and corresponding decrease in discharges to SNF. Thirty-day readmission rates were not different for patients discharged home, while there was a decrease in 30-day readmission rates for patients discharged to SNF and acute rehabilitation or LTCH. The incidence of stroke hospitalizations in California during the pandemic decreased compared to the pre-pandemic period, with 4,509 stroke admissions per month pre-pandemic and 4,000 stroke admissions per month during the pandemic. This decrease in stroke presentations is consistent with previous publications.3 , 4 , 6 , 16 , 17 Furthermore, the pandemic may be marked by more medically complex patients as demonstrated by greater Elixhauser comorbidity index scores. The reasons for the decrease in stroke presentations remains unclear, but potential explanations include a decrease in the number of mild stroke presentations (e.g., patients remaining at home in an attempt to avoid COVID-19 exposure in the hospital), delays in presentation of major strokes leading to death outside the hospital, and missed stroke diagnoses. Mild stroke presentations have been reported to be lower during the pandemic lockdown period.4 , 16 During the pandemic period, there was a significant increase in the frequency of patients discharged home, and corresponding decrease in the frequency of patients discharged to SNF. This change in discharge practice patterns was driven by older adults; however, shifts toward home discharge were consistent across sex, race and ethnicity subgroups. This increase in the rate of discharge to home is particularly notable given the higher complexity of stroke patients admitted in the pandemic period, as measured by Elixhauser index. Limited data is published regarding changes in stroke discharge disposition during the pandemic. One small study found no significant difference in disposition.3 Two recent studies found patients admitted with COVID-19 and acute ischemic stroke were more likely to have discharge destination other than home.18 , 19 However, the overall rate of acute ischemic stroke in COVID-19 patients is relatively low; approximately 1%.19 These studies did not specifically examine discharge destinations for the stroke population during the pandemic period. Potential explanations for a shift toward home discharges following an inpatient admission for stroke include1 decreased availability of SNF beds during the pandemic,2 patient preferences for home discharge to avoid exposure to COVID-19 in SNF, or3 avoid strict visitor limitations at SNF during the pandemic period. Despite the increased number of home discharges, 30-day readmissions for this population were unchanged compared to the pre-pandemic period. Furthermore, the rate of 30-day readmissions from SNF and acute rehabilitation or LTCH decreased. The primary readmission diagnoses were similar pre-pandemic and during the pandemic, with few exceptions. Readmissions for sepsis increased during the pandemic period which may be another marker of increased illness severity during the pandemic period as well as potential overlap with COVID-19 illness. Our findings suggests that more patients can be discharged directly home than previously practiced, potentially without compromising quality, as measured by avoidance of 30-day readmission. This could possibly result in decreased direct healthcare costs, increased patient satisfaction, and provide support for increased insurance coverage of home and community-based services.20 However, long-term functional outcomes in this patient population are unknown. In addition, it is possible that more patients were discharged home due to increased availability of family support during the pandemic, with family members more often acting as caregivers with shelter-in-place orders and the greater ability to work from home. Such availability of family support may wane over time as pandemic restrictions decline and fewer people work from home. Further research may focus on design and outcomes for rehabilitation at home programs for patients post-stroke. Limitations Our study utilized the California SID, an administrative database. Diagnoses are based upon ICD-10 codes, and some coding errors may exist within the data. However, the sensitivity and positive predictive value of utilizing ICD-10 codes for identifying acute stroke are 99% and 93% (ischemic strokes) and 99% and 89% (hemorrhagic strokes), respectively.21 , 22 Patients included in our cohort were selected based upon admission date. Individuals admitted within the timeframe of the study but discharged after December 2020 are not contained within these data. Patients admitted during the end of December 2020 or those with more prolonged hospitalizations that were discharged after December 2020 were not included in the analysis, leading to a net loss of patients admitted during the pandemic period, and potentially excluding some sicker patients with longer length of stay. Specific metrics such as stroke severity, radiographic and laboratory findings, and baseline and discharge functional status are not measured within the State Inpatient Database. Such data may provide more insight into specific findings; however, databases with these detailed clinical characteristics are limited in size. Data is limited to California inpatient admissions, which may limit generalizability to the population as a whole, especially since COVID-19 case rates varied by state. Data only capture inpatient death and therefore does not allow for the analysis of death following discharge to be considered as a competing risk with readmission. Furthermore, our analysis of 30-day readmission rates during the pandemic compared to a pre-pandemic period may be confounded by a generally higher threshold to send patients to the hospital at the height of the COVID-19 pandemic, which may skew our results in the direction of declining readmissions. Finally, we excluded 13% of acute stroke hospitalizations in our sample due to missing encrypted patient identifiers, which were needed to track patients over hospitalizations. Observations with missing patient identifiers were younger (mean age 63 vs. 70 years, p<0.001) and more often identified as Hispanic (42% vs. 23%, p<0.001) compared to those with known patient identifiers, raising the potential for bias in our results. Despite these limitations, our study has a large number of observations in a diverse patient population that included patients with a broad mix of insurance payers, who sought care from a wide spectrum of hospital settings (e.g., large academic centers, urban community, or rural community hospitals). This allowed us to assess for potential differences in disposition and readmission for various patient characteristics across the pre-pandemic and pandemic periods. Conclusions We confirmed our hypothesis that more acute stroke patients were discharged to home during the pandemic, as opposed to facility discharge, compared to the pre-pandemic period of our study. Furthermore, this shift toward home disposition did not result in increased hospital readmissions. Disruption in normal patterns of stroke care utilization by the COVID-19 pandemic may have forced health systems to address prior inefficiencies and costly care.23 , 24 Further study is needed to identify changes in care patterns that arose during the pandemic, which may have resulted in unintended improvements without sacrificing the quality of care provided to stroke patients, and to better align payments policies with more cost-effective, patient-centered care. Dr. George takes responsibility of the accuracy of the analysis, integrity of the study, and accountability for all aspects of the work. Author Contributions Study Concept or Design: McHugh DC, George BP Acquisition of Data: Albert GP, George BP Statistical Analysis: Albert GP, McHugh DC, George BP Interpretation of Data: Albert GP, McHugh DC, George BP, Roberts DE, Kelly AG, R Okwechime, Holloway RG Drafting of the Manuscript: Albert GP, McHugh DC Critical Revision for Important Intellectual Content: Albert GP, McHugh DC, George BP, Roberts DE, Kelly AG, R Okwechime, Holloway RG Study Supervision: George BP Sources of Funding This project was supported by the American Academy of Neurology Medical Student Research Scholarship. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the funding organization. Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: George P. Albert reports financial support was provided by American Academy of Neurology. Appendix Supplementary materials Image, application 1 Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jstrokecerebrovasdis.2023.107233. ==== Refs References 1 Tsao CW Aday AW Almarzooq ZI Heart Disease and Stroke Statistics-2022 Update: A Report From the American Heart Association Circulation 145 2022 e153 e639 35078371 2 Aboul Nour H Affan M Mohamed G Impact of the COVID-19 Pandemic on Acute Stroke Care, Time Metrics, Outcomes, and Racial Disparities in a Southeast Michigan Health System J Stroke Cerebrovasc Dis 30 2021 105746 3 Jasne AS Chojecka P Maran I Stroke Code Presentations, Interventions, and Outcomes Before and During the COVID-19 Pandemic Stroke 51 2020 2664 2673 32755347 4 Nguyen-Huynh MN Tang XN Vinson DR Acute Stroke Presentation, Care, and Outcomes in Community Hospitals in Northern California During the COVID-19 Pandemic Stroke 51 2020 2918 2924 32762619 5 Thau L Siegal T Heslin ME Decline in Rehab Transfers Among Rehab-Eligible Stroke Patients During the COVID-19 Pandemic J Stroke Cerebrovasc Dis 30 2021 105857 6 White TG Martinez G Wang J Impact of the COVID-19 Pandemic on Acute Ischemic Stroke Presentation, Treatment, and Outcomes Stroke Res Treat 2021 2021 8653396 7 Elixhauser Comorbidity Software Refined for ICD-10-CM Healthcare Cost and Utilization Project (HCUP) October 2021 Agency for Healthcare Research and Quality Rockville, MD 8 California Department of Public Health. CDC Confirms Possible First Instance of COVID-19 Community Transmission in California. Februrary 26, 2020. https://www.cdph.ca.gov/Programs/OPA/Pages/NR20-006.aspx. Accessed April 20, 2023. 9 Centers for Disease Control and Prevention. Transcript for the CDC Telebriefing Update on COVID-19. February 28, 2020. https://www.cdc.gov/media/releases/2020/t0228-COVID-19-update.html. Accessed April 20, 2023. 10 California Office of Governor Newsom. Governor Newsom Declares State of Emergency to Help State Prepare for Broader Spread of COVID-19. March 04, 2020. https://www.gov.ca.gov/2020/03/04/governor-newsom-declares-state-of-emergency-to-help-state-prepare-for-broader-spread-of-covid-19/. Accessed April 20, 2023. 11 Vahidy FS Donnelly JP McCullough LD Nationwide Estimates of 30-Day Readmission in Patients With Ischemic Stroke Stroke 48 2017 1386 1388 28389607 12 Austin PC Fine JP. Practical recommendations for reporting Fine-Gray model analyses for competing risk data Stat Med 36 2017 4391 4400 28913837 13 Austin PC Lee DS Fine JP. Introduction to the Analysis of Survival Data in the Presence of Competing Risks Circulation 133 2016 601 609 26858290 14 Fine JP Gray RJ. A Proportional Hazards Model for the Subdistribution of a Competing Risk Journal of the American Statistical Association 94 1999 496 509 15 Gray RJ. A Class of K-Sample Tests For Comparing The Cumulative Incidence of a Competing Risk Annals of Statistics 16 1988 1141 1154 16 Kristoffersen ES Jahr SH Faiz KW Stroke admission rates before, during and after the first phase of the COVID-19 pandemic Neurol Sci 42 2021 791 798 33428057 17 Yoo A Guterman EL Hwang DY Impact of the COVID-19 Pandemic on Inpatient Utilization for Acute Neurologic Disease The Neurohospitalist 2022 in revision 18 Qureshi AI Baskett WI Huang W Acute Ischemic Stroke and COVID-19: An Analysis of 27 676 Patients Stroke 52 2021 905 912 33535779 19 Rubens M Saxena A Ramamoorthy V Hospital Outcomes among COVID-19 Hospitalizations with Acute Ischemic Stroke: Cross-Sectional Study Results from California State Inpatient Database Brain Sci 12 2022 20 Johnson M Berwick DM. Medicare 2.0-A Vision for the Future of America's Health Insurance Plan JAMA 2022 21 Hsieh MT Hsieh CY Tsai TT Performance of ICD-10-CM Diagnosis Codes for Identifying Acute Ischemic Stroke in a National Health Insurance Claims Database Clin Epidemiol 12 2020 1007 1013 33061648 22 Hsieh MT Huang KC Hsieh CY Validation of ICD-10-CM Diagnosis Codes for Identification of Patients with Acute Hemorrhagic Stroke in a National Health Insurance Claims Database Clin Epidemiol 13 2021 43 51 33469381 23 George BP Kelly AG. Rethinking Regional Neurologic Care in the Coronavirus Disease 2019 Era JAMA Neurol 77 2020 1061 1062 32597937 24 O'Neal MA Zecavati N Yu M Effects of Fragmentation and the Case for Greater Cohesion in Neurologic Care Delivery Neurology 2021
PMC010xxxxxx/PMC10288318.txt	==== Front J Transp Geogr J Transp Geogr Journal of Transport Geography 0966-6923 1873-1236 Published by Elsevier Ltd. S0966-6923(23)00112-6 10.1016/j.jtrangeo.2023.103640 103640 Article Travel before, during and after the COVID-19 pandemic: Exploring factors in essential travel using empirical data Yang Chao a Wan Zhiyang a Yuan Quan a⁎ Zhou Yang a Sun Maopeng b a Urban Mobility Institute, the Key Laboratory of Road and Traffic Engineering, Ministry of Education at Tongji University, College of Transportation Engineering, Tongji University, 4800 Caoan Road, Shanghai 201804, PR China b College of Transportation Engineering, Chang'an University, Xi'an 710064, China ⁎ Corresponding author at: Urban Mobility Institute, Department of Transportation Engineering, Tongji University, Shanghai 201804, China. 23 6 2023 23 6 2023 10364023 11 2022 18 5 2023 20 6 2023 © 2023 Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The COVID-19 pandemic has a significant impact on daily life, leading to quarantines and essential travel restrictions worldwide in an effort to curb the virus's spread. Despite the potential importance of essential travel, research on changes in travel patterns during the pandemic has been limited, and the concept of essential travel has not been fully explored. This paper aims to address this gap by using GPS data from taxis in Xi'an City between January and April 2020 to investigate differences in travel patterns across three periods pre, during, and post the pandemic. Spatial statistical models are used to examine the major supply and demand-oriented factors that affect spatial travel patterns in different periods, and essential and nonessential socioeconomic resources are defined based on types of services. Results indicate that the spatial distribution of travel demand was highly correlated with the location of socioeconomic resources and opportunities, regardless of the period. During the “Emergency Response” period, essential travel was found to be highly associated with facilities and businesses providing essential resources and opportunities, such as essential food provider, general hospital and daily grocery supplies. The findings suggest that local authorities may better identify essential travel destinations by referencing the empirical results, strengthening public transit connections to these locations, and ultimately promoting traffic fairness in the post-pandemic era. Keywords Travel behavior Essential travel COVID-19 Spatial model GPS data ==== Body pmc1 Introduction The COVID-19 pandemic, caused by the infectious coronavirus, has had an unprecedented impact on the world since early 2020 (Zhang et al., 2020). This impact is significant and varied, affecting social operation, environment, economic growth, energy consumption, and other aspects (Rahman et al., 2021). One of the most noticeable manifestations is the changes in travel patterns. Due to restrictions on mobility, many industries, businesses, and households have been greatly affected. Global road transportation and aviation activities have dropped by an average of 50% during COVID-19 compared to 2019 (IEA, 2020). A report by Marchant on the effects of COVID-19 lockdowns on transport showed that the top 10 cities around the world saw reductions in travel by over 80% (Marchant, 2020). In fact, many of the reductions were due to non-essential travel, which governments encouraged people to avoid. The COVID-19 pandemic has highlighted the importance of understanding essential travel and its impacts on travel behaviors, not only during the pandemic but also in the post-pandemic era (Chen et al., 2021). Defining essential travel is crucial for policymakers to reconsider the most basic needs of citizens, especially during extreme circumstances like the COVID-19 pandemic. Although many countries and regions have advocated for only essential travel, the specific definition varies due to regional and cultural background differences. This concept has in general been defined as travel that is fundamental to daily life (Chen et al., 2021). In this paper, we created an operational definition of essential travel as travel that is still going on during the period of “Emergency Response to Public Health Emergencies” (hereinafter referred to as “Emergency Response”) when non-essential activities were largely constrained by the government to control the spreading of the virus. Under such definition, essential travel includes but not limited to work, buying necessities and food, and seeking healthcare, etc. This study aims to answer several questions regarding essential travel. How did daily travel patterns differ pre, during, and post the “Emergency Response”? What factors influenced people's travel behavior during different periods? How did these factors vary across different regions? The answers to these questions can provide valuable insights into understanding essential travel. He et al. (2021) pointed out that the concept of essential travel is useful for identifying the basic travel needs of different groups, especially socially disadvantaged groups. However, existing research has failed to provide sufficient empirical evidence to justify the definition of essential travel, which is difficult to observe directly during normal times. In this study, we aim to examine travel patterns during different periods of the COVID-19 pandemic using taxi travel data, and discuss how the results can inform decision-making for public transit services in the post-pandemic world. We employ spatial models to analyze the impact of essential and non-essential points of interest (POIs) on travel during the Pre, During, and Post periods. Our findings suggest that the influence of essential POIs on travel does not show a significant decrease in the During period and even slightly increases, with the coefficient of essential shopping showing a positive significance in this period only. On the other hand, the impact of non-essential POIs on travel decreases in the During period, with non-essential catering and non-essential education becoming non-significant. Furthermore, the spatial spillover effects of essential catering, essential healthcare, and y-lag indicate that in the During period, people may travel more directly to their destinations, and the influence of the surrounding areas on their destinations significantly decrease. A more comprehensive comprehension of essential travel can enable planners to estimate the demand for essential travel more precisely and make recommendations for enhancing the public transit system to provide people in different locations with more equitable access to essential goods and services. This level of equity could have a significant impact on daily life and various stages of personal development, as it is necessary to have access to various socioeconomic resources and opportunities, such as employment, healthcare, and social connections. The article will be organized as follows. Section 2 displays a careful literature review on the impact of COVID-19 on travel and the definition of essential travel in existing studies. Details about the research design, including methodologies and data will be presented in Section 3. Results are shown and discussed in Section 4. We will summarize the major findings and policy implications in the final section of the article. 2 Literature review 2.1 Individual travel during quarantine The outbreak of the COVID-19 pandemic has a significant impact on travel activities and behaviors. Past studies on infectious diseases have shown that travel restrictions play a crucial role in controlling the initial spread of such diseases (Aldila et al., 2020; Beck and Hensher, 2020; Chinazzi et al., 2020; de Bruin et al., 2020; De Haas et al., 2020). Sharkey and Wood (2020) estimated a difference-in-difference model and found that a 1% decrease in non-essential travel would result in a 6.4% reduction in new cases on average. Villas-Boas et al. (2020) reported similar conclusions. As a result, many countries and regions implemented corresponding travel restrictions and advocated only “Essential Travel”. These restrictive policies resulted in a year-on-year decline of more than 50% in global road transportation and air travel during the same period in 2020 (IEA, 2020). Despite the negative impacts of the pandemic, researchers have also discovered some positive effects of the COVID-19 pandemic on individual travel. For instance, the pandemic has resulted in shorter travel times for public transit and a reduction in accidents. The congestion index of cities worldwide has decreased to varying degrees as well (Clarke, 2020; Hurley, 2020). The current body of research has predominantly focused on the changes in travel preferences, patterns, and means brought about by the COVID-19 pandemic. On one hand, due to various travel restrictions and the potential risk of virus transmission during travel, people's willingness to travel has decreased significantly, particularly for non-essential travel. Cui et al. (2021) reported a severe decline in the output of all transport sectors in China. Beck and Hensher (2020) observed the largest drop in outdoor recreational activities in Australia. On the other hand, in order to maintain social distancing during travel, some individuals who previously relied on public transit have shifted to using private cars (Campisi et al., 2020; Labonté-LeMoyne et al., 2020; Shakibaei et al., 2021; Zhang et al., 2020) or non-motorized modes of transportation (Bergantino et al., 2021; Teixeira and Lopes, 2020; Zhang and Fricker, 2021). Moreover, another study found that pre-existing disparities in travel behavior across socioeconomic status (SES) clusters were exacerbated during the pandemic lockdown (Kara et al., 2021). Meanwhile, recent studies have indicated that individuals are willing to travel more if the safety and health risks of travel can be mitigated and public transit can be restored. However, the recovery has fallen far short of expectations (Beck and Hensher, 2020; Przybylowski et al., 2021). The COVID-19 pandemic and the resulting quarantine measures have presented both challenges and opportunities in the field of transport planning. On the one hand, the restricted mobility caused by quarantine orders has made it difficult for people in certain areas to access essential supplies and resources. For instance, residents in suburban areas have had to rely more on nonmotorized modes of transportation to obtain what they need than their urban counterparts. Policymakers must take appropriate actions to address potential disparities in individuals' travel choices during the pandemic and ensure that everyone has equal access to necessary goods and services. On the other hand, the quarantine measures have provided a valuable opportunity to study the fundamentals of transport services under extreme circumstances. Although travel is a means of accessing socioeconomic resources, some types of travel are more essential to daily life than others. By restricting non-essential travel during the Emergency Response period, the contrasts between essential and non-essential travel have become more apparent and can be studied more closely. While previous research provided significant contributions to understanding how the pandemic has impacted travel behavior, there remains a need for more detailed and nuanced analysis to fully grasp the implications of these changes. In particular, there has been a lack of research that explores the interconnectivity between travel restrictions, quarantine orders, and the overall decrease in travel demand in relation to how it affects access to essential goods and services. Understanding this connection could prove useful for urban transport planners as they strive to address potential disparities in accessibility among individuals residing in different areas of cities. 2.2 A closer look at essential travel During the Stay-at-Home policy period, there was a significant decrease in travel demand and activities. People were advised to travel only for essential purposes. However, the definition of essential travel varied across policies, orders, and guidelines, as well as across cultures and countries. In its guidelines, the World Health Organization (WHO) recommended that individuals only undertake essential travel and noted that different countries may have varying definitions of essential travel. The WHO's definition of essential travel included travel in emergency situations and humanitarian activities, such as the travel of vital personnel, returning to one's country of origin, and obtaining essential supplies such as food, medicine, and fuel (WHO, 2020). Table 1 presents the definitions of essential travel adopted by various institutions worldwide during the COVID-19 pandemic (C.D.C.P, 2022; C.D.P.H, 2021; E.U, 2020; Government, 2020a; Health, 2021; Immigration, 2021; University, 2021). The definitions of essential travel varied across institutions and the pandemic's different periods. Governments at higher levels were less likely to define essential travel for specific purposes and impose detailed travel restrictions.Table 1 A summary of travel policies and definitions of essential travel by different institutions across the world. Table 1Institutions Time Definition of Essential Travel U.S. Centers for Disease Control and Prevention August 2021 The CDC does not impose uniform restrictions on domestic travel, allowing states to adopt different travel restrictions. Regarding international travel, the CDC identified a notice–Warning Level 3: Avoid all non-essential travel to certain destinations. California Department of Public Health April 2021 Essential travel is travel associated with the operation, maintenance, or usage of critical infrastructure or otherwise required or expressly authorized by law, including work and study, critical infrastructure support, economic services and supply chains, health, immediate medical care, and safety and security. Selected universities in the U.S. May 2021 Essential travel in university is defined as those that are necessary and cannot be postponed or handled remotely. (i.e., for graduation, academic progress, core educational, business functions of the University, etc.) New South Wales Government July 2021 Although NSW does not provide a definition of essential travel, it does specify appropriate means of transport for certain people in its guidelines. In addition, the guide defines the people who are allowed to make essential travel, and provides recommended means of transport for different travel times. In particular, it mentions that public transit, taxis or carpooling are not allowed to travel. European Commission May 2021 The EU does not give the definition of essential travel, but gives the people whose travel belongs to it, including but not limited to: healthcare professionals; seasonal workers in agriculture; transport personnel; passengers in transit; passengers travelling for imperative family reasons; third-country nationals travelling for the purpose of study, etc. Canadian Government June 2021 Some of the travel purposes that are considered essential may be: economic services and supply chains; critical infrastructure support; health (immediate medical care), safety and security; supporting goods for indigenous communities; any other activities that are deemed non-optional or non-discretionary. British Government June 2020 Essential travel may include the following travel purposes: to obtain basic necessities, including food and medical supplies and supplies for the essential upkeep, maintenance and functioning of the household, or to obtain money; to seek medical assistance; to travel for work; to fulfil a legal obligation; to access critical public services, including educational facilities, social services, etc. It is worth noting that even before the COVID-19 pandemic, researchers had been discussing essential travel. Some studies have examined the characteristics of essential travel and have used this concept to evaluate whether transport planning is adapting to travel demand, ensuring fairness, and optimizing resource allocation efficiency (Krumdieck et al., 2010; Laube et al., 2007). These studies defined essential travel as travel that contributes to people's health, work, income, and other basic needs, but relied mainly on small-scale surveys or subjective observations and explanations. Gordon et al. (1988) is one of the few quantitative studies that used the National Personal Transport Research (NPTS) in the 1980s to demonstrate that non-essential personal travel accounted for about 30% of its total travel. Krumdieck et al. (2010) categorized travel purposes into three categories: 50% for essential, 20% for necessary, and 10% for optional. However, empirical evidence is still limited at present. Since the outbreak of COVID-19, an increasing number of researchers have conducted quantitative studies on essential travel or non-essential travel in these unusual circumstances. Some scholars have evaluated the impact of non-essential travel on disease transmission. The results showed that for every 1% reduction in non-essential travel, there was an average reduction of 6.4% in new cases, but the definition of non-essential travel was uncertain (Sharkey and Wood, 2020). Another group of scholars linked essential travel to socio-economic status (SES). The research showed that COVID-19 exacerbated existing disparities in mobility between socioeconomic classes. The lower and middle socioeconomic groups mainly took long-distance and medium-distance trips for work, while the higher socioeconomic group mainly took short trips for recreational and other non-work purposes (Kara et al., 2021). There is little quantitative research on essential travel, and a big obstacle lies in the source of data. On one hand, it is difficult to separate the essential travel and non-essential travel in daily situations. On the other hand, it is difficult to obtain the travel data of various modes at the city level. However, as one of the important components of the urban transport system, taxis have the characteristics of 24-h operation and the starting and ending points are completely determined by passengers, and people are more inclined to use taxis than public transit in extreme cases such as pandemic (Tong, 2012). In addition, Xie (2018) shows that the statistical law of taxi travel behavior obeys power law distribution, which is consistent with the power law distribution characteristics of residents' travel behavior in city level, further verifying the rationality of using taxi travel to analyze residents' travel. Therefore, this paper intends to use the GPS data of taxi operation to carry out the essential travel-related modeling analysis. There is a lack of quantitative research on essential travel, and a significant challenge lies in obtaining suitable data sources. On the one hand, it is difficult to differentiate between essential and non-essential travel in everyday situations. On the other hand, it can be challenging to obtain travel data across different modes of transportation at the city level. However, taxis are a crucial component of the urban transport system due to their 24-h operation and the fact that their starting and ending points are entirely determined by passengers. Moreover, during exceptional circumstances such as a pandemic, people tend to prefer using taxis over public transit (Tong, 2012). Additionally, Xie (2018) shows that the statistical behavior of taxi travel conforms to a power law distribution, which is consistent with the power law distribution characteristics of residents' travel behavior at the city level. This finding further supports the rationality of using taxi travel data to analyze residents' travel patterns. Therefore, this study aims to utilize GPS data obtained from taxi operations to conduct modeling analysis on essential travel. The literature on essential travel has been expanded as the COVID-19 pandemic goes on, but it has not adequately shown how such travel differs from other travel in terms of temporal and spatial patterns, and how such travel is made out of purposes to obtain unevenly distributed socioeconomic opportunities. After all, travel is to help people acquire resources at reasonable prices and costs, so essential travel may be conceptually critical in ensuring the minimum level of personal and family development. Therefore, more research is needed to help understand and utilize the concept in the practices of transport planning. As mentioned above, this paper defines that “Essential Travel” refers to the travel that is still going on during the period of “Emergency Response”, so as to model it and quantitatively analyze its characteristics and influencing factors. The literature on essential travel has grown during the COVID-19 pandemic, but it has not adequately explored how this type of travel differs from other travel in terms of temporal and spatial patterns, or how it is shaped by the pursuit of unevenly distributed socioeconomic opportunities. Travel is intended to help people acquire resources at reasonable prices and costs, and essential travel may play a crucial role in ensuring a minimum level of personal and family development. Therefore, additional research is necessary to better understand and utilize the concept of essential travel in transport planning practices. As stated earlier, this paper defines essential travel as travel that continues during the “Emergency Response” period. The goal of this study is to model and quantitatively analyze the characteristics and influencing factors of essential travel. 3 Research design 3.1 The study area This study examines travel behaviors by taxi in the City of Xi ‘an between January and April 2020. Xi'an is one of the largest cities in China and a central city in the Northwest of the country. It has a population of 13 million and its built-up area reached 729 km2 in 2019. Xi'an had long been a capital city in the ancient China and thus developed a largely monocentric urban form so far. Similar to other large cities, the urban core is highly dense, while population density decreases as distance to the urban core increases (Qin et al., 2023). Fig. 1 displays the spatial distribution of taxi destinations in a workday.Fig. 1 Spatial distribution of taxi travel destinations. Fig. 1 This research focuses on Xi'an, China, for several reasons. Firstly, during the spring of 2020, when most people were making essential travel due to strict quarantine measures, Xi'an provides an excellent opportunity to observe the behaviors of essential travel. Secondly, while the pandemic impact varied among cities in China, most cities adopted similar quarantine guidelines. As a representative city of China, Xi'an can provide empirical evidence regarding the demand and supply of essential travel during this special period. Lastly, Xi'an was chosen for this study because data was collected covering different periods of the “Emergency Response”, and microscopic taxi data made these studies possible. Furthermore, it is important to note the policy of stay-at-home orders were implemented under emergency response status in Xi'an. Based on the investigation of the pandemic and expert judgment, the pandemic prevention and control command department divided different regions of the city into high, medium, and low-risk areas according to the level of epidemic risk. Residents in low-risk areas could choose to make essential travel, while residents in other areas were advised to stay at home to control the transmission of the virus as much as possible. Urban rail transit and ground public transit were temporarily suspended during the period. The study area is divided into 1 km × 1 km grid cells, and both the dependent and independent variables will be calculated at this level. Based on the previously mentioned details, it is assumed that the travel patterns observed by taxis are representative of overall travel patterns. To protect individuals' privacy, the original GPS data has been preprocessed, and travel tracks have been aggregated at the grid cell level (Zhou et al., 2022a). The dataset also includes information on travel start time, ending time, travel distance, and other relevant details. 3.2 The variables setting The study analyzed travel behaviors during different periods of the pandemic, dividing the collected data into three periods based on the start and end of the “Emergency Responses” in Xi'an: Pre, During, and Post. The “Emergency Response” in Xi'an was implemented on January 25th, 2020 and lifted on February 28th, 2020 (Government, 2020b). Fig. 2 shows significant differences in travel volumes across the three periods. In the Pre period, the daily average number of taxi-riding trips remained stable between 480,000 and 550,000, except for holidays when it would fluctuate. The number of taxi-riding trips dropped rapidly in the During period from January 21 due to the potential risks in travel. By January 25, the number of taxi-riding trips had plummeted to about 100,000, representing a decrease of more than 80%, which may have been due to the implementation of the strict Stay-at-Home policy. Over the next 10 days, it continued to drop until February 20th. In the Post period, with the loosening of the policy, the number of taxi-riding trips gradually rebounded. By the end of April, the daily passenger travel had recovered and reached about 400,000 trips a day, representing a recovery rate of about 80% compared to normal days.Fig. 2 Daily change of taxi-riding trips and Cumulative Confirmed Cases in China. Fig. 2 Assuming that the number of taxi-riding trips in each grid cell represents the actual taxi demand, we can consider it as a function of both travel demand and transport supply (see the function below Eq. (1)). Travel demands depend largely on the availability of socio-economic opportunities and resources such as restaurants, supermarkets, schools, hospitals, etc., in the vicinity of the destination. The density of each category of points of interest (POI) potentially affects an individual's likelihood to make travel decisions. Accessibility to a place is also a crucial factor that influences travel choices. The location of major transport infrastructure can attract people to visit certain places. For instance, places located in close proximity to freeways will have an advantage in people's travel choices.(1) Tripi=fTranSupplyiTravDemandi In the analysis, three independent variables are used: the average daily number of taxi-riding trips in each grid in the Pre, During, and Post periods. Table 2 describes the dependent variables. To compare the model coefficients across different periods, the dependent variables are normalized based on mean values.Table 2 Dependent variable. Table 2Dependent Variables Description Trippre Average number of trips on workday before pandemic, Jan.6-Jan.10 Tripdur Average number of trips on workday during pandemic, Feb.10-Feb.14 Trippost Average number of trips on workday after pandemic, Apr.20-Apr.24 The selection of independent variables is informed by prior research (Chung, 1997; Gutiérrez et al., 2011; Morrall and Bolger, 1996; Sung and Oh, 2011; Taylor and Fink, 2003; Taylor et al., 2009). Our focus is primarily on two dimensions of independent variables: travel demand and transport supply, and we choose appropriate forms of values for each variable. The specific variables are presented in Table 3 .Table 3 Description of independent variables. Table 3Type Variables Description Travel demand E_catering Number of essential catering POI per km2 NE_catering Number of non-essential catering POI per km2 E_shopping Number of essential shopping POI per km2 NE_shopping Number of non-essential shopping POI per km2 E_healthcare Number of essential health care POI per km2 NE_healthcare Number of non-essential health care POI per km2 E_education Number of essential education POI per km2 NE_education Number of non-essential education POI per km2 landuse_mix Herfindahl-Hirschman index (HHI) of land use mixes popden Population density (thousand person/ km2) Transport supply Airport_distance Distance to the nearest airport (km) Railway_distance Distance to the nearest railway station (km) Ramp_distance Distance to the nearest freeway ramp (km) Primary_density Road network density of the primary roads (km/km2) We would like to provide further clarification on the choice of dependent and independent variables. Firstly, we choose workdays as the dependent variables for the three periods, as they can better capture the essential travel. Additionally, as depicted in Fig. 2, there is hardly any difference in the number of taxi-riding trips between workdays and weekends during the “Emergency Response” period. Secondly, we have not included variables such as bus stop or subway station density or distance as independent variables, as public transit was completely closed during the “Emergency Response” period, leading to different variables for each period. Lastly, while there are many other potential independent variables to consider, such as secondary road density and walking distance, we have not included them due to their strong collinearity, which would impact the modeling and were abandoned in the pre-modeling. The travel demand variables used in this study serve as a proxy for localized resources and characteristics that attract travel. These variables provide an indication of how different types of socioeconomic opportunities can attract people to travel (Zhou et al., 2022b). Point of Interests (POI) data from Gaode Map Open Platform is used to measure these variables. Given the focus of this study on essential travel, we select four categories—catering, shopping, medical services, and science/culture & education—as major independent variables. Additionally, we matched the closest POI data of various types to the drop-off points of taxi passengers in the Pre and During periods, respectively (using the original latitude and longitude information of the drop-off point). Then, we associated their secondary POI classification attributes and grouped all POIs according to the secondary classification standard. The grouping statistical results are shown in Table 4 . In this study, we defined POIs that did not decrease in proportion in the During period as Essential POIs, while the rest were classified as Non-Essential POIs. The specific division results and quantity of Essential and Non-Essential POIs are shown in Fig. 3 .Table 4 Group statistics of the nearest POI from the taxi drop-off point. Table 4Primary Class Secondary Class POI Count POI Proportion Change Of Prop Primary Class Secondary Class POI Count POI Proportion Change Of Prop Pre During Pre During Pre During Pre During Catering Chinese Food Restaurant 271,845 9402 63.34% 64.16% 0.82% Healthcare General Hospital 28,701 1367 6.56% 8.90% 2.34% Foreign Food 8598 237 2.02% 1.62% −0.40% Special Hospital 62,652 2045 14.31% 13.30% −1.01% Fast Food Restaurant 84,553 3091 19.89% 21.09% 1.20% Clinic 74,140 2455 16.94% 16.00% −0.94% Leisure Food 1071 7 0.25% 0.05% −0.20% Emergency Center 1408 148 0.32% 1.00% 0.68% Coffee House 13,886 439 3.27% 3.00% −0.27% Disease Prevention Institution 1210 49 0.29% 0.29% 0.00% Tea House 13,243 408 3.11% 2.78% −0.33% Healthcare Products Store 259,805 8975 59.36% 58.40% −0.96% Ice Cream Shop 17,060 522 4.01% 3.56% −0.45% Veterinary Hospital 9785 327 2.24% 2.10% −0.14% Cake Shop 10,819 441 3.14% 3.01% −0.13% Others 102,237 3174 Dessert Sop 4108 108 0.97% 0.74% −0.23% Education Museum 3867 77 1.15% 0.64% −0.51% Others 114,755 3885 Exhibition Hall 4243 99 1.27% 0.83% −0.44% Shopping Shopping Plaza 1676 52 0.56% 0.50% −0.06% Conference & Exhibition Center 2160 48 0.64% 0.40% −0.24% Convenience Store 111,974 4049 37.70% 39.21% 1.51% Art Gallery 2277 63 0.68% 0.53% −0.15% Household Appliance 20,129 631 6.78% 6.11% −0.67% Library 4786 367 1.43% 3.07% 1.64% Supermarket 29,096 1198 9.80% 11.60% 1.81% Science & Technology Museum 545 16 0.16% 0.13% −0.03% Plants & Pet Store 14,475 479 4.87% 4.64% −0.23% Cultural Palace 2535 77 0.76% 0.64% −0.12% Home Building Materials 33,806 1114 11.38% 10.79% −0.59% Archives 359 16 0.12% 0.12% 0.00% Comprehensive Market 30,951 1214 10.42% 11.76% 1.34% Arts Organization 2529 92 0.76% 0.76% 0.00% Stationary Store 6329 206 2.13% 1.99% −0.14% Media Organization 9481 288 2.83% 2.41% −0.42% Sports Store 3253 90 1.10% 0.87% −0.22% School 79,837 3048 23.82% 25.48% 1.66% Commercial Street 4387 101 1.48% 0.98% −0.50% Scientific Research Institution 22,533 911 6.72% 7.61% 0.89% Clothing Store 32,894 920 11.07% 8.91% −2.17% Training Institution 168,721 5770 50.33% 48.23% −2.10% Special Trade House 788 19 0.27% 0.18% −0.08% Driving School 31,343 1092 9.35% 9.13% −0.22% Cosmetics Store 7260 253 2.45% 2.44% −0.01% Others 204,722 6576 Others 242,920 8214 Fig. 3 POI category and quantity. Fig. 3 The land use mix variable is derived from fine-grained land use information provided by the Gaode Map Open Platform. The Herfindahl–Hirschman index (HHI) is widely used to measure industry concentration in economics, and can be used to reflect diversity of land uses (Palan, 2010). The calculation formula of the HHI is shown in Eq. (2). A small HHI value indicates a greater mixability.(2) HHIi=∑j=1NXijXi2 Where Xi is the total number of POIs in grid i, and Xij is the total number of POIs of category j in grid i. Fig. 4 depicts the spatial distribution of catering POIs for both essential and non-essential types. It is evident from the figure that non-essential catering POIs are mainly concentrated in the urban core within the second Ring Road, while essential catering POIs are more widely distributed, covering almost all major urban subcenters (Giuliano et al., 2019). This could be because non-essential catering businesses such as tea houses and dessert houses tend to locate in densely populated areas where there is adequate demand to support their operations. Fig. 5 illustrates the spatial distribution of essential and non-essential types of shopping POIs. Although the distribution of both groups is generally similar, non-essential shopping businesses cluster in a few hotspots that are not shown in the essential shopping business map. For instance, agglomerations of home building materials markets can be easily seen in many cities due to the apparent advantages of agglomerative economies of scale. People tend to visit these agglomerations to purchase materials in bulk, and businesses located there can benefit from premium volumes of potential customers. Both figures demonstrate the different spatial patterns of essential businesses compared to non-essential ones, which could contribute to the wide spatiotemporal variations in travel demand during the COVID pandemic.Fig. 4 Spatial distribution of POIs in essential and non-essential catering businesses. Fig. 4 Fig. 5 Spatial distribution of POIs in essential and non-essential shopping businesses. Fig. 5 3.3 Spatial autocorrelation analysis Because of spatial heterogeneity of socioeconomic opportunities, the spatial patterns of dependent variables and independent variables are both non-stationary across space. Travel by taxi is distributed along the road network, but the grid is artificially generated regardless of the network. Therefore, the travel demand by taxi can be highly subject of spatial autocorrelation: such demand in a grid cell may be highly correlated with its adjacent spatial units. As spatial autocorrelation can be classified into global spatial autocorrelation and local spatial autocorrelation, we examined them by calculating the Global Moran's I and local Moran's I respectively. Global Moran's I calculated as Eq. (3), which is a global measure of spatial autocorrelation (Moran, 1950):(3) I=n∑i=1n∑j=1nwij∑i=1n∑j=1nwijxi−X¯xj−X¯∑i=1nxi−X¯2 Where n is the number of spatial units, i and j are longitudinal and latitudinal indexes, x denotes the variable, X¯is the mean of x, and wij indicates the spatial weight between i and j . Global Moran's I index is a value ranging from −1 to 1. The spatial distribution is more similar to clustering of dissimilar values if Global Moran's Index approaches −1; otherwise, it would be more similar to clustering of similar values. The value tends to be randomly distributed in space if Global Moran's Index approaches 0. In order to achieve two purposes, Anselin (1995) proposed Local Moran's I, which can be interpreted as indicators of local pockets of nonstationarity or hot spots, and be used to assess the influence of individual locations on the magnitude of the global statistic and to identify “outliers”. The expression is shown as Eq. (4):(4) Ii=nxi−X¯∑i=1nxi−X¯2∑jwijxj−X¯ When Local Moran's I index>0, it means that surrounding units have similar values to the unit; When Local Moran's I index<0, it indicates surrounding units have dissimilar values to the unit. And when Local Moran's I index = 0, it implies that surrounding units have no relationship with the unit. 3.4 Regression models If there are different degrees of spatial correlation, the traditional panel model can not be used in setting model, such as the Ordinary Least-Squares (OLS) regressions (Elhorst, 2014). Because spatial correlation will lead to the correlation of error terms in linear regression model, or will lead to biased estimation results, thus spatial regression model should be chosen. The commonly spatial models contain the spatial lag model (SLM), the spatial lag of X model (SLX) and the spatial Durbin model (SDM). The SLM model is suitable for the endogenous spatial correlation between dependent variables, focusing on the spatial spillover effect of dependent variables. The SLX model is suitable for the endogeneity of independent variables, and the adjacency matrix W in the model can be parameterized to better adapt to different spatial distributions (Vega and Elhorst, 2015). The SDM model is a general form of spatial model, which can be simplified to the first two under certain conditions, including not only the spatial lag of the dependent variable, but also the spatial lag of the independent variable. Its original form is shown as Eq. (5).(5) y=λWy+Xβ+WXθ+ε=I−λW−1Xβ+I−λW−1WXθ+ε Where, y is a vector (n1) of observations of the dependent variable; X is matrix(nk) of observations of the independent variables; λ, β and θ are vector(k1) of regression coefficients; W is spatial weight matrix and ε is a vector(n1) of error terms. SDM model theory holds that the observed values of dependent variables are influenced not only by the dependent variables in the surrounding areas, but also by the independent variables in the surrounding areas, so the key factors affecting the dependent variables can be observed more comprehensively from the endogenous and exogenous perspectives (Borst and McCluskey, 2007; Elhorst, 2014). Therefore, the total marginal effect of independent variables on dependent variables can be further divided into direct effect and indirect effect. Through Taylor expansion, it's easy to verify that I−λW−1=I+λW+λ2W2+λ3W3+…. Suppose that X contains k explanatory variables and the r−th explanatory variable is Xr=x1rx2r…xnr(n × 1), then we can get Xβ=x1…xkβ1…βk=∑r=1kβrxr, so Eq. (5) can be rewritten as Eq. (6):(6) y=∑r=1kβrI−λW−1xr+I−λW−1WXθ+ε Suppose SrW=βrI−λW−1, then Eq. (6) can be expanded into Eq. (7):(7) y1⋮yn=SrW11⋯SrW1n⋮⋱⋮SrWn1⋯SrWnnx1r⋮xnr According to Eq. (7), it can be seen that ∂yi∂xjr=SrWij. This indicates that the variable xjr in region j may have an impact on the dependent variable in any other region i, which is the spatial autocorrelation effect of the spatial model. Specifically, if j=i, then ∂yi∂xir=SrWii, which corresponds to the diagonal element of Eq. (7). This can be understood as the direct effect of the independent variable xir in region i on the dependent variable yi in the same region. Therefore, the direct effect of the independent variable x on the dependent variable y is the average of all diagonal elements in Eq. (7), while the indirect effect is the average of all non-diagonal elements (Chen, 2010). There are two parameter estimation methods for spatial models, one is generalized two-stage least-squares cross-sectional regression method (gs2sls), the other is maximum likelihood estimation method (ML). The main difference between them is that it is more effective when the error obeys the assumption of normal distribution, otherwise it is not as robust as the former, for example, in the presence of heteroscedasticity. The heteroscedasticity of the dependent variable in this paper is proved by using White test, so the parameter estimation method of gs2sls is used in this paper. 4 Results 4.1 Spatial analysis: overall distribution and autocorrelation Fig. 6 illustrates the spatial distribution of taxi travel demand in the different periods of the pandemic, revealing significant disparities in demand across the three periods. In the Pre period, taxi travel was primarily concentrated within the beltway, with numerous hotspots in the urban core. However, as a result of the Stay-at-Home policy, almost all hotspots disappeared in the During period, and travel intensity declined precipitously. In the Post period, travel demand had rebounded and had nearly returned to pre-pandemic levels.Fig. 6 Hotspots of taxi destinations in different periods. Fig. 6 Table 5 displays the global Moran's I index for the number of taxi-riding trips in the three periods. The results indicate that the global Moran's I index for all three periods is significantly greater than 0, which suggests that the number of taxi-riding trips is spatially autocorrelated. Furthermore, we calculated the local Moran's I index for the number of taxi-riding trips in each period. The results show that the spatial distribution of the local Moran's I index is similar across all three periods. Specifically, the index is mostly significant within the second Ring Road, characterized by a High-High Cluster. Outside the second Ring Road, the index is generally not significant, with only a few areas exhibiting a Low-High Cluster, such as the airport and the edge of the second Ring Road.Table 5 Global Moran's I index of the number of taxi-riding trips in three periods. Table 5Period Moran's I z-score p-value Pre 0.80 81.31 0.00 During 0.69 70.58 0.00 Post 0.81 82.30 0.00 4.2 Inferential analysis 4.2.1 Model comparison The study utilizes both global OLS regression model and spatial regression models to examine the relationship between the number of taxi-riding trips and various independent variables in different pandemic periods. Table 6 provides a comparison of the models to show how the coefficients change when controlling for the spatial autocorrelation of the dependent and independent variables. The results indicate that the Pseudo R-squared of the SLX and SDM models is larger than that of the global OLS model. Additionally, the two indexes of Log-Likelihood and AIC suggest that all three spatial models are superior to the OLS model. These findings suggest that incorporating spatial dimensions is necessary to more accurately estimate relationships of interest.Table 6 Results of models. Table 6Independent variables OLS SLM SLX SDM Pre During Post Pre During Post Pre During Post Pre During Post E_catering 0.254* 0.197* 0.486* 0.140* 0.123* 0.330* 0.103 0.137* 0.292* 0.110 0.136* 0.302* NE_catering 2.079* 1.052* 2.722* 1.804* 0.887* 2.338* 1.039*** 0.233* 1.204* 1.046* 0.234* 1.206* E_shopping 0.238 0.546*** 0.239* −0.065 0.278* −0.174 0.013 0.183 −0.105 0.009 0.182** −0.116 NE_shopping 0.0242 0.0186 0.042 −0.008 −0.002 −0.002 0.021 0.029 0.035 0.019 0.030 0.034 E_healthcare 0.610* 0.676 0.451 1.178* 1.034* 1.203* 0.006 0.225 −0.447 −0.1787 0.251 −0.692* NE_healthcare 0.2262 0.1587 −0.045 0.377** 0.279* 0.164 0.675* 0.510* 0.565 0.671* 0.514* 0.558 E_education 1.943* 1.437* 2.350* 1.079* 0.814* 1.161* 0.694* 0.318* 0.536* 0.654* 0.323* 0.475* NE_education 0.855* −0.009 1.380* 0.400 −0.286 0.757*** 0.244* −0.101 0.574*** 0.258* −0.106 0.595* landuse_mix 5.301* 5.207* 6.555 16.888* 12.912* 22.523*** 3.399* 1.540 3.562 4.001** 1.453 4.279* Airport_distance 0.159* 0.105 0.226* −0.107 −0.065* −0.147 0.943* 1.130* 1.240* 1.100* 1.090* 1.440* Railway_distance −0.191* −0.138* −0.259* 0.726* 0.514* 0.997* −1.48* −1.050* −1.900* −1.350* −1.080* −1.720*** Ramp_distance −0.227* −0.170 −0.359 0.453* 0.319* 0.587* 0.151 −0.432* 0.224 −0.008 −0.404* 0.016 Primary_density 5.475* 2.087* 6.390* 1.921* −0.181 1.593 1.963* 0.742 1.750 2.019* 0.731 1.784* Popden 2.150* 2.300* 2.850* 1.070* 1.360* 1.360*** 0.252* −0.048 0.049 0.146 −0.021 −0.104 _cons −7.523* −6.249* −9.309* −65.486* −47.878* −88.473* −16.676 −37.835* −25.898* −27.625 −34.861* −39.396 Independent variables OLS SLM SLX SDM Pre During Post Pre During Post Pre During Post Pre During Post W × E_catering 10.254* −1.082 14.417* 12.172* −1.243 17.010* W × NE_catering 157.776* 92.046* 214.344* 177.457* 88.798* 242.104* W × E_shopping −17.962* 26.841* −28.256* −16.971* 25.608* −26.495* W × NE_shopping −1.161 0.241 −0.454 −1.778 0.2953 −1.316 W × E_healthcare −134.080* −12.471 −148.218* −182.545* −5.786 −211.137* W × NE_healthcare −31.642* −84.959* −44.265* −36.040* −83.107* −50.862* W × E_education −82.778* −34.020* −108.264* −82.146* −34.052* −107.944* W × NE_education 21.401 −7.110 20.129 30.372* −7.626 30.845 W × landuse_mix 24.744 67.831 40.992 49.815 60.465 72.699* W × Airport_distance −2.840* −3.780* −3.640* −3.170* −3.690* −4.040* W × Railway_distance 6.340* 4.730* 8.390* 5.720* 4.850* 7.590* W × Ramp_distance −2.690* 1.640 −4.330** −2.290 1.600 −3.920* W × Primary_density −83.550* −65.663* −122.577* −86.957* −63.945* −129.087* W × popden 12.000* 26.600* 19.400* 18.500* 24.800* 27.500* W × Y 2.664* 2.597* 2.706* −1.795* 0.446 −1.802* R-squared 0.594 0.548 0.603 0.522 0.451 0.529 0.740 0.662 0.756 0.743 0.662 0.759 Log-Likelihood 9373 10,271 7896 9998 10,666 8572 10,539 11,031 9175 10,548 11,032 9179 AIC −18,717 −20,511 −15,761 −19,961 −21,298 −17,110 −21,018 −22,003 −18,289 −21,036 −22,005 −18,299 BIC −18,618 −20,413 −15,663 −19,850 −21,186 −16,999 −20,820 −21,806 −18,092 −20,839 −21,815 −18,102 N of Obs. 5268 Note: The dependent variable was transformed by multiplying the value by 1000. E.g.: 1000 essential catering POI is associated with 0.111 taxi trip arrivals in the SDM model. Table 6 reveals differences among the spatial models as well. SLM (Y-Lag only), SLX (X-Lag only), and SDM (X-Lag & Y-Lag) exhibit a gradual increase in R-squared and Likelihood, and a corresponding decrease in AIC. The results indicate that the SDM model performs the best, implying that controlling the spatial autocorrelation of both dependent and independent variables can effectively help estimate taxi travel demand in different pandemic periods. Compared to the global OLS model, the coefficients of several variables in the SDM model exhibit different statistical significance, while the signs of coefficients remain the same. These findings suggest that the importance of a certain variable in predicting the number of taxi-riding trips may change when the spatial autocorrelations are controlled. Moreover, the results reveal that the number of taxi-riding trips in a certain spatial unit is not only related to the influencing factors of this unit, but also to the influencing factors in adjacent units. Hence, the SDM model can better reflect the degree of influence of different variables on the dependent variable. 4.2.2 Variable association analysis According to the results in Table 6, the factors affecting the number of taxi-riding trips in different periods show substantial differences in terms of their coefficients. In the Pre and Post periods, the impacts of the independent variables are similar to a large extent, with some notable differences. Firstly, in the Post period, there is a significant suppression of travel demand for essential healthcare, which may be due to people's fear of potential risks during the pandemic. Secondly, non-essential education strengthens its impact in the Post period, likely because people choose to use taxis as a more private and safer travel option for that purpose. In contrast, the influencing factors of taxi travel demand in the During period are significantly different from those in the other two periods. Firstly, the importance of E_catering, E_shopping, and E_education does not significantly decrease and even slightly increases, as evidenced by the significant positive coefficient of E_shopping in the During period only. This strongly confirms the hypothesis of this study that essential travel is closely associated with basic socio-economic providers, such as local food stores, convenience stores, and important educational facilities. Secondly, the transport supply type variables, such as landuse_mix and Primary_density, are no longer significant in the During period. It is possible that during the Emergency Response periods, when the main purpose of travel is essential, people may not be concerned about building characteristics such as land use patterns and road network accessibility, which led to the disappearance of the effects of these variables. The SDM model also reveals the spatial spillover effects of variables, which vary across different pandemic periods. Firstly, the spillover effect of E_shopping is negative in the Pre and Post periods, but positive in the During period. This suggests that in the Pre and Post periods, the surrounding areas suppress E_shopping in a region, while in the During period, they drive E_shopping in a region. This may be due to limited supply of necessary goods during the Emergency Response period, leading people to travel to areas with concentrated convenience stores, thus increasing the probability of completing essential shopping. Secondly, in the During period, the spillover effects of E_catering, E_healthcare, NE_education, and Y-lag are no longer significant. This may indicate that, in the case of virus transmission, people directly go to their destination without the need for more contact with nearby facilities. For example, under normal circumstances, a person may not only look for restaurants in the geographic unit where the taxi arrives but also look for drinks in nearby units after dinner. However, during the Emergency Response period, they may only focus on the unit where the target restaurant is located. Thus, spillover effects are greatly compromised in the During period. Based on these findings, the SDM model can better reveal the complex relationships of variables that change over different periods. 4.2.3 Combined spatial effect analysis Appendix A presents the direct, indirect, and total marginal effects of each independent variable in the SDM model across three different periods. The marginal effects of the independent variables on the dependent variables can be divided into two parts: direct effects from local units and indirect effects from surrounding geographical units. It is important to note that direct impact and indirect impact should be treated independently since they may represent localized impact and sub-regional impact, respectively, in the urban environment. Therefore, when the direct effect is offset by the indirect effect and the total effect may not be significant, it does not necessarily mean that the variables are irrelevant. On the contrary, the influence of the variables varies within and outside geographical units. The results presented in Appendix A reveal several key findings. Firstly, in the Pre and Post periods, the direct effects of the independent variables on taxi travel demand are notably smaller than the indirect effects (which are an order of magnitude larger). This indicates that the demand for taxi travel in a given grid cell is significantly influenced by the demand from its surrounding region. In contrast, in the During period, only direct effects are observed while indirect effects are not statistically significant. This suggests that the demand for taxi travel during this period was more constrained to the destination, with little spillover effect to the surrounding areas. Secondly, for variables such as essential shopping, healthcare, and essential education, the direct effect is generally positive, but the indirect effect is strongly negative or insignificant. As a result, the total spillover effect is negative or insignificant. This finding highlights the complex spatial relationships among different variables at the regional level, which cannot be adequately captured by a simple OLS model. Hence, a more sophisticated spatial model like SDM is required for accurate analysis. Overall, the results demonstrate the importance of considering both direct and indirect effects when examining the impact of independent variables on taxi travel demand. Moreover, the findings underscore the need for a spatially explicit model that can account for the complex spatial relationships among variables. 5 Discussion and conclusion The empirical spatial model has demonstrated that taxi travel is influenced by various factors, such as individual travel demands and transport supply. Building on the existing literature, this paper specifically investigates the shifts in taxi travel demand during different pandemic periods and identifies unique travel patterns during the Emergency Response period, which helps redefine essential travel in a more precise manner. Additionally, by examining the travel disparities across different pandemic periods, we can provide more effective optimization strategies for public transit to better serve essential travel in the post-pandemic era. First, the SDM model proves to be a better method for characterizing the relationship between taxi travel demand and various factors. Given that taxi travel demand is influenced by numerous factors and its spatial distribution is highly uneven, this paper compares the global OLS model to a range of spatial models. The SDM model, which demonstrates superior performance, not only estimates the spatial relationship between taxi travel demand and various factors, but also calculates the spatial spillover effect between local and surrounding regions. As a result, public administration officials may need to take into account neighboring units when predicting travel demand in each geographic unit. For example, in order to estimate travel demand for healthcare services, facilities within a certain distance should be considered to account for spillover effects, although such effects could vary across different periods. Secondly, this paper provides a clearer definition of essential travel by contrasting travel differences across pandemic periods. The contribution of essential travel purposes to taxi travel demand do not decrease significantly and even slightly increase, while that of non-essential travel purposes dropped significantly. This indicates the critical role of essential travel during the extreme conditions of society-wide quarantine. People prioritize visiting resources and facilities that provide the most vital services, such as convenience stores to obtain daily supplies, rather than purchasing luxury goods in large shopping plaza. These differences have been identified through empirical analysis, leading to a more explicit concept of essential travel. Thirdly, in the During period, taxi travel demand was found to be highly associated with localized resources, while the impact of surrounding areas was significantly reduced compared to the Pre and Post periods. Therefore, essential travel during this period is closely related to local businesses and facilities that provide essential services. Consequently, addressing travel demand from a spatial perspective during the pandemic requires a context-based approach. While socioeconomic resources can typically be accessed over a wide geographic area during normal times, this is much less feasible during events like the COVID-19 pandemic. Similarly, post-pandemic, the differences in mobility between individuals will also need to be considered when planning for essential resource provision. Ensuring local access to essential resources should be the primary focus of facility planning by public administration in order to enhance the welfare of those in need. The findings of this study can inform the optimization of public transit services in a hierarchical manner during the post-pandemic period. In order to ensure that all citizens have access to essential resources, it is crucial to accurately identify the destinations of essential travel and efficiently connect individuals to these locations. To assess the adequacy of the current public transit system in meeting this demand, we further analyzed the coverage of major essential travel destinations by bus in a grid cell-based manner (Table 7 and Fig. 7 ). Table 7 indicates that the bus system can meet 66.32%, 77.22%, and 65.29% of travel demand in the Pre, During, and Post periods, respectively. In the During period, bus services can cover more travel destinations than the other two periods, regardless of grid cell category. These results suggest that the current public transit system has done a relatively good job in matching essential demand as compared to non-essential demand. However, the findings also highlight the need to prioritize improving transit services, particularly by expanding the bus network to grid cells with the highest quartiles of essential travel demand but without a current bus line. For instance, Fig. 7 illustrates that the grid cells with red boundaries in the first two quartiles have high essential travel demand, but no public transit line passes through them. Therefore, these areas should be given priority consideration in the optimization of the public transit system.Table 7 Percent of travel demand covered by public transit. Table 7Level Pre During Post Count Covered Percent Count Covered Percent Count Covered Percent 0 3053 317 10.38% 4153 925 22.27% 3015 315 10.45% 1 553 240 43.40% 278 163 58.63% 563 225 39.96% 2 554 356 64.26% 279 199 71.33% 563 338 60.04% 3 554 378 68.23% 279 232 83.15% 563 406 72.11% 4 554 495 89.35% 279 267 95.70% 564 502 89.01% Overall 66.3% 77.2% 65.3% Fig. 7 Travel demand and public transport line coverage. Fig. 7 The clarification of the concept of essential travel is a significant contribution of this study. The existing literature lacks a uniform standard for defining essential travel, which is typically associated with travel for food, work, medicine, etc. This paper proposes a novel approach to define essential travel based on local travel demand and transport supply, providing a decision-making basis for optimizing public transit that caters to essential travel in the post-pandemic era. The public administration has been increasingly concerned about the inequity in access to spatial opportunities among socially disadvantaged people. The findings of this research can provide evidence for identifying communities' capability of providing essential services. At the end of this paper, we want to acknowledge some limitations of research design and data analysis. This paper focuses on the spatial distribution of taxi travel demand in different periods with regard to the pandemic, and mainly discusses its patterns relative to building environment, facility accessibility and population. Some more refined characteristics, such as population age structure, and family income were not included in this study due to the data unavailability. Admittedly, understanding these factors may be of great help to modeling and estimating essential travel demand. In addition, although spatial models are found to be one of the best choices in this study, we expect to see other advanced methods such as machine learning techniques to explore this topic in future research. Finally, due to the limited data, we failed to obtain data on all modes of travel in different periods and had to rely on taxi GPS tracks only. In addition, since the taxi data does not have the attributes of travel purpose, we cannot carry out statistics on whether the travel is essential or not according to the purpose. In the future, more research may be needed to apply the empirical model to other cities and countries, so as to explore essential travel from a more general perspective. Author contributions The authors confirm contribution to the paper as follows: Chao Yang: Methodological Support, Research Design, and Editing; Zhiyang Wan: Data analysis, Model building and validation, Writing - original draft, and Manuscript Draft Preparation; Quan Yuan: Data curation, Research Design, and Manuscript Revision; Zhou Yang: Research Design, and Editing; Maopeng Sun: Data Collection, and Editing. Appendix A Spatial effect analysis of SDM Unlabelled TableExplanatory variables Pre During Post Direct effect Indirect effect Total effect Direct effect Indirect effect Total effect Direct effect Indirect effect Total effect E_catering 0.103 4.208* 4.311* 0.136* −2.067 −1.931 0.293* 5.803* 6.096* NE_catering 0.948* 61.700* 62.648* 0.261* 155.513 155.774 1.071* 84.096* 85.167* E_shopping 0.019 −5.969 −5.951 0.190 44.937 45.127 −0.101 −9.213* −9.314* NE_shopping 0.020 −0.637 −0.617 0.030 0.540 0.570 0.034 −0.483 −0.448 E_hospital −0.077 −64.035* −64.112* 0.249 −9.926 −9.677 −0.574 −73.567* −74.141* NE_hospital 0.692* -13.088* −12.396* 0.489* −144.974 −144.486 0.587* −18.180* −17.592* E_education 0.701* −29.280* −28.579* 0.313* −59.313 −59.000 0.536* −38.134* −37.598* NE_education 0.242 10.510* 10.752* −0.108 −13.422 −13.530 0.578* 10.436 11.014 landuse_mix 3.977 14.982 18.959* 1.472 106.903 108.375 4.242* 22.779 27.021 Airport_distance 1.100* −1.810* −0.704* 1.090* −5.610 −4.520 1.450* −2.330* −0.880* Railway_distance −1.350* 2.860* 1.510* −1.080* 7.640 6.560 −1.730* 3.750* 2.020*** Ramp_distance −0.007 −0.799 −0.806 −0.404* 2.480 2.070 0.018 −1.380 −1.370 Primary_desity 2.070* −31.831* −29.761* 0.711 −111.285 −110.574 1.858* −46.372* −44.514* Popden 0.136 6.420* 6.560* −0.013 43.400 43.400 −0.120 9.690* 9.570* Notes: for significant correlation ‘’ = 0.01; ‘’ = 0.05; ‘’ = 0.1. Data availability Data will be made available on request. ==== Refs References Aldila D. Khoshnaw S.H. Safitri E. Anwar Y.R. Bakry A.R. Samiadji B.M. Anugerah D.A. Gh M.F.A. Ayulani I.D. Salim S.N. A mathematical study on the spread of COVID-19 considering social distancing and rapid assessment: the case of Jakarta, Indonesia Chaos, Solitons Fractals 139 2020 110042 Anselin L. Local indicators of spatial association—LISA Geogr. Anal. 27 1995 93 115 Beck M.J. Hensher D.A. Insights into the impact of COVID-19 on household travel and activities in Australia–The early days under restrictions Transp. Policy 96 2020 76 93 Bergantino A.S. Intini M. Tangari L. Influencing factors for potential bike-sharing users: an empirical analysis during the COVID-19 pandemic Res. Transp. Econ. 86 2021 101028 Borst R. McCluskey W. Comparative evaluation of the comparable sales method with geostatistical valuation models Pac. Rim Prop. Res. J. 13 2007 106 129 de Bruin Y.B. Lequarre A.-S. McCourt J. Clevestig P. Pigazzani F. Jeddi M.Z. Colosio C. Goulart M. Initial impacts of global risk mitigation measures taken during the combatting of the COVID-19 pandemic Saf. Sci. 128 2020 104773 C.D.C.P Domestic Travel during COVID-19 2022 Centers for Disease Control and Prevention Available at: https://www.cdc.gov/coronavirus/2019-ncov/travelers/travel-during-covid19.html#quick-reference C.D.P.H Travel Advisory 2021 California Department of Public Health Available at: https://www.cdph.ca.gov/programs/CID/DCDC/pages/COVID-19/Travel-Advisory.aspx Campisi T. Basbas S. Skoufas A. Akgün N. Ticali D. Tesoriere G. The impact of COVID-19 pandemic on the resilience of sustainable mobility in Sicily Sustainability 12 2020 8829 Chen Q. Advanced Econometrics and Stata Application 2010 Chen X. Guo Y. Yang C. Ding F. Yuan Q. Exploring essential travel during COVID-19 quarantine: evidence from China Transp. Policy 111 2021 90 97 Chinazzi M. Davis J.T. Ajelli M. Gioannini C. Litvinova M. Merler S. Pastore y Piontti A. Mu K. Rossi L. Sun K. The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak Science 368 2020 395 400 32144116 Chung K. Estimating the Effects of Employment, Development Level and Parking Availability on CTA Rapid Transit Ridership: From 1976 to 1995 in Chicago, Proceedings 1997 Clarke S. The traffic data that shows the road into—and out of—Covid-19 lockdown Guardian 27 2020 Cui Q. He L. Liu Y. Zheng Y. Zhou M. Measuring the impacts of the COVID-19 pandemic on China's transport sectors based on the CGE model coupled with a decomposition analysis approach Transp. Policy 103 2021 De Haas M. Faber R. Hamersma M. How COVID-19 and the Dutch ‘intelligent lockdown'change activities, work and travel behaviour: evidence from longitudinal data in the Netherlands Transport. Res. Interdisc. Perspect. 6 2020 100150 E.U The Temporary Restriction on Non-essential Travel into the EU and the Possible Lifting of Such Restriction 2020 European Union Available at: https://eur-lex.europa.eu/eli/reco/2020/912/oj Elhorst J.P. Spatial Econometrics from Cross-Sectional Data to Spatial Panels 2014 Springer Giuliano G. Kang S. Yuan Q. Agglomeration economies and evolving urban form Ann. Reg. Sci. 63 2019 377 398 Gordon P. Kumar A. Richardson H.W. Beyond the journey to work Transp. Res. Part A Gen. 22 1988 419 426 Government B. The Health Protection (Coronavirus, Restrictions) (England) Regulations 2020 British Government Government S.P. Shaanxi Launched a Class I Emergency Response to Public Health Emergencies 2020 Gutiérrez J. Cardozo O.D. García-Palomares J.C. Transit ridership forecasting at station level: an approach based on distance-decay weighted regression J. Transp. Geogr. 19 2011 1081 1092 He Q. Rowangould D. Karner A. Palm M. Larue S. Covid-19 Pandemic Impacts on Essential Transit Riders: Findings from a U.S. Survey 2021 SocArXiv Health N.G. Appropriate transportation for essential travel for certain groups of people 2021 Available at: https://www.health.nsw.gov.au/Infectious/factsheets/Pages/essential-travel.aspx Hurley D. Flattening traffic congestion: an analysis of how COVID-19 has changed our driving patterns 2020 Available at: https://towardsdatascience.com/flattening-traffic-congestion-b668b1fcda0f IEA Global Energy Review 2020 International Energy Agency Available at: https://www.iea.org/reports/global-energy-review-2020 Immigration T.V.O.C. Canada refines definition of essential travel during coronavirus 2021 Available at: https://www.cicnews.com/2020/05/canada-refines-definition-of-essential-travel-during-coronavirus-0514270.html#gs.c12y8t Kara A. Lea H.T. Millera H. What Is Essential Travel? Socio-Economic Differences in Travel Demand during the COVID-19 Lockdown 2021 Annals of the American Association of Geographers Krumdieck S. Page S. Dantas A. Practice Urban form and long-term fuel supply decline: a method to investigate the peak oil risks to essential activities Transp. Res. A Policy Pract. 44 2010 306 322 Labonté-LeMoyne É. Chen S.-L. Coursaris C.K. Sénécal S. Léger P.-M. The unintended consequences of COVID-19 mitigation measures on mass transit and car use Sustainability 12 2020 9892 Laube M.M. Lyons W.M. Allan D. System planning for quality transit projects Transp. Res. Rec. 1992 2007 20 27 Marchant A. What Can Traffic Data Tell us about the Impact of the Coronavirus 2020 Moving World Moran P.A. Notes on continuous stochastic phenomena Biometrika 37 1950 17 23 15420245 Morrall J. Bolger D. The relationship between downtown parking supply and transit use ITE J. 66 1996 32 Palan N. Measurement of Specialization the Choice of Indices 2010 (FIW working paper) Przybylowski A. Stelmak S. Suchanek M. Mobility behaviour in view of the impact of the COVID-19 pandemic—public transport users in Gdansk case study Sustainability 13 2021 364 Qin Z. Liang Y. Yang C. Fu Q. Chao Y. Liu Z. Yuan Q. Externalities from restrictions: examining the short-run effects of urban core-focused driving restriction policies on air quality Transp. Res. Part D: Transp. Environ. 119 2023 103723 Rahman M.M. Fattah I. Ong H.C. Mahlia T. Alam M.A. Impact of COVID-19 on the social, economic, environmental and energy domains: lessons learnt from a global pandemic Sustain. Product. Consumpt. 26 2021 343 359 Shakibaei S. De Jong G.C. Alpkökin P. Rashidi T.H. society Impact of the COVID-19 pandemic on travel behavior in Istanbul: a panel data analysis Sustain. Cities 65 2021 102619 Sharkey P. Wood G. The causal effect of social distancing on the spread of SARS-CoV-2 2020 Available at: https://osf.io/preprints/socarxiv/hzj7a/ Sung H. Oh J.-T. Transit-oriented development in a high-density city: identifying its association with transit ridership in Seoul, Korea Cities 28 2011 70 82 Taylor B.D. Fink C.N. The factors influencing transit ridership: a review and analysis of the ridership literature 2003 Available at: https://escholarship.org/uc/item/3xk9j8m2 Taylor B.D. Miller D. Iseki H. Fink C. Practice Nature and/or nurture? Analyzing the determinants of transit ridership across US urbanized areas Transp. Res. A Policy Pract. 43 2009 60 77 Teixeira J.F. Lopes M. The link between bike sharing and subway use during the COVID-19 pandemic: the case-study of New York's Citi bike Transport. Res. Interdisc. Perspect. 6 2020 100166 Tong X. The analysis of City Dweller's travel choice behavior based on taxi GPS trajectory data Comput. Telecommun. 2012 University D. COVID-19 Essential Travel Guidance 2021 Drexel University Available at: https://drexel.edu/procurement/about/covid-19-operations/travel-guidance/ Vega S.H. Elhorst J.P. The SLX model J. Reg. Sci. 55 2015 n/a-n/a Villas-Boas S.B. Sears J. Villas-Boas M. Villas-Boas V. Are we StayingHome to Flatten the Curve? Working Paper Series 2020 Department of Agricultural Resource Economics, UC Berkeley WHO Coronavirus disease (COVID-19) travel advice 2020 Available at: https://www.who.int/news-room/q-a-detail/coronavirus-disease-covid-19-travel-advice-for-the-general-public Xie K. Study on Trip Behavior Using Taxi GPS Dataset 2018 Beijing Institute of Technology Zhang Y. Fricker J.D. Quantifying the impact of COVID-19 on non-motorized transportation: a Bayesian structural time series model Transp. Policy 103 2021 11 20 Zhang L. Darzi A. Ghader S. Pack M.L. Xiong C. Yang M. Sun Q. Kabiri A. Hu S. Interactive covid-19 mobility impact and social distancing analysis platform Transp. Res. Rec. 03611981211043813 2020 Zhou Y. Zhang Y. Yuan Q. Yang C. Guo T. Wang Y. The smartphone-based person travel survey system: data collection, trip extraction, and travel mode detection IEEE Trans. Intell. Transp. Syst. 23 12 2022 23399 23407 Zhou Y. Yuan Q. Ding F. Chen M. Yang C. Guo T. Demand, mobility, and constraints: exploring travel behaviors and mode choices of older adults using a facility-based framework J. Transp. Geogr. 102 2022 103368
PMC010xxxxxx/PMC10288319.txt	==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 The Author(s). Published by Elsevier Ltd. S0264-410X(23)00758-2 10.1016/j.vaccine.2023.06.069 Article Immediate adverse reactions following COVID-19 vaccination among 16-65-year-old Danish citizens Torp Hansen K. a1⁎ Kusk Povlsen F. ab1 Hammer Bech B. a Nygaard Hansen S. a Ulrikka Rask C. cd Fink P. ce Jørgensen T. fg Nielsen H. hi Meinertz Dantoft T. f Marie Thysen S. f Rytter D. a a Department of Public Health, Aarhus University, DK-8000 Aarhus, Denmark b Department of Quality and Patient Involvement, Aarhus University Hospital, DK-8200 Aarhus, Denmark c Department of Clinical Medicine, Aarhus University, DK-8200 Aarhus, Denmark d Department of Child and Adolescent Psychiatry, Aarhus University Hospital, DK-8200 Aarhus, Denmark e Research Clinic for Functional Disorders and Psychosomatics, Aarhus University Hospital, DK-8200 Aarhus, Denmark f Center for Clinical Research and Prevention, Copenhagen University Hospital – Bispebjerg and Frederiksberg, DK-2200 Copenhagen, Denmark g Department of Public Health, Faculty of Health and Medical Sciences, University of Copenhagen, DK-2200 Copenhagen, Denmark h Department of Infectious Diseases, Aalborg University Hospital, DK-9100 Aalborg, Denmark i Department of Clinical Medicine, Aalborg University, DK-9000 Aalborg, Denmark ⁎ Corresponding author at: Department of Public Health, Aarhus University, Bartholins Allé 2, Aarhus, 8000, Denmark. 1 Equal first authors 23 6 2023 23 6 2023 17 4 2023 14 6 2023 21 6 2023 © 2023 The Author(s) 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction There is sparse knowledge of immediate adverse reactions following COVID-19 vaccination. Objective This study aimed to estimate the frequency and number of immediate adverse reactions following COVID-19 vaccination in a Danish population. Methods The study used data from the Danish population-based cohort study BiCoVac. The frequencies of 20 self-reported adverse reactions were estimated for each vaccine dose stratified by sex, age, and vaccine type. Also, the distributions of number of adverse reactions following each dose were estimated stratified by sex, age, vaccine type, and prior COVID-19 infection. Results A total of 889,503 citizens were invited and 171,008 (19%) vaccinated individuals were included in the analysis. The most frequently reported adverse reaction following the first dose of COVID-19 vaccine was redness and/or pain at the injection site (20%) while following the second and third dose, tiredness was the most frequently reported adverse reaction (22% and 14%, respectively). Individuals aged 26-35 years, females, and those with a prior COVID-19 infection were more likely to report adverse reactions compared with older individuals, males, and those with no prior COVID-19 infection, respectively. Following the first dose, individuals vaccinated with ChAdOx1-2 (AstraZeneca) reported more adverse reactions compared with individuals vaccinated with other vaccine types. Individuals vaccinated with mRNA-1273 (Moderna) reported more adverse reactions following the second and third dose compared with individuals vaccinated with BNT162b2 (Pfizer-BioNTech). Conclusion The frequency of immediate adverse reactions was highest among females and younger persons, however, most of the Danish citizens did not experience immediate adverse reactions following COVID-19 vaccination. Keywords COVID-19 vaccination Adverse reactions ==== Body pmc1 Background In January 2020 WHO declared COVID-19 an international crisis for public health [1]. This led to several political and health-related initiatives [2], [3]. In December 2020, the first COVID-19 vaccine was authorized by the European Medicines Agency (EMA) [4] and within a few months, three other vaccine candidates were approved [5], [6], [7] Development of a new vaccine is usually a process that takes several years. Due to the urgency of the COVID-19 pandemic, the vaccine development occurred at an accelerated rate. The implementation of COVID-19 vaccines was monitored closely by the authorities with registration of adverse reactions (AR) [8]. Reports of thromboembolic events in individuals vaccinated with the ChAdOx1-2 vaccine led to a rapid assessment of adverse events using the Danish registers [9]. As a reaction to the findings, the Danish authorities suspended the ChAdOx1-2 vaccine [10] and it was later excluded from the Danish vaccination program [11]. Several studies have examined ARs to the different COVID-19 vaccine types [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22]. However, most of the published studies are small and many were conducted only among healthcare staff [22], [15], [16], [17], [18], [19], [20], thus in selected populations, which are not representative for the general population. The most commonly reported ARs include pain at the injection site [12], [13], [15], [17], [18], [20], [21], [22], headache [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], fatigue [12], [13], [16], [17], [18], [20], [21], [22], joint pain [21], [22], [15], [16], [17], [18], myalgia [15], [16], [17], [19], [20], [21], [22], shiver/chills [16], [17], [20], [21], and fever [14], [16], [21], [22]. In the present study, we report the frequency of immediate ARs following COVID-19 vaccination in Danish citizens and describe the distribution of the individual ARs as well as number of immediate ARs by dose and according to predefined subgroups (sex, age, vaccine type, and prior COVID-19 infection). 2 Methods 2.1 Study setting The study was conducted in the Danish general population. In Denmark, the COVID-19 vaccines were introduced late December 2020 [23]. Healthcare staff, citizens aged 65 and above, and citizens with increased risk of severe COVID-19 disease (e.g. severe chronic disease [24]) and their relatives were first in line for COVID-19 vaccines. Hereafter the COVID-19 vaccines were gradually made available to all Danish adults, starting with the oldest age groups [25]. Individuals above 18 years were recommended to take the third COVID-19 vaccine dose 140 days after the second dose [26]. 2.2 The BiCoVac cohort BiCoVac is a Danish population-based cohort [27]. The BiCoVac cohort was established to obtain information on immediate and long-term symptoms following COVID-19 vaccination. A random sample of Danish citizens, living in Denmark at the beginning of April 2021, and born between 1956 and 2004 were obtained from the Danish Central Person Register (25 % from each birth year). A total of 911,613 Danes aged 16-65 years were selected for the study. The baseline questionnaires were sent out through a national digital mail system (e-Boks) in May 2021. The baseline questionnaire was timed to obtain information before the COVID-19 vaccines were made available for most Danes. The distribution of the first and second follow-up questionnaires was planned according to the vaccination plan so that most participants would have been offered their first and second dose at the time of the questionnaire. A third follow-up questionnaire was distributed approximately 1 year after baseline. Individuals aged 30-34 years did not receive the third follow-up questionnaire, due to an error in the distribution. Participants were only invited for follow-up questionnaires if they had answered the previous questionnaire. The questionnaires were available in Danish, English, and Arabic. In the present study, participants could be included if they had received one or more COVID-19 vaccines, based on information from the questionnaires. 2.3 Variables The questionnaires contained information on 20 suspected ARs following COVID-19 vaccination. For each dose of the vaccine, participants were asked if they had experienced any of the following ARs in the week following vaccination: Redness and/or pain at the injection site, skin rash, nausea, vomiting, fever, shivering/chills, tiredness, general malaise, joint pain, muscle pain, headache, diarrhoea, dizziness, swollen lymph nodes, facial swelling, facial paralysis, pain in the arm and legs, allergic reaction, shortness of breath, or bruising/bleeding under the skin. Participants were provided with the following options for each of the ARs 1) No, 2) Yes, mild symptoms, 3) Yes, moderate symptoms, and 4) Yes, severe symptoms. For the analyses, we dichotomized the variables to having experienced the AR following vaccination (Yes, moderate symptoms or Yes, severe symptoms) or not having experienced the AR (No or Yes, mild symptoms). Based on information on the individual ARs, the number of ARs was calculated for each individual and categorized as 0, 1, 2-4, 5-9 or 10+ ARs. In the analyses for each dose, we only included individuals, who had provided information on all 20 ARs. The questionnaires also contained questions regarding COVID-19 infection and vaccine type (see the Appendix for detailed descriptions of these variables). Age and sex were available from the Central Person Register. Age was categorized into the following age groups 17-25, 26-35, 36-45, 46-55, and 56+. 2.4 Statistical analyses Characteristics of the study population were described according to the vaccine doses by numbers and percentages. Further, frequencies of the individual ARs were estimated for each vaccine dose overall and stratified by sex, age group, and vaccine type, respectively. The distribution of number of ARs was estimated stratified by sex, age, vaccine type, and prior COVID-19 infection, respectively. Also, in sub-analyses, the analysis stratified on prior COVID-19 infection was further stratified on sex, age, and vaccine type, respectively (first and second vaccine dose only, due to small groups when stratifying on the third dose). Differences in distributions of numbers of ARs between groups were tested with a χ2 test. The frequency and number of ARs for the ChAdOx1-2 vaccine were only estimated for the first dose, because of few individuals having received the ChAdOx1-2 vaccine as their second or third dose. Since the type of vaccine received varied by age and sex, a multiple logistic regression was used to estimate the relationship between odds of reporting 5 symptoms and above and sex, age, vaccine type, and prior COVID-19 infection (mutually adjusted). All analyses were made using Stata version 17. 2.5 Ethics The BiCoVac study was approved by the Danish Data Protection Agency under the Aarhus University comment agreement (journal number 2015-57-0002) and Aarhus University journal number 2016-051-000001, sequential number 2272 (25/3-2021). According to Danish legislation, ethical approval of survey studies based on questionnaires is not required. 2.6 Role of the funding source TrygFonden had no role in the study design; in the collection, analysis, or interpretation of data; writing the report; or in the decision to submit the article for publication. 2.7 Results Of the 911,613 individuals selected for the BiCoVac study, 22,110 individuals had no digital mailbox and were therefore not invited. In total, 889,503 individuals were invited to the BiCoVac study and 252,268 (28%) filled in the baseline questionnaire. Of these, 59% participated in the first follow-up, 43% participated in the second follow-up and 25% participated in the third follow-up. A total of 171,008 individuals responded to all ARs following the first COVID-19 vaccine dose. The corresponding numbers were 130,351 and 54,903 for the second and third dose respectively (Figure 1 ). The number of individuals included in some of the analyses differs from these totals due to missing information on one or more variables or small groups (<5).Figure 1 Flow chart The participation rate for the baseline questionnaire was 23% for males and 34% for women, and it increased with age ranging from 16%-47%. In the current study there was therefore a larger proportion of females and older individuals than in the invited population (Table 1 ). Most individuals received the BNT162b2 vaccine (79 %, 85 %, and 86 % for the first, second, and third dose, respectively), while few received the ChAdOx1-2 vaccine (10 %, <1 %, and <1 %, respectively). Most individuals did not have a COVID-19 infection prior to receiving their COVID-19 vaccine.Table 1 Characteristics of the participants by vaccine dose. 1. vaccine N (%) 2. vaccine N (%) 3. vaccine N (%) Sex Men 64,849 (38) 47,223 (36) 20,783 (38) Women 106,159 (62) 83,128 (64) 34,120 (62) Age 17-25 10,640 (6) 6,525 (5) 1,779 (3) 26-35 16,039 (9) 11,711 (9) 2,027 (4) 36-45 24,558 (14) 18,758 (14) 6,839 (12) 46-55 48,404 (28) 36,863 (28) 16,635 (30) 56+ 71,367 (42) 56,494 (43) 27,623 (50) Vaccine type BNT162b2 134,632 (79) 110,870 (85) 47,425 (86) mRNA-1273 16,623 (10) 17,914 (14) 6,690 (12) ChAdOx1-2 16,921 (10) 463 (<1) 179 (<1) Other/unknown 2,832 (2) 1,104 (1) 609 (1) COVID-19 infection Prior infection 9,283 (5) 7,216 (6) 4,498 (8) No prior infection 146,898 (86) 111,101 (85) 42,220 (77) Unknown 14,827 (9) 12,034 (9) 8,185 (15) 2.8 Frequencies of adverse reactions A total of 38 %, 36 %, and 25 % reported one or more ARs following the first, second, and third dose of COVID-19 vaccine, respectively (Figure 2 ). The proportion reporting severe symptoms was the same after both the first and second vaccine dose (11%), but it was lower after the third dose (6%). The most frequently reported ARs were redness and/or pain at the injection site (Dose 1: 20 %, Dose 2: 15 %, Dose 3: 9%) and tiredness (Dose 1: 19 %, Dose 2: 22 %, Dose 3: 14%).Figure 2 Prevalence of reported individual moderate or severe adverse reactions following COVID-19 vaccination by vaccine dose. The types of the most frequently reported ARs were similar for men and women. However, most ARs were reported more frequently by women than men (Supplementary figure 1). Also, the ARs reported were similar across age groups, although most ARs were reported more frequently by individuals younger than 36 years compared with the older age groups (Supplementary Figure 2, Figure 3, Figure 4 ).Figure 3 Percentage of respondents reporting 0, 1, 2-4, 5-9 or 10+ moderate or severe immediate adverse reactions following COVID-19 vaccination after first, second and third dose stratified by sex. * p value < 0.001 Figure 4 Percentage of respondents reporting 0, 1, 2-4, 5-9 or 10+ moderate or severe immediate adverse reactions following COVID-19 vaccination after first, second and third dose stratified by age. * p value < 0.001 The frequencies of each AR differed between the vaccine types: 53 % vaccinated with ChAdOx1-2 experienced tiredness after the first dose compared with 21 % for mRNA-1273 and 14 % for BNT162b2 (Supplementary figure 5 ).Figure 5 Percentage of respondents reporting 0, 1, 2-4, 5-9 or 10+ moderate or severe immediate adverse reactions following COVID-19 vaccination after first, second and third dose stratified by vaccine type. * p value < 0.001 2.9 Number of adverse reactions Overall, women reported more ARs than men (Figure 3) and the number of ARs reported was generally lower for the higher age groups. This pattern was present for all vaccine doses (Figure 4). Stratified by vaccine type, individuals vaccinated with ChAdOx1-2 reported more ARs following the first dose (71.7 % reported 1 or more and 6.5 % reported 10 or more ARs) compared with individuals vaccinated with BNT162b2 (32.1 % reported 1 or more and less than 0.4 % reported 10 or more ARs) and mRNA-1273 (51.3 % reported 1 or more and 0.8 % reported 10 or more ARs). Following the second and third dose, individuals vaccinated with mRNA-1273 reported more ARs compared with individuals vaccinated with BNT162b2 (Figure 5). In general, individuals with prior COVID-19 infection reported more ARs than those with no prior COVID-19 infection (Figure 6 ). This was also the case when further stratified for sex and vaccine type, respectively (Supplementary figures 6 and 7). The same pattern was observed for the first dose when stratified by age. However, there was no statistically significant difference in the number of ARs between those who had a prior COVID-19 infection and those who did not for the second dose for participants in age groups below 46 years of age (Supplementary figure 8).Figure 6 Percentage of respondents reporting 0, 1, 2-4, 5-9 or 10+ moderate or severe immediate adverse reactions following COVID-19 vaccination after first, second and third dose stratified by prior COVID-19 infection. * p value < 0.001 When mutually adjusting for sex, age, vaccine type, and prior COVID-19 infection, men and individuals with higher age had lower odds of reporting 5 or more ARs following COVID-19 vaccination compared with women and individuals aged 16-25 years (Table 2 ). This pattern was observed for all vaccine doses. Individuals with a prior COVID-19 infection had higher odds of reporting 5 or more ARs compared with individuals with no prior COVID-19 infection (1. Dose: OR=2.79 (CI95% 2.61;2.99) 2. Dose: OR=1.52 (CI95% 1.41;1.64) 3. Dose: OR=1.59 (CI95% 1.40;1.79)).Table 2 Odds ratios for reporting 5 or more immediate adverse reactions for vaccine type, sex, age and prior COVID-19 infection by vaccine dose. 1. vaccine 2. vaccine 3. vaccine n (%) OR (95% CI) n (%) OR (95% CI) n (%) OR (95% CI) Vaccine type BNT162b2 4033 (3) 1.00 6365 (6) 1.00 1633 (4) 1.00 mRNA-1273 1052 (7) 1.95 (1.81-2.10) 4551 (28) 5.12 (4.89-5.35) 695 (12) 3.08 (2.80-3.40) ChAdOx1-2 5830 (38) 15.51 (14.80-16.25) Sex Women 9054 (9) 1.00 8749 (12) 1.00 1886 (7) 1.00 Men 1861 (3) 0.51 (0.48-0.54) 2167 (5) 0.44 (0.42-0.46) 442 (3) 0.40 (0.36-0.44) Age 16-25 1050 (11) 1.00 799 (14) 1.00 116 (8) 1.00 26-35 1674 (12) 0.83 (0.76-0.92) 2153 (20) 0.97 (0.89-1.07) 171 (12) 1.06 (0.82-1.36) 36-45 2012 (9) 0.65 (0.59-0.71) 2151 (13) 0.79 (0.72-0.87) 380 (7) 0.78 (0.63-0.97) 46-55 3002 (7) 0.55 (0.50-0.59) 2691 (8) 0.58 (0.53-0.64) 719 (5) 0.70 (0.57-0.86) 56+ 3177 (5) 0.43 (0.39-0.46) 3122 (6) 0.43 (0.39-0.47) 942 (4) 0.55 (0.45-0.67) COVID-19 infection No prior infection 9486 (7) 1.00 9964 (9) 1.00 1983 (5) 1.00 Prior infection 1429 (16) 2.79 (2.61-2.99) 952 (13) 1.52 (1.41-1.64) 345 (8) 1.59 (1.40-1.79) Individuals vaccinated with ChAdOx1-2 and mRNA-1273 had higher odds (OR=15.51 (CI95% 14.80;16.25) and OR=1.95 (CI95% 1.81;2.10), respectively) of reporting 5 or more ARs following the first vaccine dose compared with individuals vaccinated with BNT162b2. After the second and third dose individuals vaccinated with mRNA-1273 had higher odds than individuals vaccinated with BNT162b2. 3 Discussion The present study is one of the first large population-based studies in Denmark to investigate immediate ARs following the first, second, and third COVID-19 vaccine dose, respectively. The most frequently reported ARs following COVID-19 vaccination were redness and/or pain at the injection site after the first dose, and tiredness after the second and third dose. In general, women reported more ARs than men and younger participants reported more ARs than older participants. These patterns were observed for all vaccine doses. Individuals with prior COVID-19 infection reported more ARs following COVID-19 vaccination than individuals with no prior COVID-19 infection. For the first dose, individuals who received ChAdOx1-2 reported the most ARs compared to individuals receiving mRNA-1273 and BNT162b2. 3.1 Strengths and limitations A strength of the study is the population-based design and large sample size of the BiCoVac cohort. Further, it is a strength that data was obtained during the pandemic and timed according to the implementation of the COVID-19 vaccine schedule in the general Danish population. However, this was also a limitation as vaccine schedules and recommendations were continuously adapted. Thus, despite efforts to distribute the questionnaires according to the vaccination schedule, questionnaires were not distributed according to the vaccination date for the specific individual, which could potentially increase the risk of recall bias. The BiCoVac cohort had a low response rate at baseline varying with age and sex of the invited individuals, with lowest participation among men and younger individuals. The study can therefore be subject to selection issues, which could affect both the overall prevalence of ARs in the population and in the comparison of subgroups. It can be speculated that individuals, who experience ARs are more likely to participate. This could lead to an overestimation of ARs. However, it could also be speculated that the low participation of particularly younger individuals may have caused an underestimation of the prevalence of ARs in the population, as we found more ARs in younger individuals. When comparing the prevalence of ARs after the third dose with that of the first and second dose, it is also important to bear in mind, that only few individuals aged 30-34 years were included in the analyses for the third dose. Also, the time from vaccination to answering the questionnaire on symptoms was longer for the third vaccine dose compared with the first and second dose. Hence participants may have underestimated the prevalence of immediate symptoms after the third dose. The study aimed to describe the distributions of each AR and the number of ARs according to each dose with descriptive analyses. In addition, an adjusted analysis supported the pattern observed in the descriptive analyses. 4 Comparison with other studies Other studies found pain at the injection site [17], [18], [20], [21], [22], fatigue [16], [17], [18], [20], [21], [22], and headache [16], [17], [18], [19], [20], [21], [22] to be the most common ARs following COVID-19 vaccination. These studies mainly included health professionals and other selected populations and are thus potentially not directly comparable with our study in a general population. However, the types of ARs reported in these studies were similar to those identified in the present study. However, the proportions of the different ARs varied considerably in the literature: Previous studies report proportions of headache varying between 10 % and 45 % [22], [16], [17], [18], [19], [20], while the present study found proportions in headache between 7 % and 12 % depending on dose. These differences might be explained by different definitions of ARs (the present study only count moderate and severe symptoms), differences in sample sizes, and different study populations (all the studies have healthcare workers as their population) [22], [16], [17], [18], [19], [20]. In correspondence with the present study, most previous studies found a higher frequency of symptoms in women compared with men [18], [20], [22] which could be due to women generally reporting more somatic symptoms than men [28]. Also, in line with the present study, previous studies found that younger individuals reported more ARs than older individuals [22], [18], [19], [20]. A study by Beatty et al. found that individuals vaccinated with mRNA-1273 experienced more ARs than individuals vaccinated with BNT162b2 [29], which corresponds with the finding that more individuals vaccinated with BNT162b2 presented with serological vaccine hyporesponsiveness at day 90 compared with individuals vaccinated with mRNA-1273 [30]. In the present study, a similar pattern was observed with more individuals reporting ARs following mRNA-1273 than following BNT162b2. A study in UK found that for the first dose, individuals vaccinated with ChAdOx1-2 reported systemic adverse effects more frequently than individuals vaccinated with BNT162b2. However, the opposite was found for local adverse effects [31]. Similarly, another study found that individuals vaccinated with ChAdOx1-2 reported systemic adverse events more frequently after the first dose than individuals vaccinated with mRNA-1273 and BNT162b2. After the second dose individuals vaccinated with mRNA-1273 reported systemic adverse events more frequent than individuals vaccinated with the two other types [32]. The present study also found more frequent ARs following COVID-19 vaccination with ChAdOx1-2 at the first dose compared to mRNA-1273 and BNT162b2. We however did not have enough participants receiving ChAdOx1-2 to investigate immediate symptoms following the second and third dose of ChAdOx1-2. We found that individuals with prior COVID-19 infection reported more ARs compared with individuals with no prior COVID-19 infection, which could reflect immunological recall [33]. Similar findings were observed in other studies [19], [29], [34], but not all [18]. A study among hospital workers found that individuals with prior COVID-19 infection were more likely to report symptoms after the first COVID-19 vaccine dose (but not the second dose) compared with individuals with no prior COVID-19 infection [35]. A study among individuals in nursing homes did not find higher rates of more severe ARs in individuals with prior COVID-19 infection than without prior COVID-19 infection [36], potentially due to a weaker immunological response in the elderly. The prevalence of COVID-19 infection reported in the study and in the Danish population at the time was comparable. Hence the prevalence in the current study was between 5-8%, varying with vaccine dose, compared with a prevalence between 5-13% in the general Danish population at the time most Danish citizens got vaccinated (May 2021-December 2021) [37]. 4.1 Conclusion In this large population-based study, we found that the majority of individuals vaccinated did not experience the investigated immediate ARs following COVID-19 vaccination. Redness and/or pain at the injection site were the most frequently reported ARs following the first dose of COVID-19 vaccine. Tiredness was the most frequently reported AR following the second and third dose. Females, younger individuals, and those with a prior COVID-19 infection reported ARs more frequently. Individuals vaccinated with ChAdOx1-2 reported the most ARs and individuals vaccinated with BNT162b2 reported the least ARs. 5 Authors’ roles KTH, FKP, BHB, SMT, and DR participated in the conception of the study and design, writing of the article, and interpretation of the results. KTH, FKP, and DR performed the analysis. KTH and FKP wrote the first draft of the manuscript. All authors critically revised, commented on, and approved the final manuscript. Funding This work was supported by TrygFonden (id-number: 153678). Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability The authors do not have permission to share data. ==== Refs References 1 SSI. Tidslinje for covid-19 2022 [Available from: https://covid19.ssi.dk/-/media/arkiv/subsites/covid19/presse/tidslinje-over-covid-19/covid-19-tidslinje-lang-for-2020-2022-version-2---december-2022.pdf. [Last accessed: 14/2 2023] 2 Andersen TM. Effekterne af ikke-farmaceutisk intervention under Covid-19 pandemien - En oversigt Finansministeriet2021 [Available from: https://fm.dk/media/25165/5-baggrundspapir-effekterne-af-ikke-farmaceutisk-intervention-under-covid-19-pandemien-en-oversigt.pdf. [Last accessed: 3 Perra N. Non-pharmaceutical interventions during the COVID-19 pandemic: A review Phys Rep. 913 2021 1 52 33612922 4 EMA recommends first COVID-19 vaccine for authorisation in the EU: European Medicines Agency; 2020 [Available from: https://www.ema.europa.eu/en/news/ema-recommends-first-covid-19-vaccine-authorisation-eu. [Last accessed: 08-02 2022] 5 EMA recommends COVID-19 Vaccine Moderna for authorisation in the EU: European Medicines Agency; 2021 [Available from: https://www.ema.europa.eu/en/news/ema-recommends-covid-19-vaccine-moderna-authorisation-eu. [Last accessed: 14/2 2023] 6 EMA recommends COVID-19 Vaccine AstraZeneca for authorisation in the EU European Medicines Agency2021 [Available from: https://www.ema.europa.eu/en/news/ema-recommends-covid-19-vaccine-astrazeneca-authorisation-eu. [Last accessed: 9/3 2023] 7 EMA recommends COVID-19 Vaccine Janssen for authorisation in the EU European Medicines Agency2021 [Available from: https://www.ema.europa.eu/en/news/ema-recommends-covid-19-vaccine-janssen-authorisation-eu. [Last accessed: 14/2 2023] 8 Generelt om vacciner mod COVID-19: Lægemiddelstyrelsen [Available from: https://laegemiddelstyrelsen.dk/da/nyheder/temaer/generelt-om-vacciner-mod-covid-19/. [Last accessed: 03-03-2022 2022] 9 Pottegård AL, L.C.; Karlstad, Ø.; Dahl, Jesper; Andersen, Morten; et al. Arterial events, venous thromboembolism, thrombocytopenia, and bleeding after vaccination with Oxford-AstraZeneca ChAdOx1-S in Denmark and Norway: population based cohort study. BMJ. 2021(373). 10 Sundhedsstyrelsen. Pausering af brug af COVID-19 Vaccine AstraZeneca i det danske COVID 19 vaccinationsprogram 2021 [Available from: https://www.sst.dk/-/media/Udgivelser/2021/Corona/Vaccination/Notater/Notat-om-pausering-af-COVID-19-Vaccine-AstraZeneca-i-det-danske-COVID-19-vaccinationsprogram.ashx. [Last accessed: 04.11. 2022] 11 Sundhedsstyrelsen. Danmark fortsætter vaccineudrulning uden AstraZeneca-vaccinen 2021 [Available from: https://www.sst.dk/da/Nyheder/2021/Danmark-fortsaetter-vaccineudrulning-uden-AstraZeneca-vaccinen. [Last accessed: 14.04.21 12 Auster O. Finkel U. Dagan N. Barda N. Laufer A. Balicer R.D. Short-term Adverse Events After the Third Dose of the BNT162b2 mRNA COVID-19 Vaccine in Adults 60 Years or Older JAMA Netw Open. 5 4 2022 e227657 35435973 13 Menni C. May A. Polidori L. Louca P. Wolf J. Capdevila J. COVID-19 vaccine waning and effectiveness and side-effects of boosters: a prospective community study from the ZOE COVID Study Lancet Infect Dis. 22 7 2022 1002 1010 35405090 14 Chen P.-Y. Wu B.-J. Su M.-C. Lin Y.-H. Chiang S.-C. Wu J.-C. Risk Factors and Incidence Rates of Self-Reported Short-Term Adverse Events of COVID-19 Vaccine Booster Dose Vaccines (Basel) 10 7 2022 1115 35891279 15 Ali M.D. Almadan L.Z. Alghamdi R.A. Alghamdi A.S. Almarhoon S.A. Hassan Y.AM. Evaluation of Prevalence of Side-Effects Associated with Booster Dose of mRNA-Based COVID-19 Vaccine Among Healthcare Workers in Eastern Province, Saudi Arabia: A Descriptive Cross-Sectional Study Infect Drug Resist. Volume 15 2022 4335 4346 16 Kadali R.A.K. Janagama R. Peruru S. Malayala S.V. Side effects of BNT162b2 mRNA COVID-19 vaccine: A randomized, cross-sectional study with detailed self-reported symptoms from healthcare workers Int J Infect Dis. 106 2021 376 381 33866000 17 Im J.H. Kim E. Lee E. Seo Y. Lee Y. Jang Y. Adverse Events with the Pfizer-BioNTech COVID-19 Vaccine among Korean Healthcare Workers Yonsei Med J. 62 12 2021 1162 34816647 18 Tosun S. Ozkan Ozdemir H. Erdogan E. Akcay S. Aysin M. Eskut N. Adverse events report of inactivated COVID-19 vaccine from 4040 healthcare workers Postgrad Med. 134 1 2022 104 110 34705583 19 Quiroga B. Sánchez-Álvarez E. Goicoechea M. de Sequera P. COVID-19 vaccination among Spanish nephrologists: Acceptance and side effects J Healthc Qual Res. 36 6 2021 363 369 34244126 20 Riad A. Hocková B. Kantorová L. Slávik R. Spurná L. Stebel A. Side Effects of mRNA-Based COVID-19 Vaccine: Nationwide Phase IV Study among Healthcare Workers in Slovakia Pharmaceuticals (Basel) 14 9 2021 873 34577573 21 Mohammed R.A. Garout R.M. Wahid S. Ayub F. Firas ZinAlddin L.M. Sultan I. A Survey on the Side Effects of Pfizer/BioNTech COVID-19 Vaccine Among Vaccinated Adults in Saudi Arabia Cureus. 13 11 2021 e19222 34873547 22 Borroni E. Consonni D. Cugno M. Lombardi A. Mangioni D. Bono P. Side effects among healthcare workers from a large Milan university hospital after second dose of BNT162b2 mRNA COVID-19 vaccine Med Lav. 112 6 2021 477 485 34939617 23 Hvor mange danskere bliver vaccineret mod covid-19? Statens Serum Institut2020 [Available from: https://www.ssi.dk/aktuelt/nyheder/2020/12_hvor-mange-danskere-bliver-vaccineret-mod-covid-19_29122020. [Last accessed: 20/02 2023] 24 Sundhedsstyrelsen. Personer med øget risiko ved COVID-19 - Fagligt grundlag. 2021 25 Sundhedsstyrelsen. Vaccinationskalender 2021 [Available from: https://www.sst.dk/-/media/Udgivelser/2021/Corona/Vaccination/Kalender/Vaccinationskalender-01072021.ashx?sc_lang=da&hash=2F626F7E6433723A88DE3B36228F8C03. [Last accessed: 03-04 2023] 26 Sundhedsstyrelsen. Tilbud om 3. stik mod covid-19. 27 Bicovac [Available from: https://ph.au.dk/bicovac. [Last accessed: 03-04 2023] 28 Barsky A.J. Peekna H.M. Borus J.F. Somatic symptom reporting in women and men J Gen Intern Med. 16 4 2001 266 275 11318929 29 Beatty A.L. Peyser N.D. Butcher X.E. Cocohoba J.M. Lin F. Olgin J.E. Analysis of COVID-19 Vaccine Type and Adverse Effects Following Vaccination JAMA Netw Open. 4 12 2021 e2140364 34935921 30 Søgaard O.S. Reekie J. Johansen I.S. Nielsen H. Benfield T. Wiese L. Characteristics associated with serological COVID-19 vaccine response and durability in an older population with significant comorbidity: the Danish Nationwide ENFORCE Study Clin Microbiol Infect. 28 8 2022 1126 1133 35283313 31 Menni C. Klaser K. May A. Polidori L. Capdevila J. Louca P. Vaccine side-effects and SARS-CoV-2 infection after vaccination in users of the COVID Symptom Study app in the UK: a prospective observational study Lancet Infect Dis. 21 7 2021 939 949 33930320 32 Kant A. Jansen J. van Balveren L. van Hunsel F. Description of Frequencies of Reported Adverse Events Following Immunization Among Four Different COVID-19 Vaccine Brands Drug Saf. 45 4 2022 319 331 35314943 33 Chen Y. Tong P. Whiteman N. Sanjari Moghaddam A. Zarghami M. Zuiani A. Immune recall improves antibody durability and breadth to SARS-CoV-2 variants Science Immunology. 7 78 2022 34 Krammer F. Srivastava K. Alshammary H. Amoako A.A. Awawda M.H. Beach K.F. Antibody Responses in Seropositive Persons after a Single Dose of SARS-CoV-2 mRNA Vaccine N Engl J Med. 384 14 2021 1372 1374 33691060 35 Debes A.K. Xiao S. Colantuoni E. Egbert E.R. Caturegli P. Gadala A. Association of Vaccine Type and Prior SARS-CoV-2 Infection With Symptoms and Antibody Measurements Following Vaccination Among Health Care Workers JAMA Intern Med. 181 12 2021 1660 34398173 36 Bardenheier B.H. Gravenstein S. Blackman C. Gutman R. Sarkar I.N. Feifer R.A. Adverse Events Following One Dose of mRNA COVID-19 Vaccination Among US Nursing Home Residents With and Without a Previous SARS-CoV-2 Infection J Am Med Dir Assoc. 22 11 2021 2228 2232 34534492 37 Coronavirus (COVID-19) Cases Our World in Data2023 [Available from: https://ourworldindata.org/covid-cases. [Last accessed: 08-06 2023]
PMC010xxxxxx/PMC10288320.txt	==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 The Author(s). Published by Elsevier Ltd. S0264-410X(23)00743-0 10.1016/j.vaccine.2023.06.058 Article Public Perception of COVID-19 Vaccines through Analysis of Twitter Content and Users Saleh Sameh N. ab⁎ McDonald Samuel A. bc Basit Mujeeb A. ab Kumar Sanat bd Arasaratnam Reuben J. a Perl Trish M. a Lehmann Christoph U. be Medford Richard J. ab a Department of Internal Medicine, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390 b Clinical Informatics Center, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390 c Department of Emergency Medicine, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390 d Lebanon Trail High School, 5151 Ohio Dr, Frisco, TX 75035 e Departments of Pediatrics, Bioinformatics, Population & Data Sciences, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390 ⁎ Corresponding author at: 5323 Harry Hines Blvd, Dallas, TX 75219 23 6 2023 23 6 2023 4 6 2021 3 5 2023 15 6 2023 © 2023 The Author(s). Published by Elsevier Ltd. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background With the global continuation of the COVID-19 pandemic, the large-scale administration of a SARS-CoV-2 vaccine is crucial to achieve herd immunity and curtail further spread of the virus, but success is contingent on public understanding and vaccine uptake. We aim to understand public percepion about vaccines for COVID-19 through the wide-scale, organic discussion on Twitter. Methods This cross-sectional observational study included Twitter posts matching the search criteria ((‘covid’ OR ‘coronavirus’) AND ‘vaccine’) posted during vaccine development from February 1st through December 11th, 2020. These COVID-19 vaccine related posts were analyzed with topic modeling, sentiment and emotion analysis, and demographic inference of users to provide insight into the evolution of public attitudes throughout the study period. Findings We evaluated 2,287,344 English tweets from 948,666 user accounts. Individuals represented 87.9% (n=834,224) of user accounts. Of individuals, men (n=560,824) outnumbered women (n=273,400) by 2:1 and 39.5% (n=329,776) of individuals were ≥40 years old. Daily mean sentiment fluctuated congruent with news events, but overall trended positively. Trust, anticipation, and fear were the three most predominant emotions; while fear was the most predominant emotion early in the study period, trust outpaced fear from April 2020 onward. Fear was more prevalent in tweets by individuals (26.3% vs. organizations 19.4%; p<0.001), specifically among women (28.4% vs. males 25.4%; p<0.001). Multiple topics had a monthly trend towards more positive sentiment. Tweets comparing COVID-19 to the influenza vaccine had strongly negative early sentiment but improved over time. Interpretation This study successfully explores sentiment, emotion, topics, and user demographics to elucidate important trends in public perception about COVID-19 vaccines. While public perception trended positively over the study period, some trends, especially within certain topic and demographic clusters, are concerning for COVID-19 vaccine hesitancy. These insights can provide targets for educational interventions and opportunity for continued real-time monitoring. Keywords COVID-19 Vaccine Twitter Social Media Public Opinion COVID-19 Vaccines SARS-CoV-2 Vaccination Vaccination Refusal Vaccine Hesitancy Natural Language Processing Sentiment Analysis Topic Modeling Demographic Inference ==== Body pmc1 Introduction With the global continuation of the COVID-19 pandemic, the large-scale administration of a SARS-CoV-2 vaccine (referred from here on as the COVID-19 vaccine) is crucial to achieve herd immunity and curtail further spread of the virus1. As governments work to approve and distribute safe and effective vaccines,2 important questions regarding vaccination willingness persist: What are the attitudes and perceptions of the public3 to these vaccines and how can they affect vaccine uptake4? These questions are important to develop an education and outreach approach to achieve the desired vaccine penetration to achieve herd immunity5. In 2019, prior to the COVID-19 pandemic, the World Health Organization (WHO) had identified vaccine hesitancy as one of the top 10 greatest global health threats6. While surveys on attitudes and perception of a COVID-19 vaccine show significant vaccine hesitancy among the general population[7], [8], [9] and health care providers[10], [11], studies remain small in size, tend to focus on local participants, are prone to sampling error from non-probability sampling and reporting bias, and perhaps most poignantly, cannot capture real-time changes in vaccine willingness. Crowdfunding platforms may provide an indication of emerging community needs related to COVID-19 but fail to provide a continuous assessment of community sentiment12. Twitter, the microblogging platform, with over 187 million daily monetizable active users,13 serves as a robust medium to better understand wide-scale, organic public perception about the COVID-19 vaccine. With nearly 400 million mentions, #COVID19 was the most used hashtag on Twitter in 202014. Social media has become increasingly recognized for its rapid information dissemination (whether accurate or not) and dispersion of sentiment that quickly crosses geographic and social boundaries15. Analysis of social media text can inform real-time changes and evolution in population-level attitudes[16], [17]. As evident with the rise of the “infodemic” during the COVID-19 pandemic, Twitter has become a particularly useful data source in public health and healthcare-related research18 and has been repeatedly used to study public sentiment and understand trends throughout the COVID-19 pandemic[19], [20], [21], [22], [23], [24]. Earlier in the COVID-19 pandemic, we were able to demonstrate initial public sentiment regarding the virus, its origin and spread, and measures to limit its spread25 as well as early support for social distancing26 on Twitter. Social media, and specifically Twitter, has been shown to be a major factor in vaccine uptake and should be monitored and potentially used for interventions to address vaccine hesitancy27. Examining sentiments towards the influenza A H1N1 vaccine in 2009 showed that projected vaccination rates based on Twitter sentiment were similar to vaccination rates estimated by traditional phone surveys used by the Centers for Disease Control and Prevention (CDC)28. A previous study noted information exposure on Twitter may account for differences in human papillomavirus (HPV) vaccine uptake that are not accounted for by socioeconomic factors like education, insurance, or poverty29. Another study noted that there is a significant relationship between social media use by the public and organized action and public doubts of vaccine safety30. Content analysis and themes around COVID-19 vaccine hesitancy were evaluated in a subset of Canadian tweets31 and in tweets following the announcement of successful vaccine trials32. Multiple studies thus far have shown the importance and value of analyzing social media to understand public sentiment and discussion about COVID-19 vaccination. One study explored the network map of worldwide Facebook users to examine the relationships and growth in anti-vaccination views and relates these theoretically to COVID-19 vaccination33. In a study from China, analysis of 1.75 million Weibo messages between January and October 2020 found positive trends in COVID-19 vaccine acceptance, but revealed areas of misinformation34. One study from Australia collected 31,100 English tweets related to COVID-19 vaccine and identified sentiment, emotion, and topics35. In a study in the United Kingdom and the United States, analysis of 300,000 Twitter and Facebook posts showed sentiments that correlated with national surveys36. We aimed to apply content and sentiment analysis on COVID-19 vaccine related tweets as well as analysis of the responsible, originating user accounts to provide insight into the evolution of public attitudes about the COVID-19 vaccines over time. We hypothesized that content analysis from the start of the pandemic will identify important themes of discussion (especially those with negative sentiment or evidence of misinformation) throughout the vaccine development process that would inform health care officials, public health agencies, and policy makers and could be used to aid in the outreach and educational interventions for the COVID-19 vaccine to the general public. 2 Methods 2.1 Data Source We performed a cross-sectional observational study of English-language tweets obtained by matching the keywords ((‘covid’ OR ‘coronavirus’) AND ‘vaccine’) from February 1, 2020 to December 11th, 2020. December 11th was chosen as an end date to mark the United States Food and Drug Administration’s first emergency use authorization of a COVID-19 vaccine37. We used the snscrape library38 to obtain (“scrape”) tweets identified through Twitter's advanced search tool, which returns a relevant sample of tweets. We manually reviewed a random subsample of 1,000 tweets and verified the tweets’ relevance to the topic of COVID-19 vaccination. We extracted 21 and 20 variables related to the tweets and to the posting user accounts, respectively (Supplemental Table S1 and S2). 2.2 Data Processing We measured total daily tweets and completed descriptive statistics for collected variables. We applied natural language processing techniques to process, analyze, and visualize the text from tweets. To preprocess the tweet text for analysis (“cleaning”), we removed hyperlinks, user tags, and words of little analytical value. We also returned words to their root form and segmented text into one- and two-word terms. Further details are discussed in Supplemental Appendix A. We visualized the top 300 processed terms as a word cloud with larger font size representing greater term frequency. All analyses were conducted using Python, version 3.8.2 (Python Software foundation). Institutional review board approval was not required because this study used only publicly available data. 2.3 Sentiment Analysis Sentiment analysis describes the affect of a piece of text — the intrinsic attractiveness or aversiveness of a subject such as events, objects, or situations39. We used the Valence Aware Dictionary and sEntiment Reasoner (VADER)40 to analyze the sentiment polarity of a tweet. VADER, a lexicon and rule-based tool, was particularly designed for sentiment analysis for social media text. In addition to regular words, VADER leverages punctuation, emoticons, emojis, sentiment-laden slang words and acronyms, as well as syntax and capitalization schemas to inform labeling of a positive, neutral, and negative score for each document. These three scores were combined to form a normalized, weighted composite score. Overall positive (≥0.05), neutral (-0.05 to 0.05), and negative (≤-0.05) sentiments are defined at standardized composite score thresholds. When sentiment has been aggregated, we refer to an average sentiment of ≥0.05 as positive and ≤-0.05 as negative. Trends in sentiment over time were determined using the Mann-Kendall trend test. We also manually labeled sentiment for a random sample of 1000 tweets to evaluate VADER performance. VADER had a weighted average F1 score of 0.77. The negative sentiment had precision of 0.79 and recall of 0.65. We used the TextBlob library41 to label each tweet from a range of 0 (objective) to 1 (subjective) where objective tweets relay factual information and subjective tweets typically communicate an opinion or belief. Finally, we used the NRCLex library to label words within each tweet with corresponding emotional affects (i.e., Plutchik’s wheel of emotions which include anger, anticipation, fear, disgust, joy, sadness, surprise, and trust) based on the National Research Council Canada (NRC) affect lexicon42. This wheel of emotions was used since they can be naturally paired into opposites (i.e., joy-sadness, anger-fear, trust-disgust, anticipation-surprise). Based on these labels, we identified tweets with their primary emotion and visualized how the proportion of eligible tweets (i.e., those with an identified primary emotion) with a particular primary emotion changed over time. 2.4 Topic Modeling After cleaning the tweets to distill analyzable text as described in the methods, we applied a machine learning algorithm called Correlation Explanation (CorEx)43 to identify clusters of topics for all tweets. CorEx identifies the most informative topics based on a set of latent factors that best explain the correlations in the data in turn maximizing the total correlation or the multivariate mutual information44. We used CorEx as opposite to a generative model like Latent Dirichlet Allocation to avoid making assumptions and specifications of hyperparameters. Each document (in our case, each tweet) may include multiple topics. We iterated through 2 to 20 clusters by analyzing the distribution of topic coherence (TC) for each topic and evaluating how much each additional topic enhances the overall TC. The number of topics was increased until further topics no longer made a substantial contribution to the overall TC. This process is analogous to selecting a cutoff eigenvalue when conducting topic modeling using Latent Semantic Analysis (LSA). That resulted in 15 topics for the topic model. We presented the top 20 words for each topic cluster to author CUL without prior access to individual tweets from the dataset to manually label a theme for each topic. The manually labeled topic labels were reviewed by two other authors SNS and RJM with unanimous agreement. We visualized the monthly distribution of topics over time and utilized a heat map to visualize how the mean sentiment of each topic has changed per month. 2.5 User Exploration and Demographic Inference Given that each tweet has one authoring account, we identified all unique user accounts in our dataset and provided descriptive statistics with metadata available for the users, including the launch date of the account, followers (accounts following them), follows (accounts they follow), lifetime posts, likes, and media shared, as well as profile pictures, description information, and verified status (badge to indicate an account of public interest that has been verified to be authentic). To better understand demographic differences, we applied a previously validated deep learning system through the m3inference library45 to infer the account user as an individual or an organization based on multimodal input that includes username, display name, description, and profile picture image. If the account is labeled as an individual, the gender (female or male) and age group (≤18 years old, 19-29 years old, 30-39 years old, and ≥40 years old) are then labeled. Each label using the algorithm has an accompanying probability. The automatic demographic detection was particularly designed for Twitter profiles for health-related cohort studies46. We provided summary statistics for the demographics identified and stratified sentiment and subjectivity analyses by the different demographic groups to evaluate for differences. We used Mann-Whitney U and χ2 where appropriate to determine significance. Alpha level of significance was set a priori at 0.05 and all hypothesis testing was two-sided. We did not adjust for multiple comparisons as this was an exploratory study and should be interpreted as hypothesis-generating. 3 Results A total of 2,356,285 tweets were extracted for the study period, of which 2,287,344 tweets were English-only and included for evaluation. The tweets were generated by 948,666 accounts which had been active for an average of 6.9 years (interquartile range [IQR], 2.6 - 10.0) with a median of 267 (IQR, 55 - 1,100) followers and 3,600 (519 - 15,572) lifetime likes. Only 2.9% (n=27,443) of accounts were verified (Table 1 ). Of the tweets analyzed, 54% (n=1,235,575) had a link, 40.1% (n=916,585) mentioned other twitter accounts, 18.1% (n=414,173) used hashtags, and 11.9% (n=273,278) contained media like an image or video. In terms of engagement, 41.3% (n=943,639), 24.0% (n=548,863), and 20.7% (n=473,204) of tweets received likes, replies, and retweets, respectively (Table 1). Individuals (vs. organizations) generated 87.9% (n=834,224) of tweets. Of individuals, men (n=560,824) outnumbered women (n=273,400) by 2:1 and 39.5% (n = 329,776) of individuals were ≥40 years old (Table 1).Table 1 Tweet and user account characteristics are shown on top and inferred user demographics are shown on bottom. Tweetsn = 2,287,344 User AccountsN = 948,666 Has link(s) 1,235,575 (54.0) Years account active 6.9 (2.6 – 10.0) Mentions user(s) 916,585 (40.1) Followers 267 (55 – 1,100) Has hashtag(s) 414,173 (18.1) Following 407 (137 – 1,069) Has media 273,278 (11.9) Lifetime statuses 4,605 (1,027 – 16,365) Is quoted tweet 133,404 (5.8) Lifetime likes 3,600 (519 – 15,572) Has like 943,639 (41.3) Media shared 205 (36 – 875) Has reply 548,863 (24.0) Public lists, member 2 (0 – 12) Has retweet 473,204 (20.7) Contains description 800,619 (84.4) Has quoted tweet 183,982 (8.0) Location listed 659,720 (69.5) Twitter source Contains profile picture 902,666 (95.2) Web App 686,296 (30.0) Contains banner picture 721,542 (76.1) iPhone/iPad 660,382 (28.9) Contains link in profile 303,761 (32.0) Android 432,862 (18.9) Verified account 27,443 (2.9) TweetDeck 60,845 (2.7) User DemographicsN = 948,666 N (%) Probabilitymedian (IQR) Entity Individual 834,224 (87.9) 0.999 (0.997 – 0.999) Organization 114,443 (12.1) 0.867 (0.727 – 0.999) Sex (of individuals) Female 273,400 (32.8) 0.992 (0.949 – 0.998) Male 560,824 (67.2) 0.996 (0.980 – 0.999) Age (of individuals) <40 years old 504,448 (60.5) 0.972 (0.896 – 0.994) ≤18 years old 109,327 (13.1) 0.660 (0.528 – 0.821) 19 to 29 years old 225,360 (27.0) 0.611 (50.4 – 0.754) 30 to 39 years old 169,761 (20.4) 0.765 (0.568 – 0.936) ≥40 years old 329,776 (39.5) 0.950 (0.734 – 0.996) Daily tweets abruptly spiked to 51,176 tweets on November 9th, the day Pfizer and BioNTech announced their vaccine’s effectiveness47 (up from 4,052 tweets on November 8th) and peaked on December 8th with 55,779 tweets. Tweets from November 1st to the end of the study period on December 11th accounted for 39.8% (n = 910,593) of all tweets (Figure S1). The corpus of tweets contained over 62.5 million words and 416 million characters. The ten most commonly tweeted terms and their frequencies were as follows: “people” (228,482), “trial” (206,310), “take” (181,598), “flu” (159,043), “trump” (149,042), “first” (147,103), “make” (142,242), “test” (131,719), “need” (126,846), and “one” (122,966). Figure 1 displays a word cloud of the top 300 words with larger font size concordant with frequency.Figure 1 Word cloud of top 300 words related to COVID-19 and vaccine. Larger fonts represent higher frequency in the corpus after preprocessing text. Daily mean sentiment of tweets fluctuated congruent with news events, but overall trended positively throughout the study period (Mann-Kendall statistic=10,122; tau=0.218; p<0.001) (Figure 2 a). Several days in early to mid-March and on October 13th saw particularly negative sentiments, coinciding with news of the declaration of a pandemic by the WHO and Johnson & Johnson’s halting of their vaccine trial on October 12th48, respectively. Highest daily mean positive sentiment revolved around Moderna’s July 14th announcement of a safe vaccine with “robust immune response” in an early trial49 and Pfizer’s November 9th announcement of over 90% effectiveness of its vaccine47. Twitter accounts representing organizations had more positive sentiments than tweets from individuals (median weekly difference, 0.118; IQR, 0.091 to 0.144), but there was no significant difference in polarity for age (median weekly difference, 0.006; IQR, -0.011 to 0.019) and only minimal positive difference for males (median weekly difference, 0.030; IQR, 0.012 to 0.044) (Figure 2b-d).Figure 2 a-d. a) Mean sentiment polarity shown by day (as points) and by week (as a dashed line). Each tweet was labeled as primarily negative (-1), neutral (0), or positive (1). b) Mean weekly polarity stratified by individual versus organization. c) Mean weekly polarity stratified by gender for individual accounts. d) Mean weekly polarity stratified by age more or less than 40 years than for individual accounts. The sentiment trends were reflected by the primary emotions identified in the COVID-19 vaccine tweets by month (Figure 3 a). Fear started as the most prevalent primary emotion in nearly 40% of eligible tweets early on but decreased to under 20% by the end of the study period. Conversely, trust increased from below 20% to around 40% and outpaced fear in April 2020, maintaining as the most prevalent primary emotion thereafter. Anticipation was the second most prevalent primary emotion for most of the study period, steadily ranging from 25% to 30%. All other emotions were consistently expressed as the predominant emotion in less than 10% of eligible tweets. Individuals had an increased predominance of fear (26.3% vs. 19.4%; p<0.001) and decreased predominance of anticipation (25.9% vs. 33.6%; p<0.001) and trust (32.5% vs. 35.2%; p<0.001). For individual accounts, women had more fear (28.4% vs. males 25.4%; p<0.001) with less anticipation (23.8% vs. 26.8%; p<0.001) than men, but no significant difference in trust (32.3% vs. 32.5%, p=0.11). Those less than 40 years old had more fear (26.6% vs. 26.0%; p<0.001) and less trust (32.0% vs. 33.0%; p<0.001) (Figure 3b-d). Tweets throughout the year tended to be more objective (where 0 is fully objective and 1 as fully subjective) with limited daily variation (overall mean 0.359; std 0.028) (Figure S2).Figure 3 a-d. Percent of tweets with primary emotion per month a) overall, b) stratified by individual versus organization, c) stratified by gender for individual accounts, and d) stratified by age more or less than 40 years than for individual accounts. Only tweets with a predominant primary emotion (n = 1,489,027) are included. Table 2 shows each topic label with their key words and sample tweets. Figure S3 shows the 15 topics obtained from topic modeling with the proportion of tweets per month that contained each topic. The dominant topic (topic 15) focused on mask use and public reactions. Discussions about misinformation and conspiracy theories comprised the next most common topic, peaking in May and staying relatively consistent from July through December. Tweets related to the Indian and Russian governments’ decision on producing and using the Sputnik V vaccine (topic 2) spiked in August. Discussion of Emergency Use Authorizations (EUA) and vaccine approvals (topic 12) did not spike until November 2020 with the approval of the Pfizer and Moderna vaccines. Several topics had strong mean positive sentiments throughout the study period, including discussions of biotechnology companies and the stock market (topic 3), vaccination firsts (topic 4), vaccine development (topic 6), and EUAs (topic 12). Other topics showed a progressive trend from positive to negative throughout the study period including discussion of US politics and the election (topic 1), the FDA and CDC (topic 14), and mask use and public reactions (topic 15). Tweets comparing COVID-19 to influenza (topic 5) and its vaccine had strongly negative early sentiment but improved over time (Figure 4 ). Compared to the rest of individual users (n=810,318), those exhibiting negative sentiment posting about topic 5 (n=51,686) were proportionally more likely to be ≥40 years old (45.1% vs. 39.6%; p <0.001) and female (34.0% vs. 32.7%; p <0.001). The only other topic with persistently negative sentiment was discussion of misinformation and conspiracy theories (topic 13). Those exhibiting negative sentiment posting about topic 13 (n=166,819) were more likely than other user accounts (n=741,388) to be individuals (90.9% vs. 87.3%; p <0.001) and of those individual accounts, more likely to be female (34.4% vs. 32.4%; p<0.001).Table 2 Topic clusters identified by topic modeling. Words contributing to the model are shown in decreasing order of weighting. The topics are labeled manually based on these words. Possible Topic Label Topic # Tweets/Topic Words contributing to topic model(in ↓ order of weighting) Representative Tweet Mask use and general reactions 15 811,844 people, mask, even, dont, would, take, know, die, death, need, one, still, many, kill, risk, never, work, way, yet, wear_mask ““Pretty much what it boils down to, at this point. Ignorance, arrogance, and stupidity will end up killing LOTS of people this year, I'm afraid! Be SMART. WEAR your mask. Wash your hands. Hold off on large gatherings until a safe, effective Covid-19 vaccine arrives. [link]” Conspiracies and misinformation 13 557,301 want, think, fake, make, try, believe, conspiracy, bill_gate, money, really, gonna, real, force, shit, god, anything, anyone, hoax, black, put “@WhiteHouse Also, isnt this a RNA vaccine? Super experimental albeit dangerous, could mean with DNA as well. Human Guinea pigs. Wouldn't be surprised if the vaccine harms more then the COVID did.” Impact on lockdowns on school, work, and economy 7 406,227 wait, year, lockdown, open, month, next, life, time, long, end, last, come, back, next year, week, school, ago, day, economy, away “@[tag] @[tag] @[tag] @[tag] And even with a vaccine they will continue with the lockdowns, the social distance and the fear mongering... If not for the Covid, they will find something...” Vaccine mechanisms and immunity 10 235,625 virus, mrna, immunity, antibody, prevent, infection, disease, spread, strain, protein, herd_immunity, mutate, symptom, immune_system, sarscov, prevent infection, mutation, cell, cause, infect “If tests show one already had COVID-19 so one has antibodies and is now immune, CDC currently counts that as one infected and positive for COVID-19. After a vaccine, will every person vaccinated who therefore grows antibodies, be considered positive & infected? @realDonaldTrump” First vaccinations and recipients 4 232,613 first, world, around, world first, become, first test, receive, country, first line, first person, world news, first country, person receive, world leader, government vote, make sure_pass, yearold_woman, day government, first dos, first world “Thank the lord this is the beginning of the end: First patient receives Pfizer Covid-19 vaccine [link]” Emergency use authorizations and approvals 12 213,036 pfizer, moderna, approval, effective, pfizer_biontech, emergency_use, data, biontech, authorization, receive, regulator, next_week, approve pfizer, effective prevent, pfizer ceo, show effective, moderna effective, approve, data show, early data ““Pfizer's Covid vaccine is days away from approval after data reveals it is 95% effective [link]” US politics and election 1 212,597 trump, biden, realdonaldtrump, president, election, operation_warp, american, credit, speed, lie, take credit, joebiden, gop, democrat, potus, win, vote, joe_biden, admin, america “@realDonaldTrump If you want to take partial credit for the Covid-19 vaccine fine. You still LOST the election. In Georgia for example you are behind there by 12k votes. The recount wont change the outcome. I look forward to your predictable reply and the end of your regime.” Stock market and pharma/biotech companies 3 205,557 market, stock, news, company, good news, biotech, drug company, pharma, price, billion, drug, surge, late, update, rise, break news, positive news, investor, pharma company, announce “Markets are supported by both the cumulative upside surprises to the economy since the end of the recession and the apparently faster-than-expected progress toward a COVID-19 vaccine. [link]” Clinical trials and participants 8 205,468 trial, clinical, human trial, phase, human, volunteer, participant, oxford, begin, phase clinical, show, trial participant, result, volunteer trial, ahead_large trial, ahead_large, show_promise, immune_response, test, number “Coronavirus Vaccine Update \| Oxford’s COVID-19 vaccine trial in Brazil begins: Scientists say coronavirus jab may not work for older adults [link]” Vaccine development and supporters 6 204,083 research, development, global, effort, develop, fund, researcher, global effort, join, effort develop, access, help, target, hacker, treatment, support, dolly_parton, research development, accelerate, collaboration “As the world continues to feel the impact of COVID-19, the biopharmaceutical industry is working around the clock to identify and develop safe and effective vaccines to prevent infection, while also researching and developing new therapies to treat those infected with the virus.” Russian response and global partners 2 187,369 russia, india, via, sputnik, russian, china, putin, serum_institute, indian, covaxin, chinese, bharat_biotech, icmr, hacker_target, time india, via nbcnews, russia sputnik, indias_serum, narendramodi, possible “A Sputnik moment, president #Putin has announced that #Russia is the first country in the world to register a #Covid_19 vaccine. 10s of countries already requested it [link]” FDA and CDC 14 179,396 trump, fda, failure, administration, trump administration, fda approval, food_drug, food_drug administration, fda approve, cdc, trump admin, white_house, trump claim, president_donald trump, president_donald, cuomo, trump supporter, want, take, exist_sustainable “'@CDCgov if you try and push through an unproven vaccine because of Trump’s desperation to recover from his abysmal handling of Covid-19...good luck. No one I know myself included will be getting vaccinated.” Philanthropy and public health 9 156,183 health, public, mandatory, public health, health official, health care, public trust, official, gavi_sdg, cdc_gatesfoundation, read_billgates cdc_gatesfoundation, gavi_sdg vaccination, make mandatory, read_billgates, health expert, care_worker, health minister, health care_worker, clinton, obama_bush “@[tag] This presents a problem and crashes into the argument, should covid vaccines be mandated. I initially thought that it will need more then encouragement and common sense from the public but these vaccine deniers are going to deprive people of protection through fear. Arrest them.” Comparison to influenza 5 151,177 flu, shot, influenza, flu shot, every year, season, seasonal, die flu, every, flu death, year flu, kill, jab, take flu, first shot, virus, people, side_effect shot, compare, via “'Ok so I’m usually not super crunchy about everything but I’ve been hospitalized 2x after getting the flu shot bc of how badly I got the flu within months so I was told not to get the shot by my drs. what does that mean for COVID’s vaccine? Like what if I react the same?” Safety and side effects from trials 11 137,445 astrazeneca, safety, effect, johnson johnson, pause, study, unexplained_illness, astrazeneca trial, johnson_pause, long_term effect, oxford_university, pause trial, safety efficacy, efficacy, safety concern, resume, put hold, astrazeneca study, illness, side 'AstraZeneca COVID-19 vaccine study put on hold due to suspected adverse reaction in UK participant [link]” Figure 4 Heat map showing mean sentiment by month for each topic. Note that a tweet can include multiple topics. 4 Discussion Twitter is a rich medium that can serve as both thermometer and thermostat for the COVID-19 vaccine, which is a crucial public health strategy to combat the pandemic. It can provide insight into public perception of a COVID-19 vaccine, but can also be used to understand and combat knowledge deficits and vaccine hesitancy through information and education[50], [51]. A majority (59%) of US Twitter users regularly obtain news on Twitter, proportionally more than any other social media platform52. We analyzed nearly 2.3 million COVID-19 vaccine-related tweets in 2020, creating a dataset that exceeded the scope of related studies[53], [54] and is the largest study to date of social media posts about COVID-19 vaccination at the time of this manuscript. We evaluated public perception as the COVID-19 vaccine development went from speculative to theoretical to actual. Generally, we believe that Twitter users favored the vaccine during its development phase. Tweets with positive sentiment were more prominent than tweets with negative sentiment and trust emerged as the predominant emotion. However, there were periods of time (usually linked to events in the public news cycle), demographic subgroups, and topic clusters that had more prominent negative sentiment and emotion. Organizational accounts were significantly more positive, exhibiting more anticipation and trust and less fear. For individuals, the gender and age distribution in our dataset parallels the reported proportional share of Twitter’s global advertising audience55. Women expressed more fear and less anticipation, but by the end of the study period, that gap had narrowed. Those less than 40 years old tended to express less trust and more fear, but the margin was small. The topic most strongly associated with negative, albeit improving, sentiment was the discussion of the influenza vaccine in combination with the COVID-19 vaccine. These tweets often compared deaths and illness from both diseases or expressed general vaccine mistrust to both vaccines. Examples include: “@[user] Only time I've ever had the flu is the 2 times I got flu shots. It was not a minor case either it was the full blown flu. I refuse to get another flu shot and I also will refuse the covid vaccine” and “Flu Virus equals Flu Vaccine. Coronavirus Equals Covid-19 Vaccine...Now if the Flu shot gives you the flu, the Covid-19 Shot will give you Coronavirus....am I in the general area of Right??”. Notably, these users exhibiting negative sentiment about this topic were more likely to be ≥40 years old and female. This focused topic-demographic cluster, for example, exposes a direct opportunity for intervention to correct misinformation and mitigate vaccine hesitancy. Conversely, the emergency use authorizations of the vaccine and reports of the first vaccine recipients, which arose later in the study period, were celebrated with positive sentiment and mirror the overall increasing trend in positive sentiment and trust. While the percent population immunity needed to achieve herd immunity (either through innate or acquired immunity) for COVID-19 is not yet known, estimates have increased from 60-70% to possibly closer to 75-85%[56], [57]. Achieving herd immunity through infection would come at an untenable cost58, making the immunization effort critical to protect lives. Therefore, it was concerning to us that fear was a common and persistent predominant emotion in COVID-19 vaccine tweets. While the proportion of ‘trust’ tweets outpaced ‘fear’ tweets relatively early in the study period, approximately 20% of eligible tweets still expressed fear in association with the vaccine. If this fear translates into refusal to become immunized, we are not only likely to see a prolonged pandemic, but also further increases in COVID-19 related deaths as concerning virus variants take hold. As more people receive the vaccine in the future, we anticipate that sentiments will become more positive over time with increased trust and vaccine uptake, but this will need to be consistently studied, especially in the context of newly approved vaccines and news events. 4.1 Limitations Our study was limited by several factors. First, we recognize that our dataset is not all inclusive of tweets discussing the COVID-19 vaccine. Our tweet search criterion was narrow to ensure accuracy of captured tweets for this initial work and did not include terms such as “shot(s)”, “immunization” and “inoculation.” Moreover, despite the volume of tweets analyzed, we are limited to only a relevant sample of all tweets per Twitter’s advanced search tool. Second, we used existing tools to analyze sentiments and emotion of tweets that are not specific to health care topics, which could have skewed our analysis. Third, tweets related to COVID-19 vaccination could have been flagged or removed by Twitter for containing misinformation, but we were not privy to that context to determine how that could have affected our sample. Finally, since we targeted only tweets in English and are unable to determine geographic location for users, we are limited in making conclusions about specific countries or countries where English is not the predominant language. 5 Conclusions Leveraging 2.3 million COVID-19 vaccine related tweets in 2020, we were able to successfully explore sentiment, emotion, topics, and user demographics to elucidate important trends in public perception about the COVID-19 vaccine. Tweets were overall positive in sentiment and with growing trust. However, fear maintained as a dominant emotion raising concern regarding the willingness to receive the COVID-19 vaccine and subsets of negative sentiment emerged. Comparison to influenza and the influenza vaccine as well as discussion about conspiracy theories were important topics with negative sentiment and showed some demographic differences that could allow for informed intervention. Future work will leverage these natural language processing tools to engage in targeted messaging based on user interests and emotions. 6 Declarations Ethics approval and consent to participate: The University of Texas Southwestern Human Research Protection Program Policies, Procedures, and Guidance did not require institutional review board approval as all data were publicly available. Data Availability, Materials, & Correspondence: The data that support the findings of this study are available upon request. Please request from the corresponding author. Code Availability: The code that support the findings of this study is available upon request. Please request from the corresponding author. Conflict of interest(s): Dr. Lehmann reports stock ownership in Celanese Corporation and Colfax Corporation. There are no other competing interests. Funding: None. Authors’ contributions: Study concept and methodology/design: SNS, CUL, RJM; Data curation: SNS, SK; Analysis: SNS, SK, RJM; Interpretation of data: SNS, SK, CUL, RJM; Manuscript preparation: all authors; Manuscript reviewing and editing: all authors. All authors read and approved the final manuscript. SNS, SK, and RJM have accessed and verified the underlying data. Acknowledgements: Not applicable Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Christoph Lehmann reports a relationship with Celanese Corporation that includes:. Christoph Lehmann reports a relationship with Colfax Corp that includes. Data availability Data will be made available on request. ==== Refs References 1 Sridhar D. Gurdasani D. Herd immunity by infection is not an option Science 371 6526 2021 230 231 33446540 2 Forni G. Mantovani A. Forni, G., Mantovani, A., & COVID-19 Commission of Accademia Nazionale dei Lincei, Rome. COVID-19 vaccines Cell Death Differ 28 2 2021 626 639 33479399 3 Caserotti M. Girardi P. Rubaltelli E. Tasso A. Lotto L. Gavaruzzi T. Associations of COVID-19 risk perception with vaccine hesitancy over time for Italian residents Soc Sci Med 272 2021 113688 33485215 4 Wong M.C.S. Wong E.L.Y. Huang J. Cheung A.W.L. Law K. Chong M.K.C. Acceptance of the COVID-19 vaccine based on the health belief model: A population-based survey in Hong Kong Vaccine 39 7 2021 1148 1156 33461834 5 Feleszko W. Lewulis P. Czarnecki A. Waszkiewicz P. Flattening the Curve of COVID-19 Vaccine Rejection-An International Overview Vaccines (Basel) 9 1 2021 44 33451104 6 Ten threats to global health in 2019. https://www.who.int/news-room/spotlight/ten-threats-to-global-health-in-2019. 7 Alley S.J. Stanton R. Browne M. To Q.G. Khalesi S. Williams S.L. As the Pandemic Progresses, How Does Willingness to Vaccinate against COVID-19 Evolve? Int J Environ Res Public Health 18 2 2021 797 33477825 8 Reiter P.L. Pennell M.L. Katz M.L. Acceptability of a COVID-19 vaccine among adults in the United States: How many people would get vaccinated? Vaccine 38 2020 6500 6507 32863069 9 Malik A.A. McFadden S.M. Elharake J. Omer S.B. Determinants of COVID-19 vaccine acceptance in the US EClinicalMedicine 26 2020 100495 32838242 10 Shaw J. Assessment of U.S. health care personnel (HCP) attitudes towards COVID-19 vaccination in a large university health care system Clin Infect Dis 2021 10.1093/cid/ciab054 11 Verger P. Attitudes of healthcare workers towards COVID-19 vaccination: a survey in France and French-speaking parts of Belgium and Canada, 2020 Euro Surveill 26 2021 12 Saleh S.N. Lehmann C.U. Medford R.J. Early Crowdfunding Response to the COVID-19 Pandemic: Cross-sectional Study J Med Internet Res 23 2 2021 e25429 33523826 13 Q3 2020 Letter to Shareholders. https://s22.q4cdn.com/826641620/files/doc_financials/2020/q3/Q3-2020-Shareholder-Letter.pdf. 14 McGraw, T. Spending 2020 Together on Twitter. https://blog.twitter.com/en_us/topics/insights/2020/spending-2020-together-on-twitter.html. 15 Stieglitz S. Dang-Xuan L. Emotions and Information Diffusion in Social Media—Sentiment of Microblogs and Sharing Behavior Journal of Management Information Systems 29 4 2013 217 248 16 Depoux A. Martin S. Karafillakis E. Preet R. Wilder-Smith A. Larson H. The pandemic of social media panic travels faster than the COVID-19 outbreak Journal of Travel Medicine 27 3 2020 taaa031 32125413 17 Wilson A.E. Lehmann C.U. Saleh S.N. Hanna J. Medford R.J. Social media: A new tool for outbreak surveillance ASHE 1 2021 e50 36168466 18 Sinnenberg L. Buttenheim A.M. Padrez K. Mancheno C. Ungar L. Merchant R.M. Twitter as a Tool for Health Research: A Systematic Review Am J Public Health 107 1 2017 e1 e8 19 Chang C.-H. Monselise M. Yang C.C. What Are People Concerned About During the Pandemic? Detecting Evolving Topics about COVID-19 from Twitter J Healthc Inform Res 5 1 2021 70 97 10.1007/s41666-020-00083-3 33490856 20 Chandrasekaran R. Mehta V. Valkunde T. Moustakas E. Topics, Trends, and Sentiments of Tweets About the COVID-19 Pandemic: Temporal Infoveillance Study Journal of Medical Internet Research 22 10 2020 e22624 33006937 21 Xue J. Chen J. Hu R. Chen C. Zheng C. Su Y. Twitter Discussions and Emotions About the COVID-19 Pandemic: Machine Learning Approach Journal of Medical Internet Research 22 11 2020 e20550 33119535 22 Gallotti R. Valle F. Castaldo N. Sacco P. De Domenico M. Assessing the risks of ‘infodemics’ in response to COVID-19 epidemics Nature Human Behaviour 4 2020 1285 1293 23 Lanier H.D. Diaz M.I. Saleh S.N. Lehmann C.U. Medford R.J. De Silva D. Analyzing COVID-19 disinformation on Twitter using the hashtags #scamdemic and #plandemic: Retrospective study PLoS ONE 17 6 2022 e0268409 35731785 24 Diaz M.I. Hanna J.J. Hughes A.E. Lehmann C.U. Medford R.J. The Politicization of Ivermectin Tweets During the COVID-19 Pandemic. Open Forum Infectious Diseases 9 7 2022 ofac263 25 Medford R.J. Saleh S.N. Sumarsono A. Perl T.M. Lehmann C.U. An ‘Infodemic’: Leveraging High-Volume Twitter Data to Understand Early Public Sentiment for the Coronavirus Disease 2019 Outbreak. Open Forum Infect Dis 7 2020 ofaa258 26 Saleh S.N. Lehmann C.U. McDonald S.A. Basit M.A. Medford R.J. Understanding public perception of coronavirus disease 2019 (COVID-19) social distancing on Twitter Infect Control Hosp Epidemiol 42 2 2021 131 138 32758315 27 Systematic scoping review on social media monitoring methods and interventions relating to vaccine hesitancy. https://www.ecdc.europa.eu/en/publications-data/systematic-scoping-review-social-media-monitoring-methods-and-interventions (2020). 28 Salathé M. Khandelwal S. Meyers L.A. Assessing Vaccination Sentiments with Online Social Media: Implications for Infectious Disease Dynamics and Control PLOS Computational Biology 7 10 2011 e1002199 22022249 29 Dunn A.G. Surian D. Leask J. Dey A. Mandl K.D. Coiera E. Mapping information exposure on social media to explain differences in HPV vaccine coverage in the United States Vaccine 35 23 2017 3033 3040 28461067 30 Wilson S.L. Wiysonge C. Social media and vaccine hesitancy BMJ Global Health 5 10 2020 e004206 31 Griffith J. Marani H. Monkman H. COVID-19 Vaccine Hesitancy in Canada: Content Analysis of Tweets Using the Theoretical Domains Framework J Med Internet Res 23 4 2021 e26874 33769946 32 Boucher J.-C. Cornelson K. Benham J.L. Fullerton M.M. Tang T. Constantinescu C. Analyzing Social Media to Explore the Attitudes and Behaviors Following the Announcement of Successful COVID-19 Vaccine Trials: Infodemiology Study JMIR Infodemiology 1 1 2021 e28800 34447924 33 Johnson N.F. Velásquez N. Restrepo N.J. Leahy R. Gabriel N. El Oud S. The online competition between pro- and anti-vaccination views Nature 582 7811 2020 230 233 32499650 34 Yin F. Wu Z. Xia X. Ji M. Wang Y. Hu Z. Unfolding the Determinants of COVID-19 Vaccine Acceptance in China J Med Internet Res 23 1 2021 e26089 33400682 35 Kwok S.W.H. Vadde S.K. Wang G. Twitter Speaks: An Analysis of Australian Twitter Users’ Topics and Sentiments About COVID-19 Vaccination Using Machine Learning J Med Internet Res 2021 10.2196/26953 36 Hussain A. Artificial intelligence-enabled analysis of UK and US public attitudes on Facebook and Twitter towards COVID-19 vaccinations J Med Internet Res 2021 10.2196/26627 37 FDA Takes Key Action in Fight Against COVID-19 By Issuing Emergency Use Authorization for First COVID-19 Vaccine. https://www.fda.gov/news-events/press-announcements/fda-takes-key-action-fight-against-covid-19-issuing-emergency-use-authorization-first-covid-19. 38 Snscrape. (Github). 39 Frijda, N. H. The emotions. (Cambridge University Press ; Editions de la Maison des sciences de l’homme, 1986). 40 Hutto, C. & Gilbert, E. VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. in (2014). 41 Loria, S. Textblob: simplified text processing. (Textblob). 42 Mohammad, S. M. & Turney, P. D. Emotions evoked by common words and phrases: using mechanical turk to create an emotion lexicon. in Proceedings of the NAACL HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text 26–34 (Association for Computational Linguistics, 2010). 43 Gallagher, R. J., Reing, K., Kale, D. & Steeg, G. V. Anchored Correlation Explanation: Topic Modeling with Minimal Domain Knowledge. arXiv:1611.10277 [cs, math, stat] (2018). 44 Steeg, G. V. & Galstyan, A. Discovering Structure in High-Dimensional Data Through Correlation Explanation. arXiv:1406.1222 [cs, stat] (2014). 45 Wang, Z. et al. Demographic Inference and Representative Population Estimates from Multilingual Social Media Data. in The World Wide Web Conference 2056–2067 (ACM, 2019). doi:10.1145/3308558.3313684. 46 Yang, Y.-C., Al-Garadi, M. A., Love, J. S., Perrone, J. & Sarker, A. Automatic Gender Detection in Twitter Profiles for Health-related Cohort Studies. http://medrxiv.org/lookup/doi/10.1101/2021.01.06.21249350 (2021) doi:10.1101/2021.01.06.21249350. 47 Pfizer and Biontech announce Vaccine Candidate against Covid-19 Achieved Success in First Interim Analysis From Phase 3 Study. https://www.pfizer.com/news/press-release/press-release-detail/pfizer-and-biontech-announce-vaccine-candidate-against. 48 Staff A.J.M.C. A Timeline of COVID-19 Developments in 2020 2021 AJMC 49 Jackson L.A. Anderson E.J. Rouphael N.G. Roberts P.C. Makhene M. Coler R.N. An mRNA Vaccine against SARS-CoV-2 - Preliminary Report N Engl J Med 383 20 2020 1920 1931 32663912 50 Subbaraman N. This COVID-vaccine designer is tackling vaccine hesitancy - in churches and on Twitter Nature 590 7846 2021 377 33574597 51 Loomba S. de Figueiredo A. Piatek S.J. de Graaf K. Larson H.J. Measuring the impact of COVID-19 vaccine misinformation on vaccination intent in the UK and USA Nat Hum Behav 5 2021 337 348 33547453 52 Shearer, E. & Mitchell, A. News Use Across Social Media Platforms in 2020. https://www.journalism.org/2021/01/12/news-use-across-social-media-platforms-in-2020/ (2021). 53 Damiano A.D. Allen Catellier J.R. A Content Analysis of Coronavirus Tweets in the United States Just Prior to the Pandemic Declaration Cyberpsychol Behav Soc Netw 23 12 2020 889 893 32813572 54 Jang H. Rempel E. Roth D. Carenini G. Janjua N.Z. Tracking COVID-19 Discourse on Twitter in North America: Infodemiology Study Using Topic Modeling and Aspect-Based Sentiment Analysis J Med Internet Res 23 2 2021 e25431 33497352 55 Kemp, S. Digital 2020 Global Digital Overview. https://p.widencdn.net/1zybur/Digital2020Global_Report_en (2020). 56 Fontanet A. Cauchemez S. COVID-19 herd immunity: where are we? Nat Rev Immunol 20 2020 583 584 32908300 57 McNeil Jr., D. G. https://www.nytimes.com/2020/12/24/health/herd-immunity-covid-coronavirus.html. The New York Times (2021). 58 Randolph H.E. Barreiro L.B. Herd Immunity: Understanding COVID-19 Immunity 52 5 2020 737 741 32433946
PMC010xxxxxx/PMC10288321.txt	==== Front Learn Individ Differ Learn Individ Differ Learning and Individual Differences 1041-6080 1041-6080 Elsevier Inc. S1041-6080(23)00064-X 10.1016/j.lindif.2023.102320 102320 Article Keep going, keep growing: A longitudinal analysis of grit, posttraumatic growth, and life satisfaction in school students under COVID-19 Casali Nicole a Feraco Tommaso b⁎ Meneghetti Chiara b a Max Planck Institute for the Study of Crime, Security and Law, Freiburg, Germany b Department of General Psychology, University of Padova, Padua, Italy ⁎ Corresponding author at: via Venezia 8, Padova, Italy. 23 6 2023 23 6 2023 1023205 1 2023 17 6 2023 21 6 2023 © 2023 Elsevier Inc. All rights reserved. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The COVID-19 pandemic disrupted students' daily life, but grit could have sustained students' wellbeing by helping them work hard and stay goal-oriented over time despite adversity. Gritty students may also have interpreted COVID-19-related adversity as an opportunity to grow, thus displaying higher levels of post-traumatic growth. In this study, 445 students in grades 6–12 (160 males, Mage = 14.25, SDage = 2.11) completed measures of grit and life satisfaction at the beginning (Time 1) and at the end (Time 2) of the school year, together with a measure of posttraumatic growth. A longitudinal SEM model shows that perseverance positively relates to posttraumatic growth, indirectly favoring life satisfaction at Time 2. In conclusion, perseverance, rather than consistency, appeared to have sustained students' positive adjustment to the COVID-19 pandemic. Teaching students how to nurture this quality can have important beneficial effects for their wellbeing under adverse conditions. Keywords Grit Posttraumatic growth Life satisfaction COVID-19 Adolescents ==== Body pmc1 Educational relevance and implications of this research The present study investigated the impact of grit (perseverance of effort and consistency of interests) on well-being and posttraumatic growth following the COVID-19 pandemic. Our findings revealed a direct relationship between perseverance at the beginning of the school year and posttraumatic growth at the end of the school year. Moreover, we observed an indirect effect where perseverance also positively influenced life satisfaction. This suggests that fostering perseverance can help students face adversities, supporting their well-being and facilitating growth in challenging circumstances. Grit is a widely studied construct in educational psychology, as it seems to support several positive academic and nonacademic outcomes, remarkably academic achievement, self-regulated learning, and life satisfaction (Duckworth et al., 2007; Muenks et al., 2017), also in longitudinal studies which even better highlight the effect of this construct over time (Jiang et al., 2019; Postigo et al., 2021b; Tang et al., 2020). In the context of the COVID-19 pandemic, it may have supported students' life satisfaction and provided room for students to positively reappraise such a stressful situation (the so-called posttraumatic growth), as it is suggested to make individuals interpret life challenges in a more positive way, as an inevitable component of their goal-striving, in line with top-down theories of subjective well-being (Diener, 1984; Diener & Ryan, 2009), coping theories (Carver & Connor-Smith, 2010) and more recent conceptualizations of the role of positive personal qualities under adversity (Casali et al., 2022; Waters, Allen, & Arslan, 2021). This might have been particularly crucial in Italy, which was the first Western country hit by the pandemic, and Italian students who had to face prolonged periods of school closure, social distancing and interruption of any extracurricular activity (UNESCO, 2021). Therefore, elucidating the grit-posttraumatic growth-life satisfaction links may help researchers and practitioners alike to better understand how students successfully coped with learning in a highly stressful situation as a pandemic can be. In this study, we aim to examine the longitudinal relations between the two separate facets of grit (perseverance of effort and consistency of interest), life satisfaction at both the beginning and end of the school year, and posttraumatic growth possibly reported at the end the of the school year in a sample of students in grades 6–12. 2 Literature review 2.1 Grit Grit was first defined by Duckworth et al. (2007) as perseverance and passion for long-term goals, which means working hard and maintaining one's interest and effort over time even under adverse situations. Operationally, grit was conceived as having two facets converging in an overarching grit factor: Consistency of interests (keeping the same objective for a reasonable amount of time without losing interest), and perseverance of effort (strive despite failures). Recent studies (Credé, 2018; Credé et al., 2017; Ponnock et al., 2020; Rimfeld et al., 2016) have questioned this structure and in particular the role of consistency of interests, which seems to have a much lower predictive role compared to perseverance of effort. Indeed, Credé et al. (2017) meta-analysis did not support grit's structural validity (i.e., perseverance of effort and consistency of interests did not satisfactorily converge into a second-order factor). Consequently, authors (Credé, 2018; Datu, 2021; Guo et al., 2019) have proposed to study the two dimensions separately and to focus more on the perseverance facet (Credé et al., 2017). Other studies have championed the need to revise the original grit scales, and modified versions have been developed using newer techniques, including the Academic Grit Scale (Clark & Malecki, 2019), that specifically focuses on academic achievement in youth and was developed under the item response theory (IRT), the Oviedo Grit Scale (Postigo et al., 2021a), which followed expert guidelines on instrument development, or the (Morell et al., 2021), which specifically addresses grit's temporal dimension. All these scales resulted unidimensional. Similarly, more recent investigations of the Short Grit Scale also suggest grit to be unidimensional (Gonzalez et al., 2020). Lastly, it has been suggested to include adaptability as either a third dimension of grit (see the triarchic model by Datu et al., 2017) or a key component of grit together with perseverance of effort, to be preferred to consistency of interests, at least when considering gifted students (Datu et al., 2022). These modified versions have proved particularly reliable in collectivistic contexts such as the Philippines and China. Following this debate on grit facets, in the present study we decided to consider perseverance of effort and consistency of interests as separate dimensions (supported by meta-metanalytic evidence, Credé et al., 2017), and our expectations centered around perseverance as it appears to be the subdimension contributing the most to life satisfaction (Credé et al., 2017). More specifically, we anticipated that perseverance (rather than consistency) would be positively related to both our variables of interest, that is, posttraumatic growth and life satisfaction. 2.2 Posttraumatic growth Posttraumatic growth consists of the positive enduring changes reported by individuals after encountering major stressors (e.g., the COVID-19 pandemic) that invalidate the way they see the world (Calhoun & Tedeschi, 2014; Tedeschi & Calhoun, 1995). Posttraumatic growth changes generally occur in five domains (i.e., openness to new possibilities, interpersonal relationships, personal strengths, appreciation of life, and the spiritual domain) and have been found in both adults and adolescents (Meyerson et al., 2011). Meta-analytic evidence (Vishnevsky et al., 2010) shows that women tend to report higher posttraumatic growth and that age moderates this effect, with older women reporting higher PTG. However, this construct is not exempt from criticism. As reviewed by Infurna and Jayawickreme (2019), posttraumatic growth, especially when measured through retrospective recalling, may be better seen as a coping strategy rather than actual personality change. Only a few prospective longitudinal studies have been conducted to examine actual growth in the aftermath of adverse events, and none of them with adolescents (Harmon & Venta, 2021). Given the lack of evidence, we preferred to err on the side of caution and conceptualize posttraumatic growth in terms of coping rather than personality change, even though the role of posttraumatic growth may be relevant to student learning in pandemic conditions. In particular, posttraumatic growth could be regarded as a form of meaning-focused coping, in which individuals draw on their values to positively reappraise a stressful experience (Carver & Connor-Smith, 2010); this kind of coping is especially likely when the stressful situation is uncontrollable. In this sense, the positive changes reported under the COVID-19 pandemic, few months after the first wave in adults and high-school graduates (Chen et al., 2021; Yu et al., 2021), as well as in adolescents (Waters, Allen, & Arslan, 2021), may rather signal a positive adaptation to the pandemic situation. Importantly, scarce attention has been paid in these studies to the individual characteristics possibly explaining why some individuals experience such growth (or positively respond to a traumatic event, a form of meaning-focused coping (Carver & Connor-Smith, 2010), while others do not. Because of the attention dedicated to grit as an important factor of success in adolescence (and adulthood), it might be of interest to understand whether gritty students are also capable of a better reframing of a difficult situation, possibly resulting in posttraumatic growth. Encouraging evidence exists, for example, that grit dimensions serve as resilience factors in at-risk adolescents (Tang et al., 2021), moderating the effect of burnout on depressive symptoms. Similarly, recent studies underline the conceptual similarity of these two constructs (Cheng et al., 2023), and newer conceptualizations of (academic) grit explicitly integrated resilience to precisely emphasize how grit is an effortful goal pursuit despite delays in progress and adversity (Clark & Malecki, 2019). Overall, this evidence strengthens the idea that grit is a resilience factor for students, facilitating both positive reappraisal and life satisfaction. 2.3 Life satisfaction Born in the positive psychology research field, life satisfaction represents the cognitive component of subjective wellbeing (Diener, 1984), and can be defined as a global assessment of one's contentment with their life conditions. Even though it is considered as quite stable compared to the affective component of subjective wellbeing, life satisfaction can be affected by gender (with females usually reporting lower life satisfaction than males, see Chen et al., 2019), age (with a decrease over the course of adolescence, an in very old age, e.g., Baird et al., 2010; Fergusson et al., 2015), and external events (Anusic & Schimmack, 2016). In particular, longitudinal studies suggest that life satisfaction significantly decreased from pre- to during the pandemic in adolescents (Magson et al., 2021; von Soest et al., 2020). However, these studies considered only the immediate effects of the pandemic (around two months after lockdown was declared), while there is a lack of research on longer term effects on life satisfaction. Since life satisfaction has multiple cascading effects (e.g., on mental health; Fergusson et al., 2015), a deeper understanding of the individual features potentially protecting students from experiencing such decrease in life satisfaction is warranted. 2.4 Grit, posttraumatic growth, and life satisfaction Grit, and especially perseverance, is regarded as a crucial intrapersonal competency across several important taxonomies, ranging from the National Research Council classification of the 21st century competencies, to the VIA classification of character strengths (Peterson & Seligman, 2004), to the World Economic Forum (2016) character qualities for 21st century students, as it is suggested to be associated with a higher level of life satisfaction in students (Bruna et al., 2019; Credé et al., 2017; Feraco et al., 2023; Singh & Jha, 2008). Furthermore, there is growing interest in understanding the psychological mechanisms that link grit to life satisfaction. In this vein, there is evidence that basic psychological needs satisfaction plays a relevant mediating role in the relationship between grit and life satisfaction (Jiang et al., 2019; Jin & Kim, 2017). In these studies, grit is conceptualized as a dispositional quality that favors needs satisfaction, which in turn leads to greater subjective wellbeing. But what about adverse situations? Which mechanisms can explain the grit-wellbeing relationship in such contexts? Grit seems to be of special relevance under stressful conditions (Bono et al., 2020). In line with classic coping theorizations (Carver & Connor-Smith, 2010), personal qualities like grit and coping strategies such as posttraumatic growth have mediating effects on well-being, namely grit facilitates posttraumatic growth, which in turn influences well-being, therefore favoring adjustment to the stressful situation. Even more precisely, according to recent theorizations in positive psychology, intrapersonal features as grit favor growth following major stressors through a “building effect”, that is, the possibility to turn a crisis into an opportunity to develop new practices, processes and visions, which in turn leads to better mental health (Waters, Algoe, et al., 2021). In this sense, grittier students should be better equipped to keep a positive attitude even under adverse situations, being better able to interpret them as opportunities for growth, as shown by studies conducted during the COVID-19 pandemic in student populations (Bono et al., 2020; Wang et al., 2023). Accordingly, perseverance has been associated with growth following traumatic life events (Peterson et al., 2008) and earthquakes (Duan & Guo, 2015). Similarly, in a pandemic situation grit can directly positively impact life satisfaction by helping students stay focused and keep pursuing their objectives, thus maintaining a steady level of subjective satisfaction (Bono et al., 2020). In turn, posttraumatic growth has been suggested to relate to life satisfaction: Reporting positive changes after a major traumatic event brings about greater life satisfaction by helping individuals find meaning in the traumatic event, and integrate it within one's vision of life, therefore acting as a positive coping strategy (Chen et al., 2021; Triplett et al., 2012). At the same time, being satisfied with one's life conditions paves the way to experiencing higher posttraumatic growth (Tomaszek & Muchacka-Cymerman, 2020). Therefore, it could be that students who were generally satisfied with their life at the beginning of the school year (Time 1) may have experienced greater growth following COVID-19 pandemic (Time 2), and this in turn may have positively affected their subsequent life satisfaction. These considerations are in line with top-down theories of subjective well-being (Diener, 1984; Diener & Ryan, 2009) positing that relatively stable personal qualities (like grit) contribute to subjective well-being (life satisfaction) by impacting the way individuals react to events and interpret reality (e.g., a global pandemic). In this view, individuals holding more positive attitudes should appraise certain situations more positively than those with negative attitudes. This is the case for gritty individuals, who tend to retain optimistic views even in the face of adversity and setbacks (Duckworth et al., 2007; Jin & Kim, 2017), which in turn should lead them to appraise stressful situations as opportunities for growth, which may prompt the development of higher levels of life satisfaction. Indeed, recent evidence suggests that grit predicts posttraumatic growth following COVID-19 pandemic in university students (Wang et al., 2023). More in general, a systematic review (Henson et al., 2021) showed how personality traits (such as conscientiousness, which is closely linked to grit) are among the main determinants of posttraumatic growth. 2.5 Rationale and hypotheses To our knowledge, no studies have investigated the relation between Duckworth's grit and posttraumatic growth, except for a study examining the validity of the PTGI measure (Silverstein et al., 2018), nor its longitudinal effect on life satisfaction. Moreover, there is a lack of knowledge on the longer-term effects of COVID-19 pandemic on life satisfaction and which personal qualities can support it. The present study aims at assessing the direct and indirect relations longitudinally linking grit components (perseverance of effort and consistency of interests), life satisfaction at two time-points (to account for its possible variation over time, Magson et al., 2020), and posttraumatic growth in school-aged students. More specifically, based on previous studies and top-down theoretical considerations on the relationship between personal qualities, meaning-focused coping strategies, and well-being (Carver & Connor-Smith, 2010) as well as grit's building effects (Waters, Algoe, et al., 2021), we hypothesize the following:Hypothesis 1 Perseverance of effort at Time 1 (beginning of school year) will be positively associated with both posttraumatic growth and life satisfaction at Time 2 (end of school year); Hypothesis 2 Posttraumatic growth at Time 2 will mediate the relationship between perseverance at Time 1 and life satisfaction at Time 2; Hypothesis 3 (explorative): Posttraumatic growth at Time 2 will mediate the relationship between life satisfaction at Time 1 and life satisfaction at Time 2; those students who were already satisfied with their life may have more available resources and therefore be more prone to report growth following the COVID-19 pandemic. 3 Materials and methods 3.1 Participants A total of 558 students in grades 6–12 (i.e., 11–18 years old) completed the first part of the study (i.e., the measurement of grit and life satisfaction) and 494 (88.5 %) completed the second part because some of them were absent from school. Finally, complete1 records for 445 (160 males, Mage = 14.25, SDage = 2.11) students matched between Time 1 and Time 2 and were included in the analysis.2 Students participated on a voluntary basis after their parents or the students themselves (if 18 years old) provided their informed consent. The study was approved by the Ethical Committee of the University of [BLINDED]. All participants were informed about the purposes of the study and gave their written informed consent in accordance with the Declaration of Helsinki (World Medical Association, 2013). Power analysis was performed via simulation prior to data collection. We simulated 10,000 data sets for different sample sizes starting from a theory-based covariance matrix with small-to-medium hypothetical correlations (i.e., r = 0.30) between grit, posttraumatic growth, and life satisfaction. For each simulated data set, the hypothesized model (see Results section) was fitted, and the results were saved for power calculation. Assuming a significance level of α = 0.05, it emerged that with 400 participants power was equal to 0.74 and with 500 to 0.87. 3.2 Materials The Short Grit Scale (Duckworth & Quinn, 2009; validated in Italian by Sulla et al., 2018) was administered at Time 1 only. This involves eight items on a 5-point Likert scale measuring two facets of grit (i.e., passion for long-term goals): consistency of interest, i.e., to keep focused on the same interests for a long time (four items, e.g., “New ideas and projects sometimes distract me from previous ones”), and perseverance of effort, i.e., to maintain one's effort despite failures (four items, e.g., “Setbacks don't discourage me”). Four items were reversed to calculate the scores. The scale displayed acceptable-to-satisfactory internal consistency in the Italian version (α = 0.76 for the overall mean score and the consistency subscale, α = 0.61 for perseverance) and acceptable in the present study as well (α = 0.75, 0.67, and 0.71 respectively). Given the heated debate on grit's internal structure, we evaluated it in our sample, by comparing the unidimensional model (considering grit as a unique factor composed by the eight items), the two-correlated factors model (considering consistency of interests and perseverance of effort separately), and the hierarchical model (considering the convergence of consistency and perseverance into a second-order grit factor). The results indicated better fit for the two-factor model (χ 2 (19, N = 445) = 23.75, p < .01, CFI = 0.99, NNFI = 0.99, RMSEA = 0.02, 90 % confidence interval for RMSEA [0.00, 0.05]) than the unidimensional one (χ 2 (20, N = 558) = 95.02, p < .01, CFI = 0.92, NNFI = 0.89, RMSEA = 0.08, 90 % confidence interval for RMSEA [0.07, 0.10]), while the hierarchical model did not even converge. The Revised Posttraumatic Growth Inventory for Children (Kilmer et al., 2009) was administered at Time 2 only. This involves 10 items on a 4-point Likert scale (from 0 = “I did not experience any change” to 3 = “I experienced change to a very great degree”) measuring posttraumatic growth. The scale was translated in Italian following a back-translation procedure and instructions were adapted to suit to the COVID-19 pandemic. The questionnaire includes two items for each of five domains of posttraumatic growth: Others (e.g., “I learned how nice/helpful people can be”), new possibilities (e.g., “I have a chance to do things I couldn't”), personal strength (e.g., “I can handle big problems better”), spirituality (e.g., “My faith/belief in God is stronger”), and appreciation of life (e.g., “I know what is important to me”). A total score is calculated as indicated in the validation study. A confirmatory factor analysis run on our data supported the factorial structure of the original scale: χ2 (35, N = 454) = 89.81, p < .01, CFI = 0.96, NNFI = 0.94, RMSEA = 0.06, 90 % confidence interval for RMSEA [0.04, 0.07]. The scale showed good internal consistency (Cronbach's α = 0.85, Kilmer et al., 2009; Cronbach's α = 0.76 for the current sample). The Satisfaction With Life Scale (Diener et al., 1985; Italian validation by Di Fabio & Gori, 2016) was administered at both Times 1 and 2. This contains five items scored on a 7-point Likert scale (from 1 = “completely disagree” to 7 = “completely agree”) and measures overall life satisfaction (e.g., “The conditions of my life are excellent”). The average score was calculated. The scale showed good internal consistency (Cronbach's α = 0.85, Di Fabio & Gori, 2016; Cronbach's α = 0.85 at Time one and 0.84 at Time two for the current sample). Moreover, multigroup CFA supported strict measurement invariance across time-points, as shown in Table 1 .Table 1 Measurement invariance for SWLS across time-points (Time 1 vs. Time 2). Table 1 Df Χ2 ΔΧ2 ΔDf p Configural 10 0.53 Metric 14 3.79 3.26 4 .52 Scalar 18 5.03 1.24 4 .87 Strict 23 9.78 4.75 5 .45 Note. Df = degrees of freedom, Χ2 = chi squared, ΔΧ2 = difference in chi squared, ΔDf = difference in degrees of freedom. 3.3 Procedure Italy was the first European country to face the COVID-19 outbreak in February 2020. To respond to the pandemic situation, schools were closed (March 5th) and lessons moved online for the rest of the academic year. The following academic year (2020/2021) was again characterized by school closures and restrictions that varied depending on the regional COVID-19 situation, causing schools to close and reopen on multiple occasions. Moreover, also when opened, students had to follow strict rules regarding social contacts, use of masks, and use of school spaces. At the beginning of the school year (September 2020), schools were contacted by email or phone to explain the project. A total of 27 classes from five different schools located in four different Italian regions agreed to participate to the study. Consent forms were provided to schools that showed interest, and the teachers of the participating classes distributed them to students (if 18 years old) or their parents. After the signed consent forms were returned, a Qualtrics link was provided to the teachers, and the students completed the questionnaires at two time points during school time under the supervision of a trained psychologist. During the first data collection (between October 2020 and the beginning of January 2021), personal information (e.g., gender, age, class) were collected, together with the responses to the Short Grit Scale and the Satisfaction With Life Scale. During the second data collection (between the last two weeks of May and the first two weeks of June 2021), students answered to the two questionnaires about posttraumatic growth and life satisfaction. Completion of all the questionnaires required no more than 20 min per class. The interval between the two data collection was chosen to cover the entire academic year. 3.4 Data analysis All analyses were run using R (R Core Team, 2020). Preliminarily, we inspected the internal structure of the self-report measures used in the study by means of confirmatory factor analysis (CFA), using the package lavaan (Rosseel, 2012). We considered items as ordinal and adopted diagonally weighted least squares (DWLS) as estimator. The goodness of fit to the data for each model was examined using multiple indices: The comparative fit index (CFI); the non-normed fit index (NNFI); and the root mean squared error of approximation (RMSEA). Models with CFI and TLI values of 0.95 or more (Bentler & Bonett, 1980), and RMSEA values of 0.08 or less (Schermelleh-engel et al., 2003) should be considered adequate. To assess measurement invariance between SWLS at Time 1 and SWLS at Time 2, we adopted multigroup CFA and compared configural (Time 1 vs. Time 2), metric (equal loadings), scalar (equal loadings and intercepts), and strict (equal loadings, intercepts, and residuals) models by means of ANOVA, where p > .05 indicate the models do not significantly differ. Then, we used paired t-tests to compare life satisfaction scores at Time 1 and 2. Lastly, we fitted a longitudinal SEM model (see Fig. 1 ) to examine the multivariate associations between grit dimensions (i.e., consistency of interests and perseverance of effort, considered as correlated latent factors), posttraumatic growth, and life satisfaction. In particular, we estimated i) the direct associations of the two grit's dimensions at Time 1 and of life satisfaction at Time 1 with posttraumatic growth at Time 2 and life satisfaction at Time 2 ii) and the direct associations of posttraumatic growth at Time 2 with life satisfaction at Time 2. The indirect effects of posttraumatic growth at Time 2 on the relation between grit's facets at Time 1 and life satisfaction at Time 2, as well as between life satisfaction at Times 1 and 2 were also calculated (Fig. 2 ).Fig. 1 Graphical representation of the hypothesized model. Fig. 1 Fig. 2 The model results. Note. Numbers represent standardized beta coefficients, and dotted lines represent non-significant relations. Age and gender were added as covariates in every relation. All loadings are significant for p < .001. ***p < .001. Fig. 2 Given that age and gender are related with the variables considered (Chen et al., 2019; Fergusson et al., 2015; Vishnevsky et al., 2010), they were both entered as covariates in all the specified regressions. 4 Results Table 1 shows the means, standard deviations, and correlations between all the variables. 4.1 Life satisfaction at Times 1 and 2 Life satisfaction at Times 1 and 2 was compared through t-test to assess any changes. The results indicated that life satisfaction at Time 1 was significantly higher than at Time 2 (t(888) = 3.70, p < .001, d = 0.25). 4.2 Association between grit, post-traumatic growth, and life satisfaction The longitudinal SEM model displayed satisfactory fit indices (χ 2 (386, N = 445) = 988.07, p < .01, CFI = 0.97, NNFI = 0.97, RMSEA = 0.06, 90 % confidence interval for RMSEA [0.055, 0.064]). The results (see Table 2 ) showed that perseverance of effort at Time 1 (β = 0.20, p < .001) and life satisfaction at Time 1 (β = 0.23, p < .001) significantly related to posttraumatic growth at Time 2. Second, posttraumatic growth at Time 2 (β = 0.26, p < .001) and life satisfaction at Time 1 (β = 0.64, p < .001), but not perseverance of effort at Time 1 (β = −0.08, p > .05), significantly related to life satisfaction at Time 2. Consistency of interests was not significantly related to any of the outcome measures. Descriptively, age and gender resulted significantly negatively related with life satisfaction at Time 1, with males and younger students reporting higher satisfaction (β = −0.10 and β = −0.18, respectively). Moreover, male students also reported slightly higher posttraumatic growth at Time 2 (β = −0.07, p = .01) and lower consistency of interests at Time 1 (β = 0.11, p = .01) (Table 3 ).Table 2 Means, standard deviations, and correlations between all study variables. Table 2 M SD 1. 2. 3. 4. 5. 6. 1.Age 14.26 2.11 2.Consistency of interests T1 13.21 2.82 −0.01 3.Perseverance of effort T1 13.35 2.92 −0.04 0.41⁎⁎⁎ 4.Life satisfaction T1 24.02 6.59 −0.19⁎⁎ 0.20⁎⁎⁎ 0.42⁎⁎⁎ 5.Life satisfaction T2 22.37 6.7 −0.15⁎⁎ 0.17⁎⁎⁎ 0.31⁎⁎⁎ 0.60⁎⁎⁎ 6.Posttraumatic growth T2 23.16 5.38 −0.11⁎ 0.08 0.22⁎⁎⁎ 0.29⁎⁎⁎ 0.39⁎⁎⁎ 7.Gendera – – 0.17⁎⁎⁎ 0.11⁎ 0.02 −0.15⁎⁎ −0.19⁎⁎⁎ −0.12⁎ a Polychoric correlations were calculated. ⁎ p < .05. ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table 3 Complete results for direct and indirect effects analyzed. Table 3Dependent variable Predictor SE z β Direct effects Life satisfaction T2 Consistency of interests T1 0.08 1.74 0.09 Life satisfaction T2 Perseverance of effort T1 0.10 −1.34 −0.08 Life satisfaction T2 Posttraumatic growth T2 0.04 9.63 0.26⁎⁎⁎ Life satisfaction T2 Life satisfaction T1 0.07 14.49 0.64⁎⁎⁎ Life satisfaction T2 Gender 0.11 −0.68 −0.06 Life satisfaction T2 Age T1 0.03 0.24 0.01 Posttraumatic growth T2 Consistency of interests T1 0.05 −1.09 −0.05 Posttraumatic growth T2 Perseverance of effort T1 0.06 3.71 0.20⁎⁎⁎ Posttraumatic growth T2 Life satisfaction T1 0.03 7.65 0.23⁎⁎⁎ Posttraumatic growth T2 Gender 0.07 −2.46 −0.08⁎ Posttraumatic growth T2 Age T1 0.02 −1.55 −0.05 Life satisfaction T1 Gender 0.06 −3.61 −0.11⁎⁎⁎ Life satisfaction T1 Age T1 0.02 −6.99 −0.22⁎⁎⁎ Consistency of interest Gender 0.09 2.63 0.11⁎⁎ Consistency of interest Age T1 0.02 −0.21 −0.01 Perseverance of effort Gender 0.08 −1.49 0.05 Perseverance of effort Age T1 0.02 −1.00 −0.06 Indirect effects Life satisfaction T2 Consistency T1 x Posttraumatic growth T2 0.02 −1.07 −0.01 Life satisfaction T2 Perseverance T1 x Posttraumatic growth T2 0.08 3.22 0.05⁎⁎ Life satisfaction T2 Life satisfaction T1 x Posttraumatic growth T2 0.01 8.15 0.06⁎⁎⁎ Correlations Consistency of interest Perseverance of effort T1 0.03 21.29 0.62⁎⁎⁎ Consistency of interest Life satisfaction T1 0.02 12.43 0.29⁎⁎⁎ Perseverance of effort Life satisfaction T1 0.02 24.61 0.53⁎⁎⁎ Note. SE = standard error; z = test statistic; β = standardized beta coefficient. ⁎ p < .05. ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Posttraumatic growth fully mediated the relation between perseverance of effort and life satisfaction at Time 2 (β = 0.05, p = .001); it also partially mediated the relation between life satisfaction at Times 1 and 2 (β = 0.06, p < .001). 5 Discussion and conclusions Since the beginning of COVID-19 pandemic, life has deeply changed for nearly everybody, and especially school-aged students, whose scholastic and personal conditions had to completely be re-arranged. This abrupt and massive shift has been suggested to affect not only students' academic life, but also their subjective wellbeing, i.e., their satisfaction with life (Magson et al., 2021; von Soest et al., 2020). In this scenario, the present study drew on a top-down approach to well-being (Diener, 1984), classical coping theories (Carver & Connor-Smith, 2010) and recent theorizations on the role of positive psychology in a pandemic (Waters, Algoe, et al., 2021) to investigate whether grit, an intraindividual quality identified as one of the most prominent noncognitive factors sustaining students (Lavy, 2020), longitudinally supported students' life satisfaction as well as promoting a positive reappraisal of the pandemic, as expressed by posttraumatic growth reported at the end of the school year. Preliminarily, our findings evidenced the two correlated factors model as the one best fitting our data, as it performed better than both the unidimensional model and the hierarchical one. This result contradicts recent evidence in support of grit's unidimensionality (Clark & Malecki, 2019; Gonzalez et al., 2020; Postigo et al., 2021a), while being in line with previous findings on a high-school sample (Muenks et al., 2017). Generally, it may be argued that the original grit scale has a varying internal structure (Morell et al., 2021) and that more rigorously developed measures exhibit more stable structures and possibly better represent the grit construct. Then, the longitudinal SEM model provided us with a deeper understanding of the role of grit components with respect to life satisfaction and posttraumatic growth. It emerged that only perseverance of effort at Time 1 was significantly related to posttraumatic growth following the COVID-19 pandemic, while the direct relation of perseverance at Time 1 with life satisfaction at Time 2 was not confirmed. These results partially support our hypothesis (H1) and advocate for a more prominent role of the perseverance component compared to the consistency one, as recently suggested by other studies (Credé, 2018; Credé et al., 2017; Rimfeld et al., 2016). Moreover, these findings indicate that perseverance of effort can enable a more positive appraisal of a traumatic experience such as a global pandemic, possibly making students more able to keep working hard in spite of this stressful situation, extending previous findings on adults to adolescents (Duan & Guo, 2015; Peterson et al., 2008; Silverstein et al., 2018). In other words, our results favor perseverance's “building effect” and ability to positively influence positive reappraisal, i.e., being able to transformatively get through a crisis by developing a new outlook, which can then lead to improved wellbeing. Even more interestingly, and in accordance with our hypotheses (H2−H3), posttraumatic growth at Time 2 fully mediated the relationship between perseverance of effort at Time 1 and life satisfaction at Time 2, and partially mediated the relationship between life satisfaction at Times 1 and 2. As for the first mediation effect, perseverance appeared to act as a dispositional individual resource that enables growth at the end of the school year (the abovementioned building effect), which in turn seems to make students more satisfied with their life (a positive cascading effect of the building effect). Furthermore, our results point to a bidirectional virtuous circle, that is, life satisfaction at Time 1 positively affected posttraumatic growth at Time 2, which in turn was related to greater life satisfaction at Time 2, even after accounting for life satisfaction at Time 1 (Tomaszek & Muchacka-Cymerman, 2020). Better understanding such intervening mechanisms could shed light on the actual processes that bring grittier people to achieve higher life satisfaction in their lives. These results support a top-down view of well-being as being affected by personality traits (in our case, perseverance of effort) that enable a more positive appraisal of life situations (as expressed by posttraumatic growth). Moreover, our findings align with the idea that posttraumatic growth can be viewed as a coping strategy (Infurna & Jayawickreme, 2019) focused on finding meaning even in the face of an uncontrollable, highly stressful situation (the COVID-19 pandemic). In this sense, perseverance of effort may be understood as a resilience factor (Tang et al., 2021) that facilitates such meaning-making coping, enabling a more positive view of one's life. Prospective studies should be conducted to better understand whether this line of thought is preferable to conceiving posttraumatic growth in terms of personality change. Lastly, age and gender showed some small, yet significant effects. More specifically, male and younger students displayed higher life satisfaction at the beginning of the school year, in line with meta-analytical evidence indicating a slightly higher subjective wellbeing in males in European countries (Chen et al., 2019). Nevertheless, it should be kept in mind that male students were slightly underrepresented in our sample, making it harder to generalize this finding. As for age, our result is in line with some previous evidence showing a small decline in life satisfaction with the onset of adolescence (Baird et al., 2010; Cavallo et al., 2015; Goldbeck et al., 2007; Moksnes et al., 2013). This decrease may be seen as a physiological phenomenon due to increasing challenges associated with the transition from childhood to adulthood. Male students also reported greater posttraumatic growth at Time 2 and lower consistency of interests at Time 1. These results are not in line with meta-analytic evidence showing higher posttraumatic growth in women than men (Vishnevsky et al., 2010). However, it should be noted that this meta-analysis only considered adult samples and that moderator analysis showed gender differences to be more pronounced with age. As for consistency of interests, previous studies did not evidence gender differences in this subdimension (Duckworth & Quinn, 2009); future studies with more balanced samples would be needed to better ascertain this point. Another limitation of the present study is the lack of pre-pandemic measures and of a measure of grit at Time 2. This means we cannot assess whether grit and satisfaction with life changed already as a result of the pandemic. Adding a third wave of data collection would have allowed to strengthen our results by providing evidence of enduring positive effects of grit and posttraumatic growth on life satisfaction, as well allowing us to test for reciprocal effects of our variables. Future studies could incorporate multiple waves and also focus on the moderating effects of grit components on well-being. Furthermore, we did not examine students' perceptions about COVID-19, which may have strengthened our understanding of posttraumatic growth as meaning-focused coping strategy. Last, our choice of fitting a single model for both waves might be regarded as a limitation. We did so to specifically examine the contemporary associations of grit with life satisfaction at both time-points, as well as exploring the effect of baseline life satisfaction on posttraumatic growth. Despite these shortcomings, this study supports the building effect of perseverance of effort with respect to posttraumatic growth, and a reciprocal positive influence of posttraumatic growth and life satisfaction. Fostering perseverance through dedicated training programs could therefore prove particularly relevant in stressful contexts as a means of building students' resources and help them keep and further develop a positive outlook on the situation itself and their life conditions. This could be done as a preventative universal measure to teach students how to bring forth their perseverance in everyday life, as well as a means to manage stressful situations. Such preventative interventions may also have cascading positive implications effects on students' academic outcomes, including self-regulated learning and academic achievement. Our findings also suggest that this kind of work may benefit older and female students even more, provided they seem to be at increased risk of experiencing lower life satisfaction and posttraumatic growth. Other than specific training programs, teachers could support their students' goal-striving especially when they encounter setbacks, encouraging them to find strategies to continue working towards their objectives. Informed consent for human participants The study was approved by the University of Padova's Ethics Committee for Research in Psychology. All the participants (or their parents) signed a consent form before they participated in the study. CRediT authorship contribution statement Nicole Casali: Conceptualization, Methodology, Formal analysis, Writing – original draft, Writing – review & editing. Tommaso Feraco: Conceptualization, Data curation, Writing – original draft, Writing – review & editing. Chiara Meneghetti: Writing – review & editing, Supervision. Declaration of competing interest This research did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors. The authors have no potential conflicts of interest to report. Data availability Data are available on Figshare, https://doi.org/10.6084/m9.figshare.19403054. Acknowledgments The present work was conducted as part of the Dipartimenti di Eccellenza research program (DM 11/05/2017 n. 262), supported by a grant from 10.13039/501100003407 MIUR to the Department of General Psychology, University of Padua. 1 Some students were only present for the second part of the study (they were absent during the first day of data collection), while others failed to report their subject's code correctly, making it impossible to match their records. 2 Please note that the posttraumatic growth data were already published in a different manuscript (Author et al., 2022) ==== Refs References Anusic I. Schimmack U. Stability and change of personality traits, self-esteem, and well-being: Introducing the meta-analytic stability and change model of retest correlations Journal of Personality and Social Psychology 110 2016 766 781 10.1037/pspp0000066 26619304 Baird B.M. Lucas R.E. Donnellan M.B. Life satisfaction across the lifespan: Findings from two nationally representative panel studies Social Indicators Research 99 2010 183 203 10.1007/s11205-010-9584-9 21113322 Bentler P.M. Bonett D.G. Significance tests and goodness of fit in the analysis of covariance structures Psychological Bulletin 88 3 1980 588 606 10.1037/0033-2909.88.3.588 Bono G. Reil K. Hescox J. Stress and wellbeing in urban college students in the U.S. during the COVID-19 pandemic: Can grit and gratitude help? International Journal of Wellbeing 10 3 2020 Articolo 3 https://internationaljournalofwellbeing.org/index.php/ijow/article/view/1331 Bruna M.O. Brabete A.C. Izquierdo J.M.A. Reliability generalization as a seal of quality of substantive meta-analyses: The case of the VIA inventory of strengths (VIA-IS) and their relationships to life satisfaction Psychological Reports 122 3 2019 1167 1188 10.1177/0033294118779198 29848214 Calhoun L.G. Tedeschi R.G. Handbook of posttraumatic growth: Research and practice 2014 Routledge Carver C.S. Connor-Smith J. Personality and coping Annual Review of Psychology 61 1 2010 679 704 10.1146/annurev.psych.093008.100352 Casali N. Feraco T. Meneghetti C. Character strengths sustain mental health and post-traumatic growth during the COVID-19 pandemic. A longitudinal analysis Psychology & Health 37 12 2022 1663 1679 10.1080/08870446.2021.1952587 34288790 Cavallo F. Dalmasso P. Ottová-Jordan V. Brooks F. Mazur J. Välimaa R. Gobina I. Gaspar de Matos M. Raven-Sieberer U. Positive Health Focus Group Trends in life satisfaction in European and North-American adolescents from 2002 to 2010 in over 30 countries European Journal of Public Health 25 Suppl. 2 2015 80 82 10.1093/eurpub/ckv014 25805795 Chen R. Sun C. Chen J.-J. Jen H.-J. Kang X.L. Kao C.-C. Chou K.-R. A large-scale survey on trauma, burnout, and posttraumatic growth among nurses during the COVID-19 pandemic International Journal of Mental Health Nursing 30 1 2021 102 116 10.1111/inm.12796 33107677 Chen X. Cai Z. He J. Fan X. Gender differences in life satisfaction among children and adolescents: A meta-analysis Journal of Happiness Studies 21 6 2019 2279 2307 10.1007/S10902-019-00169-9 Cheng E.H. Kam C.C.S. Cui T. Revisiting grit: How much does it overlap with resilience? International Journal of Educational Research 119 2023 102187 10.1016/j.ijer.2023.102187 Clark K.N. Malecki C.K. Academic grit scale: Psychometric properties and associations with achievement and life satisfaction Journal of School Psychology 72 2019 49 66 10.1016/j.jsp.2018.12.001 30819462 Credé M. What shall we do about grit? A critical review of what we know and what we don't know Educational Researcher 47 9 2018 606 611 10.3102/0013189X18801322 Credé M. Tynan M.C. Harms P.D. Much ado about grit: A meta-analytic synthesis of the grit literature Journal of Personality and Social Psychology 113 3 2017 492 511 10.1037/pspp0000102 27845531 Datu J.A.D. Beyond passion and perseverance: Review and future research initiatives on the science of grit Frontiers in Psychology 11 2021 10.3389/fpsyg.2020.545526 Datu J.A.D. Yuen M. Chen G. Development and validation of the Triarchic Model of Grit Scale (TMGS): Evidence from Filipino undergraduate students Personality and Individual Differences 114 2017 198 205 10.1016/j.paid.2017.04.012 Datu J.A.D. Yuen M. Fung E. Zhang J. Chan S. Wu F. The satisfied lives of gifted and gritty adolescents: Linking grit to career self-efficacy and life satisfaction The Journal of Early Adolescence 42 8 2022 1052 1072 10.1177/02724316221096082 Di Fabio A. Gori A. Measuring adolescent life satisfaction: Psychometric properties of the satisfaction with life scale in a sample of Italian adolescents and young adults Journal of Psychoeducational Assessment 34 5 2016 501 506 10.1177/0734282915621223 Diener E. Subjective well-being Psychological Bulletin 95 1984 542 575 10.1037/0033-2909.95.3.542 6399758 Diener E. Emmons R.A. Larsen R.J. Griffin S. The satisfaction with life scale Journal of Personality Assessment 49 1 1985 71 75 10.1207/s15327752jpa4901_13 16367493 Diener E. Ryan K. Subjective well-being: A general overview South Africa Journal of Psychology 39 4 2009 391 406 10.1177/008124630903900402 Duan W. Guo P. Association between virtues and posttraumatic growth: Preliminary evidence from a Chinese community sample after earthquakes PeerJ 3 2015 e883 10.7717/peerj.883 Duckworth A.L. Peterson C. Matthews M.D. Kelly D.R. Grit: Perseverance and passion for long-term goals Journal of Personality and Social Psychology 92 6 2007 1087 1101 10.1037/0022-3514.92.6.1087 17547490 Duckworth A.L. Quinn P.D. Development and validation of the Short Grit Scale (Grit–S) Journal of Personality Assessment 91 2 2009 166 174 10.1080/00223890802634290 19205937 Feraco T. Resnati D. Fregonese D. Spoto A. Meneghetti C. An integrated model of school students' academic achievement and life satisfaction. Linking soft skills, extracurricular activities, self-regulated learning, motivation, and emotions European Journal of Psychology of Education 38 2023 109 130 10.1007/s10212-022-00601-4 Fergusson D.M. Mcleod G.F.H. Horwood L.J. Swain N.R. Chapple S. Poulton R. Life satisfaction and mental health problems (18 to 35 years) Psychological Medicine 45 2015 2427 2436 10.1017/S0033291715000422 25804325 Goldbeck L. Schmitz T.G. Besier T. Herschbach P. Henrich G. Life satisfaction decreases during adolescence Quality of Life Research 16 6 2007 969 979 10.1007/s11136-007-9205-5 17440827 Gonzalez O. Canning J.R. Smyth H. MacKinnon D.P. A psychometric evaluation of the short grit scale European Journal of Psychological Assessment 36 4 2020 646 657 10.1027/1015-5759/a000535 33840984 Guo J. Tang X. Xu K.M. Capturing the multiplicative effect of perseverance and passion: Measurement issues of combining two grit facets Proceedings of the National Academy of Sciences 116 10 2019 3938 3940 10.1073/pnas.1820125116 Harmon J. Venta A. Adolescent posttraumatic growth: A review Child Psychiatry & Human Development 52 4 2021 596 608 10.1007/s10578-020-01047-9 32830298 Henson C. Truchot D. Canevello A. What promotes post traumatic growth? A systematic review European Journal of Trauma & Dissociation 5 4 2021 100195 10.1016/j.ejtd.2020.100195 Infurna F.J. Jayawickreme E. Fixing the growth illusion: New directions for research in resilience and posttraumatic growth Current Directions in Psychological Science 28 2 2019 152 158 10.1177/0963721419827017 Jiang X. Fang L. Lyons M.D. Is life satisfaction an antecedent to coping behaviors for adolescents? Journal of Youth and Adolescence 48 11 2019 2292 2306 10.1007/s10964-019-01136-6 31587173 Jin B. Kim J. Grit, basic needs satisfaction, and subjective well-being Journal of Individual Differences 38 2017 29 35 10.1027/1614-0001/a000219 Kilmer R.P. Gil-Rivas V. Tedeschi R.G. Cann A. Calhoun L.G. Buchanan T. Taku K. Use of the revised posttraumatic growth inventory for children Journal of Traumatic Stress 22 3 2009 248 253 10.1002/jts.20410 19462437 Lavy S. A review of character strengths interventions in twenty-first-century schools: Their importance and how they can be fostered Applied Research in Quality of Life 15 2 2020 573 596 10.1007/s11482-018-9700-6 Magson N.R. Freeman J.Y.A. Rapee R.M. Richardson C.E. Oar E.L. Fardouly J. Risk and protective factors for prospective changes in adolescent mental health during the COVID-19 pandemic Journal of Youth and Adolescence 50 1 2021 44 57 10.1007/s10964-020-01332-9 33108542 Meyerson D.A. Grant K.E. Carter J.S. Kilmer R.P. Posttraumatic growth among children and adolescents: A systematic review Clinical Psychology Review 31 6 2011 949 964 10.1016/j.cpr.2011.06.003 21718663 Moksnes U.K. Løhre A. Espnes G.A. The association between sense of coherence and life satisfaction in adolescents Quality of Life Research: an International Journal of Quality of Life Aspects of Treatment, Care and Rehabilitation 22 6 2013 1331 1338 10.1007/s11136-012-0249-9 22886139 Morell M. Yang J.S. Gladstone J.R. Turci Faust L. Ponnock A.R. Lim H.J. Wigfield A. Grit: The long and short of it Journal of Educational Psychology 113 2021 1038 1058 10.1037/edu0000594 Muenks K. Wigfield A. Yang J.S. O'Neal C.R. How true is grit? Assessing its relations to high school and college students' personality characteristics, self-regulation, engagement, and achievement Journal of Educational Psychology 109 5 2017 599 620 10.1037/edu0000153 Peterson C. Park N. Pole N. D'Andrea W. Seligman M.E.P. Strengths of character and posttraumatic growth Journal of Traumatic Stress 21 2 2008 214 217 10.1002/jts.20332 18404632 Peterson C. Seligman M.E.P. Character strengths and virtues: A handbook and classification 2004 American Psychological Association Ponnock A. Muenks K. Morell M. Seung Yang J. Gladstone J.R. Wigfield A. Grit and conscientiousness: Another jangle fallacy Journal of Research in Personality 89 2020 10.1016/j.jrp.2020.104021 Postigo Á. Cuesta M. Fernández-Alonso R. García-Cueto E. Muñiz J. Academic grit modulates school performance evolution over time: A latent transition analysis Revista de Psicodidáctica (English Ed.) 26 2 2021 87 95 10.1016/j.psicoe.2021.03.001 Postigo Á. Cuesta M. Fernández-Alonso R. García-Cueto E. Muñiz J. Temporal stability of grit and school performance in adolescents: A longitudinal perspective Educational Psychology 27 1 2021 77 84 10.5093/psed2021a4 R Core Team R: The R project for statistical computing https://www.r-project.org/ 2020 Rimfeld K. Kovas Y. Dale P.S. Plomin R. True grit and genetics: Predicting academic achievement from personality Journal of Personality and Social Psychology 111 5 2016 780 789 10.1037/pspp0000089 26867111 Rosseel Y. Lavaan: An R package for structural equation modeling and more version 0.5-12 (BETA). 37 2012 Schermelleh-engel K. Moosbrugger H. Müller H. Evaluating the fit of structural equation models: Tests of significance and descriptive goodness-of-fit measures Methods of Psychological Research 2003 23 74 Silverstein M.W. Witte T.K. Lee D.J. Kramer L.B. Weathers F.W. Dimensions of growth? Examining the distinctiveness of the five factors of the posttraumatic growth inventory Journal of Traumatic Stress 31 3 2018 448 453 10.1002/jts.22298 29958338 Singh K. Jha S.D. Positive and negative affect, and grit as predictors of happiness and life satisfaction Journal of the Indian Academy of Applied Psychology 34 2008 40 45 Sulla F. Renati R. Bonfiglio S. Rollo D. Italian students and the grit-S: A self-report questionnaire for measuring perseverance and passion for long-term goals IEEE International Symposium on Medical Measurements and Applications (MeMeA) 2018 2018 1 5 10.1109/MeMeA.2018.8438668 Tang M. Wang D. Guerrien A. A systematic review and meta-analysis on basic psychological need satisfaction, motivation, and well-being in later life: Contributions of self-determination theory PsyCh Journal 9 1 2020 5 33 10.1002/pchj.293 31177644 Tang X. Upadyaya K. Salmela-Aro K. School burnout and psychosocial problems among adolescents: Grit as a resilience factor Journal of Adolescence 86 2021 77 89 10.1016/j.adolescence.2020.12.002 33360420 Tedeschi R.G. Calhoun L.G. Trauma and transformation 1995 SAGE Tomaszek K. Muchacka-Cymerman A. Thinking about my existence during COVID-19, I feel anxiety and awe-the mediating role of existential anxiety and life satisfaction on the relationship between PTSD symptoms and post-traumatic growth International Journal of Environmental Research and Public Health 17 19 2020 7062 10.3390/ijerph17197062 32992515 Triplett K.N. Tedeschi R.G. Cann A. Calhoun L.G. Reeve C.L. Posttraumatic growth, meaning in life, and life satisfaction in response to trauma Psychological Trauma: Theory, Research, Practice, and Policy 4 2012 400 410 10.1037/a0024204 UNESCO Education: From disruption to recovery https://en.unesco.org/covid19/educationresponse 2021 Vishnevsky T. Cann A. Calhoun L.G. Tedeschi R.G. Demakis G.J. Gender differences in self-reported posttraumatic growth: A meta-analysis Psychology of Women Quarterly 34 1 2010 110 120 10.1111/j.1471-6402.2009.01546.x von Soest T. Bakken A. Pedersen W. Sletten M.A. Life satisfaction among adolescents before and during the COVID-19 pandemic Tidsskrift for Den Norske Laegeforening: Tidsskrift for Praktisk Medicin, Ny Raekke 140 10 2020 10.4045/tidsskr.20.0437 Wang S. Zhao Y. Li J. True grit and brain: Trait grit mediates the connection of DLPFC functional connectivity density to posttraumatic growth following COVID-19 Journal of Affective Disorders 325 2023 313 320 10.1016/j.jad.2023.01.022 36627056 Waters L. Algoe S.B. Dutton J. Emmons R. Fredrickson B.L. Heaphy E. Moskowitz J.T. Neff K. Niemiec R. Pury C. Steger M. Positive psychology in a pandemic: Buffering, bolstering, and building mental health The Journal of Positive Psychology 1–21 2021 10.1080/17439760.2021.1871945 Waters L. Allen K.-A. Arslan G. Stress-related growth in adolescents returning to school after COVID-19 school closure Frontiers in Psychology 12 2021 643443 10.3389/fpsyg.2021.643443 World Economic Forum New vision for education: Fostering social and emotional learning through technology. World Economic Forum https://www.weforum.org/reports/new-vision-for-education-fostering-social-and-emotional-learning-through-technology/ 2016 World Medical Association Declaration of Helsinki Ethical principles for medical research involving human subjects JAMA: The Journal of the American Medical Association 3120 20 2013 2191 2194 10.1001/jama.2013.281053 Yu Y. Yu Y. Hu J. COVID-19 among Chinese high school graduates: Psychological distress, growth, meaning in life and resilience Journal of Health Psychology 1359105321990819 2021 10.1177/1359105321990819
PMC010xxxxxx/PMC10288624.txt	==== Front Cell Chem Biol Cell Chem Biol Cell Chemical Biology 2451-9456 2451-9448 Elsevier Ltd. S2451-9456(23)00154-X 10.1016/j.chembiol.2023.05.011 Article Spike-protein proteolytic antibodies in COVID-19 convalescent plasma contribute to SARS-CoV-2 neutralization McConnell Scott A. 1 Sachithanandham Jaiprasath 1 Mudrak Nathan J. 1 Zhu Xianming 4 Farhang Parsa Alba 1 Cordero Radames J.B. 1 Wear Maggie P. 1 Shapiro Janna R. 2 Park Han-Sol 1 Klein Sabra L. 123 Tobian Aaron A.R. 4 Bloch Evan M. 4 Sullivan David J. 1 Pekosz Andrew 1 Casadevall Arturo 15∗ 1 W. Harry Feinstone Department of Molecular Microbiology and Immunology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD 21205, USA 2 Department of International Health, Johns Hopkins School of Public Health, Baltimore, MD 21205, USA 3 Department of Biochemistry and Molecular Biology, Johns Hopkins School of Public Health, Baltimore, MD 21205, USA 4 Department of Pathology, Johns Hopkins School of Medicine, Baltimore, MD 21287, USA ∗ Corresponding author 5 Lead contact 23 6 2023 23 6 2023 18 11 2022 23 3 2023 26 5 2023 © 2023 Elsevier Ltd. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Understanding the mechanisms of antibody-mediated neutralization of SARS-CoV-2 is critical in combating the COVID-19 pandemic. Based on previous reports of antibody catalysis, we investigated the proteolysis of spike (S) by antibodies in COVID-19 convalescent plasma (CCP) and its contribution to viral neutralization. Quenched fluorescent peptides were designed based on S epitopes to sensitively detect antibody-mediated proteolysis. We observed epitope cleavage by CCP from different donors which persisted when plasma was heat-treated or when IgG was isolated from plasma. Further, purified CCP antibodies proteolyzed recombinant S domains, as well as authentic viral S. Cleavage of S variants suggests CCP antibody-mediated proteolysis is a durable phenomenon despite antigenic drift. We differentiated viral neutralization occurring via direct interference with receptor binding from that occurring by antibody-mediated proteolysis, demonstrating that antibody catalysis enhanced neutralization. These results suggest that antibody-catalyzed damage of S is an immunologically relevant function of neutralizing antibodies against SARS-CoV-2. Graphical abstract McConnell et al. identify catalytic antibodies specific to the SARS-CoV-2 protein across a large panel of convalescent plasma donors. Antibody-mediated degradation of spike epitopes is correlated with viral neutralization capacity and catalysis is persistently observed despite antigenic drift of Spike in variant strains. Keywords SARS-CoV-2 catalytic antibodies antigen cleavage viral neutralization Published: June 23, 2023 ==== Body pmcIntroduction Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is responsible for the current global pandemic of coronavirus disease-2019 (COVID-19). The genome of SARS-CoV-2 is ∼79% and ∼50% conserved with respect to SARS-CoV and MERS-CoV,1 two human coronaviruses that caused serious epidemics in recent years. There is a high degree of sequence conservation in the structural and nonstructural proteins encoded by SARS-CoV-2 and SARS-CoV. One notable exception is the S protein, in which divergence is observed at key receptor-engaging residues as well as three short insertions within the N-terminal domain.2 The S protein is a homotrimeric glycoprotein with two major subunits: S1 and S2. S2 comprises the highly conserved fusion machinery, while S1 mediates host receptor binding primarily through its receptor binding domain (RBD), which interacts with angiotensin-converting enzyme 2 (ACE2) receptors. The receptor binding domains in SARS-CoV-2 prefusion spike complex transiently exchange between either “open” or “closed” conformations relative to the stem of the S protein,3 where the closed conformation is incapable of receptor binding, and the open conformation is binding-competent.4 , 5 , 6 The SARS-CoV-2 RBD has a higher binding affinity for ACE2 than the homologous receptor engaging domain in SARS-CoV,7 but the former predominantly assumes the closed configuration, which may aid immune evasion by partially masking neutralizing epitopes.3 Multiple proteolytic activation steps are required to loosen the sequestered RBD “closed” arrangement into the “open” conformation to allow ACE2 binding and subsequent dissociation of the S1 subunit. This in turn triggers a large rearrangement of the S2 fusion machinery that enables the insertion of the fusion peptide into the host membrane and initiation of viral fusion.8 , 9 , 10 Owing to its accessibility on the viral surface and critical function in viral fusion to host membranes, the S protein is a rational target for immunological and pharmaceutical disruption of viral pathogenesis. However, extensive glycosylation of the S protein blankets approximately 40% of the protein surface, sterically shielding many epitopes from immune recognition.11 As the RBD is mostly devoid of glycosylation and houses the critical receptor binding residues, many neutralizing antibodies are directed to RBD epitopes. Massive vaccination campaigns and monoclonal antibody therapies targeting the S protein of the ancestral SARS-CoV-2 strain initially demonstrated remarkable efficacy.12 , 13 , 14 , 15 , 16 However, these interventions also applied significant selective pressure that yielded novel variants capable of breakthrough infections in previously immune populations.17 , 18 COVID-19 convalescent plasma (CCP) is an alternative passive antibody therapy requiring no development and low investment, and its availability is limited only by willing donors.19 Additionally, CCP provides tailored protection against specific circulating strains within a given community.20 Thus, high-titer CCP has emerged as an attractive therapeutic approach, as it reduces the risk of hospitalization and decreases the risk of mortality for hospitalized patients who do not have any detectable antibodies or who are immunosuppressed.21 , 22 , 23 , 24 , 25 , 26 , 27 Because CCP contains neutralizing antibodies, the primary mechanism of CCP activity is thought to be as an antiviral agent. Higher SARS-CoV-2 neutralizing titers are associated with increased Fc-dependent functions, specifically increased complement activation, phagocytosis, and antibody-dependent cellular cytotoxicity against SARS-CoV-2-infected cells.28 , 29 Recent work suggests that antibody-mediated antigen catalysis is yet another mechanism by which CCP can neutralize virus.30 Based on the transition-state stabilization theory of catalysis, the existence of antibodies with catalytic properties has long been theorized: Paratopes raised against transition-state analogs corresponding to specific chemical reactions should lower activation energy barriers for chemical conversions. In 1989, the first naturally occurring antibodies that hydrolyzed their cognate antigen were described, followed by many more examples of antibodies cleaving a diverse set of antigens.30 , 31 , 32 , 33 Although many natural catalytic antibodies have been linked to deleterious pathologies, others are associated with protective roles against microbes or in homeostasis.31 For example, catalytic antibodies specific to the capsule of the pathogenic fungus Cryptococcus neoformans have been demonstrated to cleave their antigen, resulting in altered complement deposition and increased phagocytosis.30 As such, we investigated whether CCP antibodies could actively proteolyze the S protein and play a protective role in antiviral immunity. Here we report that purified antibodies from CCP catalyze the cleavage of peptide epitopes and recombinant S protein fragments, including variants of concern (VOCs) which arose after these units of CCP were collected. In addition, we show that antibody proteolysis of full-length S in the context of authentic SARS-CoV-2 virus contributes a significant portion of CCP antibody neutralization capacity, suggesting an important role for this process in viral neutralization via damage of S proteins and attenuation of receptor binding. Results Design of peptide epitopes based on interfacial residues of RBD The ACE2 interaction region of the S protein, known as the receptor binding module (RBM; Ser438-Pro507), is an extended insertion into the anti-parallel β-sheet core fold of the RBD34 , 35 , 36 (Figure S1A). The RBD-ACE2 interaction is dominated by extensive interactions with the N-terminal helix of ACE236 (Figure S1C). Antibodies with epitopes that overlap with the RBM binding site are thought to be neutralizing via competitive binding with ACE2. The complete RBM epitope is fully accessible only in the “RBD-open” conformation, but with specific angles of approach and subepitopes, certain antibodies are capable of binding regardless of RBD conformation37 , 38 , 39 (Figures S1B and S1C). Such antibodies, belonging to the class II binding mode proposed by Finkelstein et al.,38 are extremely versatile RBD binders. As such, we focused on the portion of the RBM which overlaps with the class II nAb epitope and designed two peptides intended to mimic the nAb-RBD epitope interface: RBM epitope peptide 1 (RBM1) consists of residues Lys444-Leu452 (KVGGNYNYL) and RBM2 consists of residues Val483-Ser494 (VEGFNCYFPLQS) (Figures 1B and S1C). The biological interface between RBD and ACE2 encompasses 864 Å2 of surface area, of which RBM1 and RBM2 contribute 10% and 39% of the total interface, respectively. Of the five critical residues for ACE2 engagement, RBM2 contains three (F486, Q493, and S494) and RBM1 has none.1 With respect to the immunological interface formed with a representative class II neutralizing antibody P2B-2F6, RBM1, and RBM2 encompass interfaces of 329 and 327 Å2, respectively, which together account for the complete interface (Figure S1C). Thus, cleavage of the RBM2 sequence would likely result in direct disruption of ACE2 recognizing residues, while cleavage of the more readily accessible RBM1 site could exert indirect structural changes that may interfere with ACE2 engagement.Figure 1 Antibody-mediated antigen FRET proteolysis assay with purified IgG from CP donors (A) Schematic of FRET proteolysis principle employed in this assay. Mca fluorophore and Dnp quencher pairs were appended to the N- and C-termini of each RBM epitope peptide. Intact peptides do not fluoresce when excited by a 320 nm laser line due to resonance energy quenching by the covalently attached Dnp quencher. When the intervening peptide backbone is cleaved, quenching is relieved and fluorescence is observed. (B) The locations of the two FRET peptide sequences on the tertiary structure of the receptor binding domain are highlighted (RBM1, yellow; RBM2, green). (C) FRET kinetic traces illustrate the relative activity of untreated and heat-treated convalescent plasma from four different donors (labeled CCP1-4, 2-fold dilution of CCP). See Figure S2 and Table S1 for full CCP kinetic data. (D) The concentration of purified CCP3 antibodies was titrated from total antibody concentrations of 10 mg/mL to 156 μg/mL to measure the dose dependence of antibody concentration on peptide cleavage. See Table S2 for full kinetic CCP IgG kinetic data. Each kinetic curve is displayed as the mean ± S.D. of 2 replicates, and fluorescence measurements were converted to units of molarity of proteolyzed peptide epitope using each corresponding positive control. (E) The kinetic traces of peptide proteolysis mediated by CCP3 and a set of six nonimmune plasma samples are overlaid, where 5 mg/mL of purified total IgG from each plasma sample was incubated with RBM FRET epitopes. Cleavage of RBM1 and RBM2 epitopes are displayed on the left and right, respectively. See Figure S5 and Table S3 for full nonimmune IgG kinetic data. High-throughput assay to detect RBM peptide epitope proteolytic cleavage These two RBM epitopes were synthesized as quenched fluorescent peptides to generate sensitive fluorescent resonance energy transfer (FRET) reporters of RBM-specific proteolysis. Each RBM peptide was appended with an Mca fluorophore and Dnp quencher at its N- and C-termini, respectively (Figure 1A). In a pilot screen to assess the catalytic cleavage of RBM-derived peptides by convalescent antibodies, we selected four random aliquots of CCP units obtained near the onset of the pandemic (June 2020). These CCP units varied markedly in specific antibody titers (∼5 ng/mL–5 μg/mL, Table S1), which was expected given that neutralizing titers are highly variable among individuals who have recovered from COVID-19.40 Each CCP unit was incubated with a single quenched fluorescent peptide under physiological conditions (37°C) for 5 h. Fluorescence was monitored over time and converted to molarity of substrate using the maximal fluorescence observed in proteinase K control reactions, which represent complete proteolysis of each peptide. RBM cleavage was observed for every CCP sample in our pilot study, although cleavage velocities varied (Figures 1C and S2, and Table S1). Further, catalytic activity did not increase in a dose-dependent manner, but instead diminished at higher concentrations of convalescent plasma (Figures S2 and S3, and Table S1). Such impairment of immune complex formation at exceedingly high antibody concentrations, which results in lower activity for concentrated samples above a critical threshold, is known as the hook or prozone effect.41 , 42 , 43 Observation of this phenomenon in our kinetics data strongly suggests an antibody-mediated competition phenomenon. Next, we sought to confirm that the observed cleavage was antibody-mediated, rather than from spurious cleavage from canonical proteases or other catalytic components found in human plasma. Non-immunoglobulin components in CCP were denatured by heat treatment at 55°C, a temperature routinely used to inactivate complement proteases while sparing highly thermostable antibodies.44 , 45 A comparison of the activities from heat-treated versus untreated plasma samples demonstrated that 40–55% of activity was retained after heat treatment (Figure S2 and Table S1). The attenuation of observed proteolysis may also be due to the reduction in specific antibody catalysts, as 2-fold reductions of IgG and IgM titers were observed following similar heat-inactivation procedures.46 Proteolysis rates of RBM2 across the four CCP units tested here were, on average, 32% faster than for RBM1 for untreated plasma, and 86% faster than RBM1 cleavage rates by heat-treated plasma. CCP3 has a relatively high proportion of RBD-specific IgG and exhibited up to 40% faster initial velocities of RBM proteolysis than the other CCP tested (Table S1). Interestingly, the proportion of RBD-specific antibody did not directly determine catalytic activity in the heat-treated CCP samples characterized here, as CCP containing lower levels of RBD-specific antibody demonstrated comparable cleavage activity to higher specific titer plasma (Figure S3). Heat-inactivated nonimmune plasma obtained from a donor never exposed to SARS-CoV-2 proteolyzed RBM1 and RBM2 at velocities 97% and 34% of that observed for CCP3 (Figure S4). RBD-specific antibody was depleted from CCP3 using RBD-conjugated magnetic beads, resulting in a 2.2-fold reduction of RBD-specific immunoglobulin (Figure 2A). Peptide cleavage by α-RBD Ab-depleted CCP3 relative to untreated CCP3 was attenuated by 6% and 27% for RBM1 and RBM2, respectively (Figures 2B and 2C). Thus, RBD-specific antibody has an outsized contribution to overall catalysis given that it comprised only 0.04% of the total IgG in CCP3.Figure 2 RBD-specific antibody depletion of CCP3 by adsorption to magnetic beads results in attenuation of cleavage activity (A) Depletion of RBD-specific antibody from heat-treated CCP3 using RBD-conjugated magnetic beads was confirmed by quantitative ELISA. Absorbance values at 405 nm for untreated (black) or bead-depleted CCP3 (red) are graphed as the mean ± S.D. of 2 replicates for each successive serial dilution and calculated RBD-specific IgG concentrations are indicated. (B) FRET kinetics of heat-treated CCP3 compared to heated-treated CCP3 depleted of RBD-specific antibody with RBD-conjugated magnetic beads. Fluorescence values for non-depleted CCP3 (black) and bead-depleted CCP3 (red) are displayed for the 2-fold dilution. Each sample was measured in duplicate. (C) Calculated initial velocities for depleted and non-depleted CCP3 at each serial dilution tested are displayed, with the percent initial velocity of RBD-depleted CCP3 with respect to the corresponding dilution of non-depleted CCP3 are indicated with red labels. Next, to evaluate immunoglobulin-mediated catalysis separately from any proteolytic contributions from non-antibody components in plasma, we purified IgG from CCP3, which had the highest RBD-specific antibody titers and fastest RBM2 cleavage velocity after heat treatment. We subjected this affinity-purified polyclonal CCP3 IgG to the same FRET screen using a total IgG concentration in the range of 156–10,000 μg/mL (0.04–2.78 μg/mL α-RBD IgG). We observed significant catalytic activity in the purified antibody sample, which increased in a dose-dependent manner, as total antibody concentrations were kept below the prozone threshold (Figure 1D and Table S2). Affinity-purified IgG from six nonimmune plasma donors was also characterized to establish a baseline of the cross-reactive cleavage of nonimmune IgG. We measured initial proteolytic velocities for RBM1 and RBM2 in the ranges of 0.0005–0.024 μM/h and 0.002–0.078 μM/h, respectively (Figures 1E and S5, and Table S3). Compared to the initial velocities observed by antibody isolated from CCP3 at the same total antibody concentration (0.165 and 0.255 μM/h for RBM1 and RBM2), this represents 0–15% and 1–30% of the observed velocity by CCP3 for each epitope. Importantly, we noted that the relationship between the concentration of α-RBD IgG and peptide cleavage velocity observed for the affinity-purified CCP3 IgG was comparable to that of heat-treated total CCP3, indicating that heat treatment effectively separates nonspecific cleavage by serum proteases from antibody-mediated cleavage and can be used as a proxy for cleavage by pure IgG in larger screens where individual purification is impractical (Figure S3). RBM proteolysis correlates with CCP neutralization and other attributes After establishing a robust high-throughput FRET assay for RBM epitope cleavage, we deployed our FRET proteolysis assay on two collections of diverse CCP samples to ascertain if antibody-mediated cleavage efficiency corresponded to relevant immunological characteristics. In the first library, heat-treated CCP from 46 random donors was screened for proteolytic cleavage of the RBM peptides. Significant inverse relationships were observed for cleavage of both RBM peptides and patient age (R = −0.387, p = 0.008 and R = −0.719, p < 0.001 for RBM1 and RBM2, respectively), and the negative correlation of peptide cleavage to time since diagnosis closely approached the significance threshold (R = −0.282, p = 0.057 and R = −0.281, p = 0.059) (Figure S6A). However, there was no significant correlation between S1-specific antibody titers measured by EuroImmun and RBM cleavage (Figure S6A). We then screened a second library of 126 CCP samples40 to investigate the relationship between antibody-neutralizing function and cleavage activity. We observed a weak positive correlation between total RBM1 cleavage and microneutralization assay area under the curve (AUC) values (R = 0.178, p = 0.047), indicating that antibody-mediated cleavage may contribute to the neutralization of SARS-CoV-2 (Figure S6B). We also found a correlation between RBD-specific antibody titers and the RBM1 cleavage, supporting the notion that cleavage is associated with a specific antibody. Antibody-mediated epitope cleavage is catalyzed through several mechanisms To gain a better understanding of the types of proteolytic mechanisms employed by catalytic antibodies in the in polyclonal mixture, we subjected affinity purified antibody from CCP3 to a protease inhibitor screen. We assessed the relative inhibition of five specific protease inhibitors to gain insight into the contribution of each mechanism of action in the catalytic antibody population: AEBSF (serine proteases), bestatin (aminopepsidases), E−64 (cysteine proteases), and pepstatin-A (aspartic acid proteases). We did not observe any appreciable inhibition by E−64 or pepstatin-A, but measured a 42% and 13% inhibition of RBM2 cleavage was observed for AEBSF and bestatin, respectively, indicating that a fraction of antibodies in the convalescent antibody population utilize serine protease and aminopepsidase-like mechanisms (Figure 3 ).Figure 3 Protease inhibitor screens yields mechanistic insights in antibody catalytic mechanism (A and B) RBM epitope probes were incubated with CCP3 IgG in the presence of a set of protease inhibitors to gain insight into the mechanism of action. Kinetic traces of the reaction in the presence of AEBSF, bestatin, E−64, pepstatin-A, and without inhibitor are displayed in blue, green, purple, black, and red, respectively. Panel A and B show the mean cleavage of two replicates at 5 mg/mL CCP3 IgG for RBM1 and RBM2 peptide probes, respectively (C) A heatmap of the initial velocities of each reaction + inhibitor at 5 mg/mL, normalized to the corresponding uninhibited velocity. (D) Full table of initial velocities of cleavage for CCP3 in presence of each inhibitor at all antibody concentrations. CCP antibodies cleave authentic S protein and recombinant fragments After determining that convalescent antibodies cleave RBD peptide epitopes and that this activity was associated with virus neutralization across a large CCP library, we investigated whether the catalytic antibodies could also cleave full S proteins in more stably folded tertiary configurations. Incubation of purified CCP3 antibodies with recombinant RBD (rRBD) or S1 (rS1) protein domains for 7 days resulted in clear proteolytic degradation of the full-length protein (Figures 4A and 4B). The residual band intensity of intact rRBD decreased in a dose-dependent manner with respect to rRBD incubated at 37°C in the absence of antibody. Incubation with CCP3 antibody in the range 250–2000 μg/mL total IgG (70–560 ng/mL α-RBD IgG) resulted in 20–78% degradation of rRBD. Under the same conditions, rS1 protein was degraded 21–59%. Under these reaction conditions, purified CCP3 antibody demonstrated a range of cleavage rates from 0.45–1.75 nM/h and 0.82–2.30 nM/h when rRBD or rS1 were used as substrates, respectively. Further, we confirmed that the proteolysis observed in these assays was specific to S proteins, as no detectable degradation was observed when we incubated an irrelevant protein, bovine serum albumin, in similar reaction conditions with purified antibodies from three different CCP units (Figure S7). To confirm that the observed activity extended to full-length S protein on authentic SARS-CoV-2 virus particles, we examined the effect of CCP3 IgG incubation with inactivated SARS-CoV-2 virus (Figure 4C). Although the degree of degradation was less than observed for recombinant S proteins in similar conditions (0–15% band reduction compared to no-antibody control), we also observed the concomitant accumulation of degradative S fragments.Figure 4 Recombinant RBD and S1 is degraded by purified antibody from CCP3 Digestion reactions of recombinant S protein fragments were separated by SDS-PAGE and probed for residual S protein by immunoblot. The relative amount of intact S protein remaining after each digestion reaction was quantified by normalizing band intensity of rRBD or rS1 or authentic viral S in the no-antibody control to each experimental reaction. (A) Recombinant RBD (10 μg/mL) was incubated with 250–2000 μg/mL purified IgG from CCP3 for a period of 7 days. (B) Recombinant S1 (50 μg/mL) was incubated with 250–2000 μg/mL purified IgG from CCP3 for a period of 7 days. (C) Inactivated SARS-CoV-2 virus (50 μg/mL) was incubated with 250–1000 μg/mL CCP3 antibody for a period of 10 days. Each reaction was run in duplicate and the location of S protein monomer, multimers and proteolytic fragments is indicated. Below each panel (A–C), the relative amount of intact S remaining in each digestion reaction with a gradient of purified IgG from CP was quantified by normalizing band intensity of S in the no-antibody control the same incubation. In panel C) the relative band intensity of proteolytic degradation species (defined as cumulative band intensity below the S protein monomer) was quantified and noise-corrected by subtracting intensity measured in the same search area for the no-antibody control for full virus incubations. Molecular weight ladders from the aligned Commassie-stained gels are shown for reference and labeled with L. The molecular weight marker in panel A refers also to panel B, as both experiments were on the same gel. Antibody proteolysis enhances virus neutralization Given that CCP antibody proteolyzes S protein on authentic SARS-CoV-2, we next sought to directly measure the impact of antibody catalysis on the ability of the virus to infect human cells. As traditional human enzymes have optimal activity at physiological temperature, we reasoned that lowering the temperature could substantially reduce catalysis by antibodies. Indeed, antibody-mediated cleavage of rRBD was significantly attenuated below physiological temperatures (<37°C), and nearly undetectable at 4°C (Figures 5A and 5B). Thus, we designed a plaque reduction neutralization test (PRNT) protocol that involved differential temperature preincubations of antibody with virus to separate antibody binding of epitopes from antibody-mediated epitope proteolysis (i.e., incubation with virus at physiological temperature activates antibody-mediated proteolysis, while incubation at 4°C allows binding but inhibits proteolysis). The incubation period of virus and antibody was also extended from the 1 h incubation used in standard viral neutralization assays to 6 h to allow sufficient proteolysis to occur. Given that antibody-antigen binding is an enthalpically driven reaction, we hypothesized that binding should be more favorable at a low temperature given the lessening effect of the temperature-weighted entropy term. This was confirmed by indirect ELISA where the antibody was bound to RBD antigen under the same conditions as in the modified PRNT, revealing that binding was roughly equivalent at the two temperatures (Figure 5C). Therefore, antibody-viral complexes should form with equal affinity at the two different temperatures used in this assay and differences in neutralization activity reflect only the contribution of antibody-mediated catalysis.Figure 5 Temperature dependence of RBD proteolysis by CCP enables separation of binding and catalysis (A) Recombinant RBD was incubated with affinity-purified IgG from two different CP donors (CCP1 and CCP3) for 7 days at three different temperatures, and the reactions were visualized by immunoblot. The RBD band intensities were quantified and compared for each digestion reaction at each temperature. Molecular weight ladders from the aligned Commassie-stained gels are shown for reference and labeled with L. (B) Quenched fluorescent RBM epitope peptides were incubated with serially diluted, complement-treated CCP3 antibody for 16 h and endpoint fluorescence was measured to assess peptide cleavage. Each reaction was run in duplicate on separate plates which were incubated for the same time at 4, 25, or 37°C. The average total endpoint peptide cleavage observed across all dilutions at 25°C and 4°C was 0.51- and 0.22-fold of that seen at 37°C, respectively for RBM1, and 0.59- and 0.25-fold for RBM2. Fluorescence values, normalized to 37°C at each dilution, are labeled in each cell. (C) CCP binding of rRBD is comparable at 4 or 37°C. Antibody binding at the two different temperatures was measured by indirect ELISA, where heat-treated CCP3 (left) or purified IgG from CCP3 were incubated with RBD substrates at the two temperatures and binding efficiency was measured. Measured absorbance values at 405 nm as plotted as the mean ± S.D. of 2 replicates with respect to dilution of serum or antibody. Modified PRNT assays were performed at three different temperatures using heat-treated CCP3 with SARS-CoV-2 A.3 isolate HP00076. Samples incubated at 37°C achieved a geometric mean IC50 of 619.8 (95% CI 404.9, 948.8), while the 4°C and room temperature (RT) were found to be 137.5 (95% CI 93.75, 201.8) and 118 (95% CI 54.75, 254.1), respectively (Figures 6A and 6B). The 4°C and RT showed statistically significant lower geometric mean IC50 values compared to 37°C, with the fold difference ranging from 1.16- to 5.25-fold. Next, purified IgG from CCP3 was tested in the same live virus neutralization assay. The geometric mean for the samples incubated at 37°C was IC50 of 770 μg/mL (inverse IgG dilutions; 95% CI 274.8, 2157), followed by 208.4 μg/mL (95% CI 131.9, 329.2) and 259.5 μg/mL (95% CI 253.2, 265.9; Figure 6C) for RT and 4°C. The 4°C and RT showed statistically significant lower geometric mean IC50 values than 37°C, with the fold difference ranging from 1.25- to 3.69-fold. The finding that the fold-increase observed for heat-treated CCP3, containing the full complement of antibody subtypes, was greater than purified polyclonal IgG suggests that antibodies of other isotypes are similarly catalytically active.Figure 6 Viral neutralization correlates with antibody-mediated S protein damage The CCP3 convalescent plasma or purified IgG samples were tested against three different temperatures using A.3 lineage isolates to determine virus neutralization activity. (A) Serial dilutions of heat-treated plasma were assessed for their ability to inhibit plaque formation, with plaque reduction normalized to no serum controls at the same dilution. IC50 (half maximal inhibitory concentration) was calculated using the inhibition-dose response of plaque reduction over 2-fold serial dilutions of plasma starting from 1:20 using (B) heat-treated CCP3 or (C) purified IgG from CCP3. Geometric mean titer and 95% confidence intervals are indicated and the geometric mean fold drops between temperatures is denoted above each dataset. CCP3 heat-treated plasma and affinity-purified CCP3 IgG were assessed with four and two technical replicates, respectively. Statistical significance was measured using the Mann Whitney matched rank test, ∗∗p = 0.0286. Variant S proteins are still susceptible to proteolysis by CCP obtained prior to variant emergence Given that the four main CCP samples analyzed in this study were acquired near the beginning of the pandemic when the ancestral SARS-CoV-2 strain predominated, we were interested in whether substitutions found in the S protein from subsequent VOC (VOC) affected antibody-mediated proteolysis. Mutations frequently associated with VOC within the RBM epitopes used in this study (residues K444-L452 and V483-S494) were selected for further characterization in the established RBM proteolysis FRET assay: S494P, E484K, and L452R (Figure 7A). These mutations correspond to VOCs Alpha B.1.17 (S494P), Beta/Gamma (E484K), and Delta (L452R). RBM1 and RBM2 variant sequences were again synthesized as quenched fluorescent peptides to assess the reactivity of these altered peptide epitopes. While affinity-purified CCP3 IgG proteolyzed the RBM1 epitope containing a. L452R mutation with a velocity comparable to the wildtype peptide sequence, E484K and S494P (RBM2) mutations were completely unreactive in the FRET assay (Figure 7B and Table S4). Next, rRBDs of the corresponding variants were subjected to the same immunoblot proteolysis analysis as before. In contrast to the reactivity pattern observed in variant peptides, the degree of CCP3 antibody proteolysis of each corresponding variant rRBD domain was equivalent, or even greater, to that of the wild-type rRBD (Figure 7C). However, the relative reduction in band intensity of variant rRBDs incubated without antibody suggested that variant rRBDs, especially Omicron rRBD, have intrinsically lower thermostability.Figure 7 Antibody-mediated S protein epitope cleavage for VOC The susceptibility to antibody-mediated cleavage of quenched fluorescent peptides encoding amino acid substitutions within the RBM1 and RBM2 sequences found in VOC was determined by measurement of fluorescence over time. (A, top) The location of the mutations of interest on the structure of RBD are highlighted as spheres and colored dark blue, yellow, and cyan for L452R, E484K, and S494P substitutions, respectively. The RBD is displayed as dark gray surface representation and RBM1 and RBM2 peptide epitope locations are highlighted in red, and the entire RBM is colored white. (A, middle) RBD is displayed in the same orientation as the top panel and colored dark gray. Alpha substitutions are highlighted in cyan, Beta/Gamma substitutions in yellow, Delta substitutions in dark blue, and Omicron BA.1 substitutions in green. (A, bottom) The RBD with the same color scheme and orientation as above is displayed in complex with ACE2 (PBD 6M0J). (B) Kinetic FRET traces observed when purified CCP3 IgG at concentrations ranging from 10 mg/mL to 156 μg/mL was incubated with each variant RBM FRET peptide for 5 h. See Table S4 for full proteolytic kinetics of variant RBM peptides. Each kinetic curve is displayed the mean ± S.D. of 2 replicates, and fluorescence measurements were converted to units of molarity of proteolyzed peptide epitope using each corresponding positive control. (C) Antibody-mediated proteolysis of recombinant RBD. protein with S494P, E484K, L452R/T478K, or Omicron substitutions is compared to proteolysis of wild-type recombinant RBD. Recombinant RBD (10 μg/mL) was incubated with CCP3 IgG (500 μg/mL) or alone at 37°C for 7 days. Band intensities of RBD following the incubation period, normalized to the corresponding sample at time zero, are labeled for each digestion reaction. Molecular weight ladders from the aligned Commassie-stained gels are shown for reference and labeled with L. Discussion Our study demonstrates that antibodies in CCP exhibit proteolytic activity specific to the S protein of SARS-CoV-2, contributing to their antiviral function. Using peptide substrates derived from the RBM epitope of the RBD of SARS-CoV-2, an FRET screen was established to rapidly screen CCP for epitope cleavage. We demonstrated that cleavage is an antibody-mediated process by measuring the proteolytic capacity of CCP, heat-treated plasma, and affinity-purified IgG derived from CCP. The RBM2 epitope (40% of the ACE2 interface) was almost 2-fold as reactive as RBM1 (Class II nAb interface and partial ACE2 interface). Interestingly, the α-RBD IgG concentrations in the four CCP units in our initial screen did not correlate well with proteolytic efficiency, as we observed that samples with very low α-RBD titers demonstrated measurable cleavage of both RBM epitopes. However, several lines of evidence strongly suggest that RBD-specific antibodies drive the observed proteolysis. First, the relative cleavage of epitope probes by CCP IgG compared to nonimmune IgG was significantly faster. Second, the depletion of specific antibody from plasma resulted in significant reductions in cleavage velocity, despite the very low fraction of specific antibody in the total Ig pool. We also saw a weak but significant correlation between RBM1 cleavage and RBD-specific titers in our second CCP library and no correlation for RBM2. The lack of strong correlation could be partially explained by the specific probes used in this study, which only report on cleavage at two specific RBD epitopes. Catalytic activity of α-RBD Abs recognizing alternative epitopes, or conformational epitopes contained within the probe sequences, is not captured by this screen. In particular, the RBM2 sequence features a disulfide linkage which enforces a sharp turn in the peptide backbone in the protein but is absent from the peptide probe. Further, this could reflect either a greater ratio of antibodies specific to the RBM epitopes used on the screen or a greater ratio of antibodies with catalytic properties, in the low-titer samples. It may also suggest some cross-reactivity of existing antibodies directed against endemic coronavirus.47 , 48 Alternatively, it has been posited that antibodies which undergo positive selection are less likely to be catalytic, as antigenic occupancy of B cell receptors would be reduced if antigen is proteolyzed, thus disfavoring clonal expansion during the B cell differentiation process.49 , 50 , 51 Indeed, our findings that nonimmune plasma and depleted plasma still exhibits observable RBD epitope proteolysis suggests that constitutively expressed naive antibodies may account for some nonspecific cleavage. However, the enhancement of RBD cleavage observed in high-titer CCP indicates that proteolytic activity is not completely eliminated in antibodies that undergo affinity maturation, and that greater specificity and titer may enhance the activity of such antibodies compared to naive immunoglobulins. Analysis of protease inhibitor screen revealed the potent inhibitory effect of the serine protease inhibitor AEBSF, suggesting that a large proportion of catalytic antibodies found in CCP3 use cleavage mechanisms similar to conventional serine proteases. A smaller fraction of catalytic antibodies is also inhibited by aminopepsidase inhibitors. Serine protease-like mechanisms for catalytic antibodies have been previously hypothesized, and catalytic triads reminiscent of serine protease active sites were identified in nearly a quarter of annotated catalytic antibodies with deposited structures.32 On the other hand, aminopepsidases only cleave at the amino terminus of polypeptide chains, and as such would require an initial cleavage by an endoprotease in order to further degrade the RBD. Interestingly, the aggregate effect of all inhibitors in this screen (57%) would not completely abrogate the activity of CCP3, suggesting that there are additional noncanonical, antibody-specific mechanisms at play that are not attenuated by traditional inhibitors.52 Using purified antibody from a CCP sample with high specific antibody titer and efficient proteolysis kinetics (CCP3), we observed extensive antibody-mediated proteolysis of recombinant S protein domains. Importantly, incubation of authentic SARS-CoV-2 virions with CCP3 IgG likewise resulted in S protein fragmentation. The total extent of CCP3 IgG cleavage of viral S was less than that of recombinant S domains (15% vs. 51–70% degradation of authentic virus and recombinant S1/RBD catalyzed by 1 mg/mL CCP3 IgG). This could be attributable to the β-propiolactone deactivation method used, which can introduce stabilizing crosslinks to the S protein52 , 53 or partial shielding of epitopes in the trimeric S configuration. Regardless, when viral neutralization by CCP3 was directly assessed using a modified PRNT assay, we observed significantly more potent neutralization at the catalysis-permissive temperature compared to the low temperature which permits only antigen binding. Furthermore, antibody-mediated proteolysis of the S protein appears to be a widespread phenomenon, as we observed RBM proteolysis across two large libraries of CCP and measured a correlation between antibody neutralization capacity and proteolytic efficiency of RBM epitopes. We note that this correlation may be improved if the extended antibody incubation time employed in this study (to allow for more extensive antigen cleavage) was used in the original virus neutralization measurements of the library. Additionally, our observation of an inverse correlation between proteolysis and donor age is intriguing because younger patients generally do not exhibit severe COVID-19 symptoms.54 , 55 Antibody-mediated proteolysis must necessarily work in concert with conventional blocking mAbs to exert relevant influence on the neutralization process. As antibody neutralization by blocking works on the order of minutes,56 , 57 the kinetics of S protein damage measured in vitro might at first glance seem too slow to contribute meaningfully to neutralization on the same timescale. However, while the kinetics of this process are slow compared to traditional proteases, these activities can exert significant immunological effects in vivo, given the long half-life and high titers of antibodies in circulation31 and considering that the amount of viral substrate relative to catalyst is quite low compared to the substrate concentrations processed natural enzymes. By extrapolating the in vitro proteolysis kinetics measured in this study, we calculated that every S protein in the human body at peak viral load could be proteolyzed within seconds assuming complete substrate accessibility (see Note S1 for detailed kinetics argument). Indeed, our in vitro PRNT data indicates that IC50 for CCP3 decreases 4.5- to 3-fold when proteolysis is inhibited in heat-treated plasma or purified IgG, respectively. Hence, 67–78% of the virus was neutralized via proteolysis in catalytic conditions, as only 22–33% of the antibody was required to achieve IC50 values equivalent to the same antibody in catalysis-inhibited conditions. An even more important role of antibody-mediated proteolysis may be in the destruction of epitopes undergoing antigenic drift, such as variant viral proteins. As such, we were particularly interested if S protein substitutions from important VOC escaped the proteolytic activity of CCP antibodies elicited before variant emergence. When RBM variant peptides were analyzed, no measurable cleavage was observed when S494P and E484K substitutions were introduced, likely due to abrogation of binding affinity for these altered peptide epitopes. This finding also provides further evidence that catalytic antibodies have a high degree of specificity to their cognate epitopes. However, the CCP antibody did cleave the Delta variant RBM peptide (L452R) with comparative efficiency to the wild-type sequence, implying CCP-mediated proteolysis does not rely on the leucine at this position for recognition. Therefore, at the peptide level, certain residues in the S sequence are critical for recognition and proteolysis, while other mutations can be tolerated, likely due to the relative prevalence of certain epitopes targeted by the polyclonal antibody population. Conversely, on the level of S protein domains, we observed that rRBD variants were degraded even more efficiently than the wild-type RBD. Cleavage of Beta/Gamma and Delta rRBD, but not the corresponding RBM peptides is not unexpected, given that the RBM epitopes comprise only a small subset of the total available RBD epitopes, many of which are likely cleaved by cross-reactive catalytic antibodies specific to conserved epitopes in CCP. Conversely, the enhancement of variant rRBD degradation with respect to wild-type protein was surprising. Paradoxically, while these substitutions are selected for by weakening the binding of non-catalytic antibodies, they may result in RBD structures that are better substrates for catalytic antibodies due to slight structural perturbations which may destabilize the domain and enhance the proteolytic susceptibility of certain epitopes. For example, the heavily mutated Omicron RBD is markedly destabilized, but the destabilization is partially compensated for by additional mutations in S2 or N-terminal domain in the context of the full S ectodomain.58 , 59 While S protein substitutions rapidly weaken antibody binding and thus neutralization by competitive binding, antibody proteolysis is more durable, perhaps because lower-affinity transient binding events are sufficient to permit proteolysis. In agreement with this model, it was recently reported that CCP from donors never exposed to newer variants can nevertheless neutralize novel SARS-CoV-2 variants.60 Further, CCP obtained more recently and from the same geographic community would likely contain S-specific antibody with higher binding affinity to the specific circulating S protein variant.20 Our findings are relevant to the interpretation of results from earlier studies of antibody-mediated viral neutralization, which implicated the destabilization of viral proteins driven by immune complexation with antibodies. Early evidence for viral destabilization by antibodies included the finding that a low proportion of mAbs specific to poliovirus type 1 was capable of “disruptive” neutralization which was irreversible by dissolution of the immune complexes and resulted in the release of genetic material from the virion.61 , 62 Two separate groups also reported that a single neutralizing antibody was sufficient to neutralize 2–4 receptor proteins in influenza and flavivirus.63 , 64 In the light of antibody-mediated viral proteolysis reported here, one could imagine that catalytic turnover may account for the non-stoichiometric neutralizing capacity exerted by each antibody in those studies. More recently, neutralizing antigen-binding fragments (Fabs) were shown to completely dissociate the envelope glycoprotein gp120 trimer of human immunodeficiency virus type-1 into smaller fragments, which were shed from the virion.65 Although not specifically implicated, antibody-mediated proteolysis could explain that destabilization. In a separate study, the generation of proteolytic fragments of gp120 catalyzed by purified IgM and IgA was directly observed using biotinylated substrates.49 , 66 Catalysis by natural antibodies is not limited to viral epitopes, as cleavage of Staphylococcus aureus extracellular fibrinogen-binding protein (Efb) has also been reported.51 These studies are consistent with the notion that antibody-mediated catalytic damage of important microbial virulence factors is likely a common occurrence. Limitations of the study In this study, proteolysis screening of CCP libraries was limited to the peptide epitopes selected. RBM1 and RBM2 peptides were adopted based on coverage of common epitopes of known neutralizing antibodies as well as the known functional importance to the interaction of the spike protein with the ACE2 receptor. Inclusion of additional peptide sequences derived from spike, including non-neutralizing epitopes, would provide information about the contribution of antibody proteolysis to neutralization for other domains within S1 or S2 subunits. Expansion of the epitope probe library could capture more of the total catalysis performed by CCP antibodies and is likely to improve the statistical correlations between catalytic activity and neutralization titers for each CCP unit. Further, the use of peptides as cleavage targets may underestimate the total cleavage of a particular Spike region if antibodies recognize conformational epitopes within that sequence. Additionally, though we focused our in vitro characterization of antibody catalysis on IgG derived from CCP in this study, it is plausible that antibodies of other isotypes also contribute to the overall catalysis. In this regard, we note that much of the viral neutralization capacity of CCP resides in the IgM and IgG fractions, the former of which was not specifically evaluated in this study. This analysis was limited to the bulk properties of polyclonal CCP Abs, a small fraction of which are specific to spike. Within the specific Ab fraction, it is plausible that only rare mAbs are responsible for catalytic activity. Analysis of monoclonal Abs could address this question and establish critical determinants of specific paratope or epitope sequences that enable antibody catalysis. Significance Traditionally, antibody-mediated viral neutralization is thought to occur by passive binding mechanisms. Antibodies can directly interfere with critical viral processes such as host receptor binding, conformational changes associated with membrane fusion, or the release of viral genome into the cytoplasm. Antibodies also indirectly recruit complement or effector cells to clear circulating virions. 67 , 68 However, our finding that antibody-mediated proteolysis of the receptor-engaging machinery of SARS-CoV-2 is associated with virus neutralization implies an additional active role for immunoglobulins in combating viral infections. For example, low-affinity antibodies with significant off-rates have limited neutralization potency due to the reduced timescale of competitive blocking. However, if proteolysis occurs before dissociation, viral deactivation is achieved via irreversible destabilization of viral virulence factors. Furthermore, whereas non-catalytic antibodies deactivate virus stoichiometrically, catalytic antibodies can turn over and cleave multiple molecules per antibody paratope, thus amplifying their potency. Finally, catalytic antibodies which recognize non-neutralizing epitopes can still directly impact neutralization by destabilizing the S protein tertiary and quaternary structure and would not be limited by the conformational shielding of the important epitopes, such as the RBM. Taken together, we predict that antibody-mediated catalysis is an immunologically relevant feature of the humoral immune response which works in concert with conventional antiviral activities of antibodies. Our findings suggest that characterization of the catalytic capacity of convalescent plasma or monoclonal antibodies could provide insights into an additional axis of antibody attributes that drives viral protection. STAR★Methods Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Polyclonal Rabbit RBD antibody Sino Biological Cat#40592-T62; RRID:AB_2927483 Europium-labeled Goat α-Mouse IgG Molecular Devices Cat#R8208 HRP-conjugated Goat anti-Rabbit Ig Southern Biotech Cat#4010-05; RRID:AB_2632593 Monoclonal Anti-SARS Coronavirus/SARS-Related Coronavirus 2 Spike Glycoprotein Receptor Binding Domain (RBD) Sino Biological Cat#40150-D001 Goat α-Human IgG H + L-AP Southern Biotech Cat#2040-04; Bacterial and virus strains SARS-CoV-2 clade A.3 isolate SARS-CoV-2/USA/MD-HP00076/2020 Division of Medical Microbiology, Department of Pathology, Johns Hopkins School of Medicine, Baltimore, Maryland, USA GISAID accession number: EPI_ISL_438234 Biological samples Patient convalescent plasma Johns Hopkins Hospital, Department of Pathology, Transfusion Medicine Division See Table S527,40,69,70 Chemicals, peptides, and recombinant proteins Synthetic quenched fluorescent peptide (RBM1) Peptide 2.0 [7-methoxycoumarin-4-acetic acid] (Mca) – KVGGNYNYL – K(Dnp) [2,4-dinitrophenyl] Synthetic quenched fluorescent peptide (RBM2) Peptide 2.0 [7-methoxycoumarin-4-acetic acid] (Mca) – VEGFNCYFPLQS – K(Dnp) [2,4-dinitrophenyl] Proteinase K Invitrogen Cat#25530049 AEBSF Thermo Scientific Cat#78431 Bestatin Thermo Scientific Cat#78433 E-64 Thermo Scientific Cat#78434 Pepstatin-A Thermo Scientific Cat#78436 SARS-CoV-2 Spike RBD-coupled magnetic beads Acro Biosystems Cat#MBS-K002 Recombinant RBD (wild-type) Sino Biologicals Cat#40592-V08H Recombinant RBD (S494P) Sino Biologicals Cat#40592-V08H18 Recombinant RBD (E484K) Sino Biologicals Cat#40592-V08H84 Recombinant RBD (L452R/T478K) Sino Biologicals Cat#40592-V08H90 Recombinant RBD (Omicron; G339D/S371L/S373P/S375F/K417N/N440K/G446S/S477N/T478K/E484A/Q493R/G496S/Q498R/N501Y/Y505H) Sino Biologicals Cat#40592-V08H121 Recombinant S1 (wild-type) Sino Biologicals Cat#40591-V08H SuperSignal West Pico PLUS Chemiluminescent Substrate Fisher Scientific Cat#34577 para-Nitrophenylphosphate (PNPP) Sigma-Aldrich Cat#S0942-100TAB β-propiolactone Millipore Sigma Cat#P5648-1ML DMEM Gibco-Thermo Fisher Cat#11965118 Fetal Bovine Serum Gibco-Thermo Fisher Cat#F2442-500ML Glutamine Life Technologies Cat#25030081 Sodium Pyruvate Sigma Cat#S8636-100ML Penicillin Quality Biologicals Cat#120-095-721 Streptomycin Quality Biologicals Cat#120-095-721 Methylcellulose Sigma-Aldrich Cat#435244-250G Critical commercial assays NAb Protein A/G spin columns Thermo Fisher Cat#89958 4-20% Mini-PROTEAN TGX protein gels BioRad Cat#4561096 Immobilon-P transfer membranes Millipore Cat#IPVH00010 Bicinchoninic acid assay Thermo Fisher Cat#23227 Deposited data Raw immunoblot images, raw FRET kinetics data Mendeley Data https://doi.org/10.17632/28cft6pz76.2 Experimental models: Organisms/strains VeroE6-TMPRSS2 Cell repository of the National Institute of Infectious Diseases, Japan RRID: CVCL_YQ49 Software and algorithms GraphPad Prism Graphpad Software https://www.graphpad.com/scientificsoftware/prism/ ImageJ Schneider et al.71 https://imagej.nih.gov/ij/ Adobe Illustrator Adobe https://www.adobe.com/ Excel Microsoft https://www.microsoft.com/en-us/microsoft-365/excel BioRender BioRender https://biorender.com/ R 4.2.1 R Core Team https://www.r-project.org/ Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Arturo Casadevall (acascade1@jhu.edu). Materials availability This study did not generate any new unique reagents. Experimental model and study participant details Human subjects The main CCP samples (CCP1-4) were thawed plasma samples from volunteer apheresis convalescent plasma donors obtained in June 2020 from the Transfusion Medicine Division, Department of Pathology at the Johns Hopkins Hospital. The first library of CCP samples was selected at random from a set of 300 recent convalescent plasma donors collected for two clinical trials27 , 69 (see Table S5 for detailed description of each CCP sample). The first donor cohort was composed of 22 (49%) males and 24 (51%) females with a mean age of 46 years. The second library, also collected for clinical trials, consists of 126 highly characterized CCP samples and the principal cohort of the study has been previously described.40 , 70 Briefly, for both the first and second libraries, individuals were previously diagnosed with SARS-CoV-2 infection by PCR+ nasal swab who met the standard eligibility criteria for blood donation and were collected in the Baltimore, MD, and Washington DC area (Johns Hopkins Medical Institutions, JHMI cohort). The second cohort was composed of 68 (54%) males and 58 (46%) females. 12 (10%) cases were severe enough to require hospitalization (median duration of stay 5 days; interquartile range 2–7 days). Plasma was collected from each donor approximately one month after symptom onset or the first positive PCR test in the case of mild or asymptomatic disease (see Table S5 for detailed description of each CCP sample). Human subject research was approved by both the Johns Hopkins University School of Medicine’s Institutional Review Board. Nonimmune plasma samples were randomly selected from leftover control plasma from the early treatment randomized control trial collected in November 2020 and confirmed seronegative.27 All participants provided informed written consent. Viral strains SARS-CoV-2 clade A.3 isolate (SARS-CoV-2/USA/MD-HP00076/2020 GISAID accession number EPI_ISL_438234) were used for PRNT or deactivated for use in in vitro degradation immunoblot assays. Cell lines Female VeroE6-TMPRSS2 were cultured at 37°C with 5% carbon dioxide in a humidified chamber, using complete medium consisting of minimum essential medium (MEM) supplemented with 10% fetal bovine serum, 1 mmol/L glutamine, 1 mmol/L sodium pyruvate, 100 μg/mL penicillin, and 100 μg/mL streptomycin. For each experiment, cells were plated in 6-well dishes and grown to 75% confluence. Method details Purification of antibodies from CCP Polyclonal antibodies from CCP were purified by NAb Protein A/G spin columns (Thermo Fisher) according to the manufacturer’s protocols. Purified mAbs were assessed for purity by separation with SDS-PAGE and protein staining with SimplyBlue Coomassie stain (Thermo Fisher). The total mAb concentration was then determined by Beer’s Law using absorbance at 280 nm and approximate molar absorptivity coefficients for human IgG. Purified polyclonal antibody mixtures were then heat-treated at 55°C for 30 min and centrifuged at 5000 x g to denature any selectively residual non-antibody biomolecules. FRET assay Quenched fluorescent peptides appended with 7-methoxycoumarin-4-acetic acid (Mca) and 2,4-dinitrophenyl (Dnp) at the N-terminus and C-terminal lysine sidechain, respectively, were synthesized by Peptide 2.0. Kinetic assays with the RBD FRET peptides were performed by diluting each antibody with PBS in 40 μL in an opaque microtiter assay plate (Corning 3993). To each well, 10,000–156 μg/mL of purified CCP antibody, 1- to 64-fold dilution of serum was mixed with 75 μM of RBD FRET peptide was added. Fluorescence was measured with excitation and emission wavelengths of 320 and 405 nm, respectively, and an emission cutoff filter at 320 nm, using the SpectraMax M5 plate reader (Molecular Devices) at a constant temperature of 37°C. Wells were set up in duplicate and fluorescence was measured every minute for 5 h, with a 3 s mixing step between reads. The maximal RFU value measured for the positive control proteinase K (Invitrogen; 2 mg/mL) digestion of RBM FRET peptides (PKmax) was considered complete peptide cleavage (75 μM) and the ratio of maximal RFU for each condition to PKmax was used to calculate total peptide cleavage for each antibody at 200 min, which was then used to calculate initial velocity. For the protease inhibitor screen, the assay set up was identical except for the supplementation of AEBSF (Thermo Scientific), bestatin (Thermo Scientific), E−64 (Thermo Scientific), and pepstatin-A (Thermo Scientific) at constant working concentrations of 1 mM, 40 μM, 10 μM, and 1.5 μM, respectively. For inhibitors dissolved in organic solvents (E−64, 50% EtOH; bestatin and pepstatin-A, 100% MeOH), an equivalent amount of the carrier solvent was supplemented to proteolysis reactions without inhibitor as control. RBD-specific antibody depletion from CCP SARS-CoV-2 Spike RBD-coupled magnetic beads (Acro Biosystems) were incubated with heat-treated CCP3 plasma overnight at 4°C with gentle rotation. The magnetic beads were then separated from the CCP3 using a magnetic separator, and the supernatant (containing RBD-specific Ab-depleted plasma) was removed. The efficiency of RBD-specific antibody removal was assessed by ELISA. Degradation assay of recombinant S1 or RBD recombinant protein Recombinant proteins were obtained from Sino Biologicals and reconstituted in sterile water at 0.25 mg/mL. RBD constructs contained SARS-CoV-2 amino acid sequence from R319-F541 and a C-terminal polyhistidine tags. The spike S1 construct contained SARS-CoV-2 S protein amino acid sequence Val16-Arg685 and a C-terminal polyhistidine tag. Recombinant S1 subunit or RBD was dissolved in 1XDPBS at a concentration of 50 or 10 μg/mL, respectively. Antibody was added to the reactions at concentrations ranging from 125 to 2000 μg/mL and the reactions were incubated at 37°C with rotation for a period of up to 7 days. The reactions were then separated under reducing conditions using 4–20% Mini-PROTEAN TGX protein gels (BioRad) and transferred to Immobilon-P transfer membranes (Millipore) in transfer buffer (20% methanol, 48 mM Tris, 39 mM Glycine, 0.00375% SDS (Sodium Dodecyl Sulfate)) for 10 min at 1.3 A and 25V using a Trans-Blot Turbo Transfer System (BioRad). The membranes were then blocked with 5% BSA (Bovine Serum Albumin) + 0.1% TWEEN 20 for 1 h and blotted with primary antibody overnight at 4°C in blocking solution. The membranes were then washed 5x with TBST (Tris, NaCl, 0.1% TWEEN 20) and probed with a secondary antibody for 1 h at 37°C with gentle shaking. Polyclonal Rabbit anti-RBD antibody (Sino Biological) was employed as the primary antibody, and Europium-labeled Goat α-Mouse IgG (Molecular Devices) or HorseRadish Peroxidase (HRP)- conjugated Goat anti-Rabbit Ig (Southern Biotech) was used as the secondary antibody. Membranes were then washed 5x with TBST. For HRP detection, membranes were developed with SuperSignal West Pico PLUS Chemiluminescent Substrate (Fisher Scientific), and chemiluminescent signal was detected with a ChemiDoc Imaging System (BioRad). For Europium detection, membranes were imaged using the time-resolved fluorescence ScanLater Western Blot protocol on the SpectraMax iD5 (Molecular Devices). Indirect enzyme-linked immunosorbent assays 96-well high-binding polystyrene plates (Corning, 9018) were coated with 2 μg/mL rRBD (Sino Biologicals, 40592-V08H) overnight at 4°C, with gentle shaking. The plates were then blocked with blocking solution (TBST +1% BSA) for 1 h at 37°C and then washed 5x with TBST. For antibody standards, monoclonal human anti-RBD mAb (Sino Biological) was added to wells at concentrations ranging from 50 to 0.8 μg/mL. Purified antibodies from CCP samples were added to plates at total antibody concentrations ranging from 500 to 7.8 μg/mL. Plates were incubated at 37°C for 1 h with gentle shaking and washed 5x with TBST. For detection, 50 μL Goat α-Human IgG H + L-AP (Southern Biotech) was added at a concentration of 1 μg/mL and the plates were again incubated at 37°C for 1 h and washed 5x with TBST. 50 μL 1 mg/mL of para-nitrophenylphosphate (PNPP; Sigma) substrate was added to determine the concentration of antibody in each well and the absorbance at 405 nm was measured in each well with the EMax Plus Microplate reader (Molecular Devices). Absorbances from the CCP antibody were interpolated into the standard curve to determine the percentage of RBD-specific antibody in each CCP sample. Virus inactivation procedure Vero TMPRSS2 cells in flasks with 90–100% confluence were infected with the specific SARS-CoV-2 isolate at a multiplicity of infection (MOI) of 0.01, incubated for 3days at 33°C, upon which virus supernatant was collected, centrifuged, and inactivated with the addition of 0.05% β-propiolactone and incubation at 4°C for 24 h followed by a 2 h 37°C incubation. The inactivated virus particles were purified by ultracentrifugation in a Beckman SW28TU rotor at 25,000 rpm for 1 h at 4°C with a 20% sucrose gradient and the virus pellet was resuspended in 1X PBS+. To confirm the virus had been successfully inactivated, a TCID50 assay was performed.72 A bicinchoninic acid assay (BCA; Pierce) was used to estimate the relative viral protein concentration. Plaque reduction neutralization test (PRNT) assay The SARS-CoV-2 PRNT has been described previously.73 Briefly, VeroE6-TMPRSS2 cells were seeded in a sterile 6-well plate and incubated at 37°C until 100% confluent. Heat-treated CCP3 convalescent plasma or purified IgG samples (stock = 17.1 mg/mL total IgG, 4.78 μg/mL α-RBD IgG) were diluted to 1:20 using DMEM containing 2.5% FBS (Gibco, Thermo Fisher Scientific), 1 mM glutamine (Invitrogen, Thermo Fisher Scientific), 1 mM sodium pyruvate (Invitrogen, Thermo Fisher Scientific), 100 U/mL penicillin (Invitrogen, Thermo Fisher Scientific), and 100 μg/mL streptomycin (Invitrogen, Thermo Fisher Scientific)40 and serially diluted in 2-folds until 1:2560. Each dilution and a non-plasma control were incubated with 100 plaque-forming units (PFU) of a SARS-CoV-2 clade A.3 isolate. For each experiment, three sets of diluted samples were prepared. The first set of virus-plasma mixtures were incubated for 1 h at RT, the second set of samples were incubated for 6 h at 37°C, and the third set was incubated for 6 h at 4°C. After the incubation 250 μL of each mixture was overlaid on VeroE6-TMPRSS2 monolayers in duplicate wells for 1 h at 37°C. Then the virus-plasma mixture was carefully aspirated and 2 mL of sterile 1% methylcellulose (a mixture of equal parts 2% methylcellulose (Sigma) and equal parts of 2X MEM (Gibco) supplemented with 1% Penicillin and Streptomycin, 1% GlutaMAX and 10% FBS) were overlaid into each well and incubated for 48 h at 37°C. The plates were then fixed with 4% formaldehyde overnight and stained with Napthol Blue Black. IC50 was calculated based on PFU counts using the inhibition dose-response, non-linear regression model using the GraphPad Prism 9 software. Quantification and statistical analysis Statistical analysis Data were analyzed in Excel (Microsoft, Redmond, WA) and graphed in Prism 9.0 (Graphpad, San Diego, CA). For the two CCP cohorts, Spearman correlations of epitope cleavage observed by FRET for RBM1 or RBM2 and donor age or days since PCR positive diagnosis of COVID-19, anti-Spike antibody or nAb AUC were conducted. Missing data were excluded (available case method). Analysis was performed in R 4.2.1 (R Core Team, Vienna, Austria). Statistical significance for PRNT data was measured using the Mann Whitney matched rank test, using GraphPad Prism. Two-tailed p value <0.05 was considered statistically significant. Supplemental information Document S1. Figures S1–S7, Tables S1–S4, S6 and Note S1 Table S5. Plasma characterization of individual donors in CCP library 1 and 2, related to Figures 1 and S7 Document S2. Article plus supplemental information Data and code availability • Raw FRET kinetics data for de-indentified human plasma samples and original western blot images have been deposited at Mendeley and are publicly available as of the date of publication. Additional Supplemental Items available from Mendeley Data at https://doi.org/10.17632/28cft6pz76.2. • This paper does not report original code. • All data reported in this paper and any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledgments We would like to thank Dr. Alan Scott for critical review of this manuscript. 10.13039/100000002 National Institutes of Health grants AI052733, AI152078, and HL059842 (A.C.). 10.13039/100000002 National Institutes of Health grants T32AI007417 and R01AI162381 (S.A.M.). 10.13039/100000060 National Institute of Allergy and Infectious Diseases contract N272201400007C (A.P. and J.S.). U.S. Department of Defense’s (DOD) Joint Program Executive Office for Chemical, Biological, Radiological and Nuclear Defense (10.13039/100017443 JPEO-CBRND ), in collaboration with the Defense Health Agency (10.13039/100009898 DHA ) contract number: W911QY2090012, with additional support from Bloomberg Philanthropies and State of Maryland (D.J.S., E.M.B., and A.A.R.T.). 10.13039/100000050 National Heart, Lung, and Blood Institute grant 1K23HL15182 (E.M.B.). Author contributions Conceptualization, S.A.M, M.P.W., and A.C.; Methodology, S.A.M.; Investigation, S.A.M., J.S., N.J.M., and P.A.F.; Writing – Original Draft, S.A.M.; Writing – Review and Editing, R.J.B.C., M.P.W, A.A.R.T., E.M.B., D.S., A.P., and A.C.; Resources, A.A.R.T, D.S., A.P, J.R.S., H-S.P., and S.L.K.; Supervision, A.A.R.T, A.P., and A.C. Declaration of interests None declared. Supplemental information can be found online at https://doi.org/10.1016/j.chembiol.2023.05.011. ==== Refs References 1 Hu B. Guo H. Zhou P. Shi Z.L. Characteristics of SARS-CoV-2 and COVID-19 Nat. Rev. Microbiol. 19 2020 141 154 10.1038/s41579-020-00459-7 33024307 2 Zhou P. Yang X. Lou Wang X.G. Hu B. Zhang L. Zhang W. Si H.R. Zhu Y. Li B. Huang C.L. A pneumonia outbreak associated with a new coronavirus of probable bat origin Nature 579 2020 270 273 10.1038/s41586-020-2012-7 32015507 3 Wrapp D. Wang N. Corbett K.S. Goldsmith J.A. Hsieh C.-L. Abiona O. Graham B.S. McLellan J.S. Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation Science 367 2020 1260 1263 10.1126/science.abb2507 32075877 4 Pallesen J. Wang N. Corbett K.S. Wrapp D. Kirchdoerfer R.N. Turner H.L. Cottrell C.A. Becker M.M. Wang L. Shi W. Immunogenicity and structures of a rationally designed prefusion MERS-CoV spike antigen Proc. Natl. Acad. Sci. USA 114 2017 E7348 E7357 10.1073/PNAS.1707304114 28807998 5 Gui M. Song W. Zhou H. Xu J. Chen S. Xiang Y. Wang X. Cryo-electron microscopy structures of the SARS-CoV spike glycoprotein reveal a prerequisite conformational state for receptor binding Cell Res. 27 2017 119 129 10.1038/cr.2016.152 28008928 6 Walls A.C. Xiong X. Park Y.J. Tortorici M.A. Snijder J. Quispe J. Cameroni E. Gopal R. Dai M. Lanzavecchia A. Unexpected receptor functional mimicry elucidates activation of coronavirus fusion Cell 176 2019 1026 1039.e15 10.1016/j.cell.2018.12.028 30712865 7 Yan R. Zhang Y. Li Y. Xia L. Guo Y. Zhou Q. Structural basis for the recognition of SARS-CoV-2 by full-length human ACE2 Science 367 2020 1444 1448 10.1126/SCIENCE.ABB2762 32132184 8 Xia S. Zhu Y. Liu M. Lan Q. Xu W. Wu Y. Ying T. Liu S. Shi Z. Jiang S. Fusion mechanism of 2019-nCoV and fusion inhibitors targeting HR1 domain in spike protein Cell. Mol. Immunol. 17 2020 765 767 10.1038/s41423-020-0374-2 32047258 9 Shang J. Wan Y. Luo C. Ye G. Geng Q. Auerbach A. Li F. Cell entry mechanisms of SARS-CoV-2 Proc. Natl. Acad. Sci. USA 117 2020 11727 11734 10.1073/pnas.2003138117 32376634 10 Rossi G.A. Sacco O. Mancino E. Cristiani L. Midulla F. Differences and similarities between SARS-CoV and SARS-CoV-2: spike receptor-binding domain recognition and host cell infection with support of cellular serine proteases Infection 48 2020 665 669 10.1007/S15010-020-01486-5 32737833 11 Grant O.C. Montgomery D. Ito K. Woods R.J. Analysis of the SARS-CoV-2 spike protein glycan shield reveals implications for immune recognition Sci. Rep. 10 2020 14991 10.1038/s41598-020-71748-7 32929138 12 Baden L.R. el Sahly H.M. Essink B. Kotloff K. Frey S. Novak R. Diemert D. Spector S.A. Rouphael N. Creech C.B. Efficacy and safety of the mRNA-1273 SARS-CoV-2 vaccine N. Engl. J. Med. Overseas. Ed. 384 2021 403 416 10.1056/nejmoa2035389 13 Polack F.P. Thomas S.J. Kitchin N. Absalon J. Gurtman A. Lockhart S. Perez J.L. Pérez Marc G. Moreira E.D. Zerbini C. Safety and efficacy of the BNT162b2 mRNA covid-19 vaccine N. Engl. J. Med. 383 2020 2603 2615 10.1056/NEJMoa2034577 33301246 14 Gottlieb R.L. Nirula A. Chen P. Boscia J. Heller B. Morris J. Huhn G. Cardona J. Mocherla B. Stosor V. Effect of bamlanivimab as monotherapy or in combination with etesevimab on viral load in patients with mild to moderate COVID-19: a randomized clinical trial JAMA 325 2021 632 644 10.1001/JAMA.2021.0202 33475701 15 Weinreich D.M. Sivapalasingam S. Norton T. Ali S. Gao H. Bhore R. Musser B.J. Soo Y. Rofail D. Im J. REGN-COV2, a neutralizing antibody cocktail, in outpatients with covid-19 N. Engl. J. Med. 384 2021 238 251 10.1056/NEJMOA2035002 33332778 16 Gupta A. Gonzalez-Rojas Y. Juarez E. Crespo Casal M. Moya J. Falci D.R. Sarkis E. Solis J. Zheng H. Scott N. Early treatment for covid-19 with SARS-CoV-2 neutralizing antibody sotrovimab N. Engl. J. Med. 385 2021 1941 1950 10.1056/NEJMOA2107934 34706189 17 Focosi D. Maggi F. Franchini M. McConnell S. Casadevall A. Analysis of immune escape variants from antibody-based therapeutics against COVID-19: a systematic review Int. J. Mol. Sci. 23 2021 29 10.3390/IJMS23010029 35008446 18 Harvey W.T. Carabelli A.M. Jackson B. Gupta R.K. Thomson E.C. Harrison E.M. Ludden C. Reeve R. Rambaut A. COVID-19 Genomics UK COG-UK Consortium SARS-CoV-2 variants, spike mutations and immune escape Nat. Rev. Microbiol. 19 2021 409 424 10.1038/s41579-021-00573-0 34075212 19 Casadevall A. Pirofski L.A. Joyner M.J. The principles of antibody therapy for infectious diseases with relevance for COVID-19 mBio 12 2021 03372-20 10.1128/MBIO.03372-20 20 Casadevall A. Henderson J.P. Joyner M.J. Pirofski L.A. SARS-CoV-2 variants and convalescent plasma: reality, fallacies, and opportunities J. Clin. Invest. 131 2021 e148832 10.1172/JCI148832 33621214 21 Tobian A.A.R. Cohn C.S. Shaz B.H. COVID-19 Convalescent plasma Blood 140 2022 196 207 10.1182/BLOOD.2021012248 34695186 22 Bloch E.M. Shoham S. Casadevall A. Sachais B.S. Shaz B. Winters J.L. Van Buskirk C. Grossman B.J. Joyner M. Henderson J.P. Deployment of convalescent plasma for the prevention and treatment of COVID-19 J. Clin. Invest. 130 2020 2757 2765 10.1172/JCI138745 32254064 23 Bar K.J. Shaw P.A. Choi G.H. Aqui N. Fesnak A. Yang J.B. Soto-Calderon H. Grajales L. Starr J. Andronov M. A randomized controlled study of convalescent plasma for individuals hospitalized with COVID-19 pneumonia J. Clin. Invest. 131 2021 e155114 10.1172/JCI155114 34788233 24 Estcourt L.J. Turgeon A.F. McQuilten Z.K. McVerry B.J. Al-Beidh F. Annane D. Arabi Y.M. Arnold D.M. Beane A. Bégin P. Effect of convalescent plasma on organ support-free days in critically ill patients with COVID-19: a randomized clinical trial JAMA 326 2021 1690 1702 10.1001/JAMA.2021.18178 34606578 25 Libster R. Pérez Marc G. Wappner D. Coviello S. Bianchi A. Braem V. Esteban I. Caballero M.T. Wood C. Berrueta M. Early high-titer plasma therapy to prevent severe covid-19 in older adults N. Engl. J. Med. 384 2021 610 618 10.1056/NEJMoa2033700 33406353 26 Estcourt L.J. Cohn C.S. Pagano M.B. Iannizzi C. Kreuzberger N. Skoetz N. Allen E.S. Bloch E.M. Beaudoin G. Casadevall A. Clinical practice guidelines from the association for the advancement of blood and biotherapies (AABB): COVID-19 convalescent plasma Ann. Intern. Med. 175 2022 1310 1321 35969859 27 Sullivan D.J. Gebo K.A. Shoham S. Bloch E.M. Lau B. Shenoy A.G. Mosnaim G.S. Gniadek T.J. Fukuta Y. Patel B. Early outpatient treatment for covid-19 with convalescent plasma N. Engl. J. Med. 386 2022 1700 1711 10.1056/NEJMOA2119657 35353960 28 Bonny T.S. Patel E.U. Zhu X. Bloch E.M. Grabowski M.K. Abraham A.G. Littlefield K. Shrestha R. Benner S.E. Laeyendecker O. Cytokine and chemokine levels in coronavirus disease 2019 convalescent plasma Open Forum Infect. Dis. 8 2021 ofaa574 10.1093/OFID/OFAA574 33553467 29 Natarajan H. Crowley A.R. Butler S.E. Xu S. Weiner J.A. Bloch E.M. Littlefield K. Wieland-Alter W. Connor R.I. Wright P.F. Markers of polyfunctional SARS-CoV-2 antibodies in convalescent plasma mBio 12 2021 00765-21 10.1128/mBio.00765-21 30 Crawford C.J. Wear M.P. Smith D.F.Q. d’Errico C. McConnell S.A. Casadevall A. Oscarson S. A glycan FRET assay for detection and characterization of catalytic antibodies to the Cryptococcus neoformans capsule Proc. Natl. Acad. Sci. USA 118 2021 10.1073/PNAS.2016198118 e2016198118 31 Bowen A. Wear M. Casadevall A. Antibody-mediated catalysis in infection and immunity Infect. Immun. 85 2017 e00202-17 10.1128/IAI.00202-17 28674034 32 Bowen A. Wear M.P. Cordero R.J.B. Oscarson S. Casadevall A. A monoclonal antibody to Cryptococcus neoformans glucuronoxylomannan manifests hydrolytic activity for both peptides and polysaccharides J. Biol. Chem. 292 2017 417 434 10.1074/JBC.M116.767582 27872188 33 Paul S. Volle D.J. Beach C.M. Johnson D.R. Powell M.J. Massey R.J. Catalytic hydrolysis of vasoactive intestinal peptide by human autoantibody Science 244 1989 1158 1162 10.1126/SCIENCE.2727702 2727702 34 Ju B. Zhang Q. Ge J. Wang R. Sun J. Ge X. Yu J. Shan S. Zhou B. Song S. Human neutralizing antibodies elicited by SARS-CoV-2 infection Nature 584 2020 115 119 10.1038/s41586-020-2380-z 32454513 35 Chakraborti S. Prabakaran P. Xiao X. Dimitrov D.S. The SARS coronavirus S glycoprotein receptor binding domain: fine mapping and functional characterization Virol. J. 2 2005 73 10.1186/1743-422X-2-73 16122388 36 Lan J. Ge J. Yu J. Shan S. Zhou H. Fan S. Zhang Q. Shi X. Wang Q. Zhang L. Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor Nature 581 2020 215 220 10.1038/s41586-020-2180-5 32225176 37 Yuan M. Liu H. Wu N.C. Wilson I.A. Recognition of the SARS-CoV-2 receptor binding domain by neutralizing antibodies Biochem. Biophys. Res. Commun. 538 2021 192 203 10.1016/J.BBRC.2020.10.012 33069360 38 Finkelstein M.T. Mermelstein A.G. Miller E.P. Seth P.C. Stancofski E.S.D. Fera D. Structural analysis of neutralizing epitopes of the SARS-CoV-2 spike to guide therapy and vaccine design strategies Viruses 13 2021 134 10.3390/V13010134 33477902 39 Deshpande A. Harris B.D. Martinez-Sobrido L. Kobie J.J. Walter M.R. Epitope classification and RBD binding properties of neutralizing antibodies against SARS-CoV-2 variants of concern Front. Immunol. 12 2021 2185 10.3389/FIMMU.2021.691715/BIBTEX 40 Klein S.L. Pekosz A. Park H.-S. Ursin R.L. Shapiro J.R. Benner S.E. Littlefield K. Kumar S. Naik H.M. Betenbaugh M.J. Sex, age, and hospitalization drive antibody responses in a COVID-19 convalescent plasma donor population J. Clin. Invest. 130 2020 6141 6150 10.1172/JCI142004 32764200 41 Neisser M. Wechsberg F. Ueber die Wirkungsart bactericider sera Munchner Medizinsche Wochenshcrift 48 1901 697 42 Bayne-Jones S. Equilibria in precipitin reactions J. Exp. Med. 25 1917 837 853 10.1084/jem.25.6.837 19868126 43 Weinstock C. Schnaidt M. The complement-mediated prozone effect in the Luminex single-antigen bead assay and its impact on HLA antibody determination in patient sera Int. J. Immunogenet. 40 2013 171 177 10.1111/J.1744-313X.2012.01147.X 22913810 44 Akazawa-Ogawa Y. Nagai H. Hagihara Y. Heat denaturation of the antibody, a multi-domain protein Biophys. Rev. 10 2018 255 258 10.1007/S12551-017-0361-8 29256117 45 Triglia R.P. Linscott W.D. Titers of nine complement components, conglutinin and C3b-inactivator in adult and fetal bovine sera Mol. Immunol. 17 1980 741 748 10.1016/0161-5890(80)90144-3 7432351 46 Hu X. An T. Situ B. Hu Y. Ou Z. Li Q. He X. Zhang Y. Tian P. Sun D. Heat inactivation of serum interferes with the immunoanalysis of antibodies to SARS-CoV-2 J. Clin. Lab. Anal. 34 2020 e23411 10.1002/JCLA.23411 32594577 47 Shrock E. Fujimura E. Kula T. Timms R.T. Lee I.-H. Leng Y. Robinson M.L. Sie B.M. Li M.Z. Chen Y. Viral epitope profiling of COVID-19 patients reveals cross-reactivity and correlates of severity Science 370 2020 eabd4250 10.1126/science.abd4250 32994364 48 Yaqinuddin A. Cross-immunity between respiratory coronaviruses may limit COVID-19 fatalities Med. Hypotheses 144 2020 110049 10.1016/J.MEHY.2020.110049 32758887 49 Planque S. Mitsuda Y. Taguchi H. Salas M. Morris M.K. Nishiyama Y. Kyle R. Okhuysen P. Escobar M. Hunter R. Characterization of gp120 hydrolysis by IgA antibodies from humans without HIV infection AIDS Res. Hum. Retroviruses 23 2007 1541 1554 10.1089/AID.2007.0081 18160012 50 Polosukhina D.I. Buneva V.N. Doronin B.M. Tyshkevich O.B. Boiko A.N. Gusev E.I. Favorova O.O. Nevinsky G.A. Hydrolysis of myelin basic protein by IgM and IgA antibodies from the sera of patients with multiple sclerosis Med. Sci. Monit. 11 2005 BR266 BR272 16049372 51 Brown E.L. Nishiyama Y. Dunkle J.W. Aggarwal S. Planque S. Watanabe K. Csencsits-Smith K. Bowden M.G. Kaplan S.L. Paul S. Constitutive production of catalytic antibodies to a Staphylococcus aureus virulence factor and effect of infection J. Biol. Chem. 287 2012 9940 9951 10.1074/jbc.M111.330043 22303018 52 Perrin P. Morgeaux S. Inactivation of DNA by beta-propiolactone Biologicals 23 1995 207 211 10.1006/BIOL.1995.0034 8527119 53 Uittenbogaard J.P. Zomer B. Hoogerhout P. Metz B. Reactions of beta-propiolactone with nucleobase analogues, nucleosides, and peptides: implications for the inactivation of viruses J. Biol. Chem. 286 2011 36198 36214 10.1074/JBC.M111.279232 21868382 54 Zhou F. Yu T. Du R. Fan G. Liu Y. Liu Z. Xiang J. Wang Y. Song B. Gu X. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study Lancet 395 2020 1054 1062 10.1016/S0140-6736(20)30566-3 32171076 55 Kim L. Garg S. O’Halloran A. Whitaker M. Pham H. Anderson E.J. Armistead I. Bennett N.M. Billing L. Como-Sabetti K. Risk factors for intensive care unit admission and in-hospital mortality among hospitalized adults identified through the US coronavirus disease 2019 (COVID-19)-Associated hospitalization surveillance network (COVID-NET) Clin. Infect. Dis. 72 2021 E206 E214 10.1093/CID/CIAA1012 32674114 56 McLain L. Dimmock N.J. Single- and multi-hit kinetics of immunoglobulin G neutralization of human immunodeficiency virus type 1 by monoclonal antibodies J. Gen. Virol. 75 Pt 6 1994 1457 1460 10.1099/0022-1317-75-6-1457/CITE/REFWORKS 8207410 57 Magnus C. Virus neutralisation: new insights from kinetic neutralisation curves PLoS Comput. Biol. 9 2013 e1002900 10.1371/JOURNAL.PCBI.1002900 23468602 58 Stalls V. Lindenberger J. Gobeil S.M.-C. Henderson R. Parks R. Barr M. Deyton M. Martin M. Janowska K. Huang X. Cryo-EM structures of SARS-CoV-2 Omicron BA.2 spike Cell Rep. 39 2022 111009 10.1016/J.CELREP.2022.111009 35732171 59 Javanmardi K. Segall-Shapiro T.H. Chou C.-W. Boutz D.R. Olsen R.J. Xie X. Xia H. Shi P.-Y. Johnson C.D. Annapareddy A. Antibody escape and cryptic cross-domain stabilization in the SARS-CoV-2 Omicron spike protein Cell Host Microbe 30 2022 1242 1254.e6 10.1016/J.CHOM.2022.07.016 35988543 60 Li M. Beck E.J. Laeyendecker O. Eby Y. Tobian A.A.R. Caturegli P. Wouters C. Chiklis G.R. Block W. McKie R.O. Convalescent plasma with a high level of virus-specific antibody effectively neutralizes SARS-CoV-2 variants of concern Blood Adv. 6 2022 3678 3683 10.1182/BLOODADVANCES.2022007410 35443020 61 Delaet I. Boeyé A. Capsid destabilization is required for antibody-mediated disruption of poliovirus J. Gen. Virol. 75 1994 581 587 10.1099/0022-1317-75-3-581 8126454 62 Brioen P. Rombaut B. Boeyé A. Hit-and-run neutralization of poliovirus J. Gen. Virol. 66 1985 2495 2499 10.1099/0022-1317-66-11-2495 2997379 63 Pierson T.C. Diamond M.S. A game of numbers: the stoichiometry of antibody-mediated neutralization of flavivirus infection Prog. Mol. Biol. Transl. Sci. 129 2015 141 166 10.1016/BS.PMBTS.2014.10.005 25595803 64 Taylor H.P. Armstrong S.J. Dimmock N.J. Quantitative relationships between an influenza virus and neutralizing antibody Virology 159 1987 288 298 10.1016/0042-6822(87)90466-1 3617501 65 Lee J.H. Leaman D.P. Kim A.S. Torrents de la Peña A. Sliepen K. Yasmeen A. Derking R. Ramos A. De Taeye S.W. Ozorowski G. Antibodies to a conformational epitope on gp41 neutralize HIV-1 by destabilizing the Env spike Nat. Commun. 6 2015 8167 10.1038/NCOMMS9167 26404402 66 Paul S. Karle S. Planque S. Taguchi H. Salas M. Nishiyama Y. Handy B. Hunter R. Edmundson A. Hanson C. Naturally occurring proteolytic antibodies: selective immunoglobulin M-catalyzed hydrolysis of HIV gp120 J. Biol. Chem. 279 2004 39611 39619 10.1074/JBC.M406719200 15269209 67 Klasse P.J. Neutralization of virus infectivity by antibodies: old problems in new perspectives Adv. Biol. 2014 2014 157895 10.1155/2014/157895 27099867 68 Corti D. Lanzavecchia A. Broadly neutralizing antiviral antibodies Annu. Rev. Immunol. 31 2013 705 742 10.1146/annurev-immunol-032712-095916 23330954 69 Shoham S. Bloch E.M. Casadevall A. Hanley D. Lau B. Gebo K. Cachay E. Kassaye S.G. Paxton J.H. Gerber J. Transfusing convalescent plasma as post-exposure prophylaxis against SARS-CoV-2 infection: a double-blinded, phase 2 randomized, controlled trial Clin. Infect. Dis. 76 2023 e477 e486 10.1093/CID/CIAC372 35579509 70 Patel E.U. Bloch E.M. Clarke W. Hsieh Y.H. Boon D. Eby Y. Fernandez R.E. Baker O.R. Keruly M. Kirby C.S. Comparative performance of five commercially available serologic assays to detect antibodies to SARS-CoV-2 and identify individuals with high neutralizing titers J. Clin. Microbiol. 59 2021 02257-20 10.1128/JCM.02257-20 71 Schneider C.A. Rasband W.S. Eliceiri K.W. NIH Image to ImageJ: 25 years of image analysis Nat. Methods 9 2012 671 675 10.1038/NMETH.2089 22930834 72 Dhakal S. Ruiz-Bedoya C.A. Zhou R. Creisher P.S. Villano J.S. Littlefield K. Ruelas Castillo J. Marinho P. Jedlicka A.E. Ordonez A.A. Sex differences in Lung imaging and SARS-CoV-2 antibody responses in a COVID-19 golden Syrian hamster model mBio 12 2021 e0097421 10.1128/MBIO.00974-21 34253053 73 Gordon O. Brosnan M.K. Yoon S. Jung D. Littlefield K. Ganesan A. Caputo C.A. Li M. Morgenlander W.R. Henson S.N. Pharmacokinetics of high-titer anti-SARS-CoV-2 human convalescent plasma in high-risk children JCI Insight 7 2022 e151518 10.1172/JCI.INSIGHT.151518 34855624
PMC010xxxxxx/PMC10288822.txt	==== Front Healthc Anal (N Y) Healthc Anal (N Y) Healthcare Analytics (New York, N.y.) 2772-4425 The Author(s). Published by Elsevier Inc. S2772-4425(23)00076-X 10.1016/j.health.2023.100209 100209 Article A fractional mathematical model with nonlinear partial differential equations for transmission dynamics of severe acute respiratory syndrome coronavirus 2 infection Thabet Hayman ab⁎ Kendre Subhash c a Division of Applied Mathematics, Brown University, Providence, RI 02906, USA b Department of Mathematics, University of Aden, Aden, Yemen c Department of Mathematics, Savitribai Phule Pune University, Pune, MH, India ⁎ Corresponding author. 12 6 2023 12 2023 12 6 2023 4 100209100209 29 3 2023 23 4 2023 4 6 2023 © 2023 The Author(s) 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. This study presents a fractional mathematical model based on nonlinear Partial Differential Equations (PDEs) of fractional variable-order derivatives for the host populations experiencing transmission and evolution of the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) pandemic. Five host population groups have been considered, the Susceptible, Exposed, Infected, Recovered, and Deceased (SEIRD). The new model, not introduced before in its current formulation, is governed by nonlinear PDEs with fractional variable-order derivatives. As a result, the proposed model is not compared with other models or real scenarios. The advantage of the proposed fractional partial derivatives of variable orders is that they can model the rate of change of subpopulation for the proposed model. As an efficient tool to obtain the solution of the proposed model, a modified analytical technique based on the homotopy and Adomian decomposition methods is introduced. Then again, the present study is general and is applicable to a host population in any country. Keywords Bioanalytics Mathematical biology Fractional mathematical modeling Nonlinear partial differential equations Fractional order system ==== Body pmc1 Introduction Most of the models that come from exceptionally diverse regions of investigation are all diffusion processes, however, models with non-local diffusion, where the movement of a particle is affected by genes around its forthright environment, makes a difference to depict the behavior of matter in “meeting places”, such as where an ice sheet meets the sea or the edge of a timberland fire meets a timberland. During the last decades, it has been noted that modeling nonlinear phenomena with fractional derivatives provides a better fit due to their non-local nature. Differential operators of fractional order have non-local behaviors since fractional derivatives at a point rely upon the characteristics of the entire function and not just the values in the vicinity of the point, which is helpful to enhance the performance of texture preservation. Fractional PDEs which involve fractional partial derivatives, in time or/and space, have been considered in many ways as a novel topic and they have been the subjects of several conferences due to their important applications in natural and applied sciences [1], [2], [3], [4], [5], [6], [7], [8]. Several problems in biology, ecology, ontology, and epidemiology need to be modeled in terms of fractional PDEs through mathematical modeling. Mathematical modeling is implemented to compute and evaluate parameters that are substantial for a dynamic understanding of epidemic transmission. However, scientific modeling plays a critical part in epidemiology. Results of mathematical modeling for epidemiological studies help to attract health interventions for effective disease control. Although epidemiology is a descriptive science, evaluating epidemiological data is not often possible because of the high complexity of epidemic observations and fighting against diseases. Fighting against infectious diseases, however, is like targeting a moving objective. Even though different contamination control methodologies, still make large-scale ceaseless pathogens and create reestablished challenges to irresistible illness control. Infectious diseases such as malaria, tuberculosis, severe acute respiratory syndrome (SARS), and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), have been and maintain a global threat to human health. However, SARS-CoV-2 is a highly transmissible and pathogenic coronavirus that has created a pandemic of drastic respiratory disease, called ‘coronavirus disease 2019’ (COVID-19), which threatens human health and public safety. SARS-CoV-2 is a spread infectious disease that has been a threat to humans for a long time. However, caused by a newly discovered coronavirus. Most individuals tainted with the SARS-CoV-2 infection will involve gentle to direct respiratory sickness and recoup without requiring uncommon treatment. Although many factors for reduced activation rates of SARS-CoV-2 such as the development of antibiotics and vaccines, SARS-CoV-2 is drug-susceptible and resistant. Consequently, some significant factors must be taken into account to limit the evolution of SARS-CoV-2, for example, improved diagnostic procedures and changes in the host potential population. Therefore, modeling the potential host population and developing mathematical techniques can help in the evolution and reduction of SARS-CoV-2 and inform public health interventions. Over the last four years, SARS-CoV-2 has been one of the top major causes of death and emerged as a threat to public health worldwide [9], [10], [11]. Several implementations with the rapid evolution of molecular techniques have been applied for more effective vaccines and dynamical control for SARS-CoV-2. The successful implementation of SARS-CoV-2 control depends on monitoring and epidemiological understanding of the tracks of spreading in a population. However, formulating models for the population dynamics of SARS-CoV-2 can evaluate the adequacy of diverse anticipation and mediation strategies. Some research works on modeling the evolution of infectious diseases are available in the literature [12], [13], [14], [15], [16], [17]. Recently, Chu et al. [18] the transmission mechanism of the SARS-CoV-2 virus using the non-local Atangana-Baleanu fractional-order approach based on ordinary differential equations. Moreover, Bilgil et al. [19] introduced a fractional-order mathematical model based on vaccinated and infected compartments of SARS-CoV-2. Further, DarAssi et al. [20] introduced a vigorous study of fractional order mathematical model for the SARS-CoV-2 epidemic with the Mittag-Leffler kernel. Furthermore, Haq et al. [21] developed a within-host viral kinetics model of SARS-CoV-2 under the Caputo fractional-order operator. In addition, Aba Oud et al. [22] formulated a fractional epidemic model in the Caputo sense with the consideration of quarantine, isolation, and environmental impacts to examine the dynamics of the COVID-19 outbreak. These fractional models are quite useful for understanding better the disease epidemics as well as capturing the memory and nonlocality effects. However, many challenges are still faced in bridging the gap between immunology and epidemiology. Measuring population immunity and illness may be a challenging errand and is right now a dynamic range of inquiry. Consequently, to meet these challenges, mathematical modeling is a compelling procedure for formulating the population elements that can offer assistance for disease control. Moreover, an effective mathematical model is introduced with analytical and numerical solutions. This study may help epidemic control, and provide better understanding and impact measures of the transmission of SARS-CoV-2. 1.1 Contributions of this paper Evaluating population immunity to measure the epidemiology of health and disease is an intriguing issue that is currently being researched. As a result, forecasting a potential host population and developing mathematical and computational methods may assist in the evolution and control of infectious diseases, as well as advise public health initiatives. The following are some of the advantages of this paper: 1 This paper introduces a fractional mathematical model based on nonlinear PDEs of fractional variable-order derivatives for the host population experiencing transmission and evolution of SARS-CoV-2. This model is presented for the first time in its current formulation. The proposed model represents five host population groups that have been considered, the susceptible, exposed, infected, recovered, and deceased. 2 This paper also introduces an analytical technique based on the homotopy and Adomian decomposition methods to approximate a fully general nonlinear system of nonlinear fractional PDEs. This analytical technique is applied to obtain an analytical solution for the SARS-CoV-2 model. By comparing the proposed technique with other techniques and methods available in the literature, see, for example, [23], [24], [25], [26], our technique is more general and very effective in approximating analytical solutions to general nonlinear systems of fractional PDEs. 3 Although most epidemic model solutions are numerical, obtaining analytical solutions for epidemic problems is a tricky task. Analytical solutions are often preferred over numerical solutions because they provide exact answers, whereas numerical solutions may have errors due to approximations. In this paper, an analytical solution to the SEIRD model is obtained. This solution, in its current formulation, is in general and has not been previously introduced. The graphical representations of the solution for the proposed model are also shown. The numerical results are obtained to confirm the effectiveness of the proposed technique in approximating the solution to the SEIRD model. 1.2 Organization of the paper The rest of the paper is organized as follows. In Section 2, we present basic results which are needed in the sequel. In Section 3, we introduce a SARS-CoV-2 infection model representing five host population groups, the susceptible, exposed, infected, recovered, and deceased. Moreover, an analytical technique based on the homotopy and Adomian decomposition methods is introduced to approximate analytical solutions to the proposed model. In Section 4, we obtain analytical solutions to the SEIRD model. Numerical results are obtained in Section 5 to confirm the accuracy and effectiveness of the proposed technique. The conclusion of the paper is introduced in Section 6. 2 Preliminaries In this section, we introduce basic definitions and theorems of fractional integrals and derivatives. Some of these are modified accordingly as needed in this paper and some others can be found in [27], [28], [29], [30], [31] and some references cited therein. Definition 1 Let u(x,t):R×(0,∞)→R and k−1<q(t)<k∈N, then the Riemann–Liouville time fractional partial integral of variable-order q(t) for a function u(x,t) is defined as follows: (2.1) Itq(t)u(x,t)=1Γ(q(t))∫0t(t−τ)q(t)−1u(x,τ)dτ, where Γ is the well-known Gamma function. Definition 2 Let u(x,t):R×(0,∞)→R and k−1<q(t)<k∈N, then the Riemann–Liouville time fractional partial derivative of variable-order q(t) for a function u(x,t) is defined as follows: (2.2) RLDtq(t)u(x,t)=∂k∂tk∫0t(t−τ)k−q(t)−1Γ(k−q(t))u(x,τ)dτ. Definition 3 Let u(x,t):R×(0,∞)→R and k−1<q(t)<k∈N, then the Caputo time fractional partial derivative of variable-order q(t) for a function u(x,t) is defined as follows: (2.3) Dtq(t)u(x,t)=∫0t(t−τ)k−q(t)−1Γ(k−q(t))∂ku(x,τ)∂τkdτ,Dtq(t)u(x,t)=∂ku(x,t)∂tk,q(t)=k∈N. Theorem 2.1 Let u(x,t):R×(0,∞)→R and k−1<q(t)<k∈N . Then (2.4) Itq(t)Dtq(t)u(x,t)=u(x,t)−∑j=0k−1tjj!∂ju(x,0+)∂tj,Dtq(t)Itq(t)u(x,t)=u(x,t). Theorem 2.2 Let q(t),t∈R,t>0 and k−1<q(t)<k∈N . Then (2.5) Dtq(t)tp(t)=Γ(p(t)+1)Γ(p−q(t)+1)tp(t)−q(t),k≤p(t),Dtq(t)tp(t)=0,\|p(t)\|≤k−1,t∈[a,∞),a∈R+. Theorem 2.3 Assume that the function u(x,t) is infinitely p(x) - differentiable function at a neighborhood of a point (x0,t),x0>0 . Then u(x,t) has the following space fractional power series expansion: (2.6) u(x,t)=∑k=0∞1Γ(kp(x)+1)[Dxkp(x)u(x,t)](x0,t)(x−x0)kp(x), for 0<p(x)≤1 where Dtkp(x) denotes the application of time fractional partial derivative of variable-order p(x) for k -times. Theorem 2.4 Assume that the function u(x,t) is infinitely q(t) - differentiable function at a neighborhood of a point (x,t0),t0>0 . Then u(x,t) has the fractional power series expansion: (2.7) u(x,t)=∑h=0∞1Γ(hq(t)+1)[Dthq(t)u(x,t)](x,t0)(t−t0)kq(t), for 0<q(t)≤1 where Dtkq(t) denotes the application of time fractional partial derivative of variable-order q(t) for h -times. 3 Mathematical modeling and analytical technique 3.1 Formulation of the SARS-CoV-2 infection model In this section, we study the dynamics transmission of SARS-CoV-2 through a mathematical model governed by PDEs of fractional variable-order derivatives. This study is in a general scenario, so the natural history of the disease is not used in the modeling process. However, in Fig. 1, we draw the schematic diagram of the proposed model. The arrows from e and i to s (in the beta coefficients) do not describe the natural history of SARS-CoV-2 and are considered asymptomatic infectious. Also, the exposed are those who are infected and still do not infect or still do not have symptoms. From the above diagram, we model the susceptible, exposed, infected, recovered, and deceased (SEIRD-model) of the host population’s survival during the transmission and evolution of SARS-CoV-2 in terms of PDEs with time-fractional variable-order partial derivatives as follows: (3.1) Dtq(t)s(x,t)=αn(x,t)−μs(x,t)−βs(x,t)i(x,t)−βs(x,t)e(x,t)+∇x⋅(nνs∇xs(x,t)),Dtq(t)e(x,t)=(−ρ−ϵ−μ)e(x,t)+βs(x,t)i(x,t)+βs(x,t)e(x,t)+∇x⋅(nνe∇xe(x,t)),Dtq(t)i(x,t)=ρe(x,t)+βs(x,t)i(x,t)+(−η−ζ−μ)i(x,t)+∇x⋅(nνi∇xi(x,t)),Dtq(t)r(x,t)=ζi(x,t)−ϵe(x,t)−μr(x,t)+∇x⋅(nνr∇xr(x,t)),Dtq(t)d(x,t)=ηi(x,t), subject to the following initial conditions: (3.2) s(x,t0)=s0(x)=n(x,t0)−i0(x),e(x,t0)=e0(x)=i(x,t0)=i0(x),r(x,t0)=r0(x)=n(x,t0)−s0(x)−i0(x),d(x,t0)=0,n(x,t0)=n0(x)=s0(x)+e0(x)+i0(x)+r0(x). for (x,t)∈Ω=[0,T]×[0,T],T≤1, n(x,t)>0, ∇x=∂∂x where x is the space (special location) of the host population, t is the time of the transmission of infection, α is the birth rate, μ is the general (non-SARS-CoV-2) mortality rate, β is the infection rate, ϵ is the asymptomatic recovery, ρ is the inverse of the incubation period, ζ is the infected recovery rate, η is the deceased mortality rate, and νs,νe,νi and νr are diffusion parameters respectively corresponding to the different population groups. Here s(x,t),e(x,t),i(x,t),r(x,t), and d(x,t) denote the densities of the susceptible s¯=∫Ωs(x,t)dx, exposed e¯=∫Ωe(x,t)dx, infected i¯=∫Ωi(x,t)dx, recovered r¯=∫Ωr(x,t)dx, and deceased populations d¯=∫Ωd(x,t)dx respectively where n(x,t)=s(x,t)+i(x,t)+r(x,t)+d(x,t) and Dtq(t) is the Caputo time-fractional partial derivative of a variable-order q(t). For any nonnegative values of s0,e0,i0,r0, one can claim that (3.3) Dtq(t)∫A[s(x,t)+e(x,t)+i(x,t)+r(x,t)]dA=0, where A⊂Ω, and it implies that (3.4) ∫A[s(x,t)+e(x,t)+i(x,t)+r(x,t)]dA≡n, where n is the total size of the population. Indeed, Eq. (3.4) shows that the size of the population in Ω is a constant.Fig. 1 The schematic description of SEIRD model for the SARS-CoV-2 infectious disease. 3.2 Description of the solution technique In this section, we introduce an approximate analytical method to solve a nonlinear system of variable time-fractional order PDEs of the following form: (3.5) Dtq(t)uj(x,t)=fj(x,t)+Lju¯(x,t)+Nju¯(x,t),k−1<q(t)<k,∂iuj(x,t0)∂ti=fj(x),t0≥0,i=0,1,…,k−1,j=1,2,…,m, where u¯(x,t)=(u1(x,t),u2(x,t),…,um(x,t)), and Lj,Nj are linear and nonlinear operators respectively of u¯=u¯(x,t) and its derivatives which might include other fractional derivatives of orders other than q(t), and fj(x,t) and fj(x),j=1,2,…,m, are known analytic functions and Dtq(t) is the Caputo time fractional partial derivative of variable-order q(t). In the case of fj(x,t)=0, the system (3.5) becomes in the homogeneous form. In order to solve system (3.5), we assume that the solution function uj(x,t) can be written as uj(x,t)=fj(x)gj(t), where fj(x) and gj(t) are analytical functions. Theorem 3.1 Assume that u¯(x,t):R×(0,∞)→R . For u¯(x,t)=f¯(x)g¯(t) , with assumption that f¯(x) and g¯(t) are analytical functions, then the nonlinear operator Nju¯(x,t) has the following time-fractional power series: (3.6) Nju¯(x,t)=∑n=0∞(t−t0)nq(t)Γ(nq(t)+1)Dtnq(t)Nj(u¯(x,t))\|(x,t0), for j=1,2,…,m . Theorem 3.2 Assume that a function u¯(x,t) is defined as u¯(x,t):R×(0,∞)→R , then for u¯=∑k=0∞pkα(x)u¯k(x,t) , the following property of nonlinear operators Nju¯(x,t) is satisfied the (3.7) Nju¯=Nj∑k=0∞pkq(t)u¯k=∑n=0∞[1Γ(nq(t)+1)Dpnq(t)[Nj∑k=0npkq(t)u¯k]p=0]pnq(t), where 0<q(t)<1 and j=1,2,…,m . Proof The proof of Theorem 3.2 is similar to the proof of Theorem 3 in [32]. □ Definition 4 The polynomials Pnj=Pnj(u0j,u1j,…,unj), for j=1,2,…,m, are defined as (3.8) Pnj=1Γ(nq(t)+1)Dpnq(t)[Nj∑k=0npkq(t)u¯k]p=0. Remark 3.1 Assume that Pnj=Pnj(u0j,u1j,…,unj), then by using Definition 4 and Theorem 3.1, the operators Nju¯λ can be expressed in terms of Pnj as (3.9) Nju¯=∑n=0∞Pnj,j=1,2,…,m. Theorem 3.3 Let uj(x,t) be analytic functions for j=1,2,…,m , then there exists at least a solution for the system (3.5) given by (3.10) uj(x,t)=fjt(−q(t))(x,t)+∑i=0k−1tii!fij(x)+(∑r=0∞Ljt(−q(t))u¯(r−1)+P(r−1)jt(−q(t))(u0j,u1j,…,urj)), where fjt(−q(t))(x,t), Ljt(−q(t))=It−q(t)Lju¯, and Pjt(−q(t))(u0j,u1j,…,urj)=It(−q(t))Pj(u0j,u1j,…,urj) denote the Caputo time-fractional partial integral of variable-order q(t) for fj(x,t),Lju¯ and Pj(u0j,u1j,…,urj) respectively. Proof As uj are analytic functions for j=1,2,…,m, then uj can be expressed as (3.11) uj(x¯,t)=∑r=0∞urj(x¯,t),j=1,2,…,m. Consider the following homotopy system: (3.12) Dtq(t)uλj(x,t)=λ(fj(x,t)+Lju¯(x,t)+Nju¯(x,t)), where λ∈[0,1] is an embedding parameter. Next, we assume the solutions uλj(x¯,t) for (3.5) are given by (3.13) uλj(x¯,t)=∑r=0∞λrurj(x¯,t),j=1,2,…,m. Plugging (3.13) into using (3.12) and using, Definition 1 and Theorem 2.1, the components of the series (3.13) can be obtained as follows: (3.14) u0j(x,t)=∑i=0k−1tii!fij(x),u1j(x,t)=fjt(−q(t))(x,t)+Ltj(−q(t))u¯0+P0jt(−q(t)),urj(x,t)=Ljt(−q(t))u¯(r−1)+P(r−1)jt(−q(t)), for r=2,3,…,j=1,2,…,m. By substituting (3.14) into (3.13) and by letting λ→1, it proves the existence of solution for (3.5). □ Theorem 3.4 Let B be a Banach space. Then the series of solutions given by (3.14) converges to Sj∈B for j=1,2,…,m , if there exists λj,0≤λj<1 such that, ‖unj‖≤λj‖u(n−1)j‖ for ∀n∈N . Proof See [32]. □ 4 Analytical solutions for SEIRD-model By comparing (3.1) with (3.5) for j=1,2,…,5, we have (4.1) Dtq(t)s(x,t)=f1(x,t)+L1(s,e,i,r,d)+N1(s,e,i.r,d),Dtq(t)e(x,t)=L2(s,e,i,r,d)+N2(s,e,i,r,d),Dtq(t)i(x,t)=L3(s,e,i,r,d)+N3(s,e,i,r,d),Dtq(t)r(x,t)=L4(s,e,i,r,d)+N4,(s,e,i,r,d),Dtq(t)d(x,t)=L5(s,e,i,r,d), By comparing the system (4.1) with the system (3.5), we get (4.2) f1(x,t)=αn,f2(x,t),…,f5(x,t)=0,L1(s,e,i,r,d)=−μs(x,t),L2(s,e,i,r,d)=(−ρ−ϵ−μ)e(x,t),L3(s,e,i,r,d)=ρe(x,t)+(−η−ζ−μ)i(x,t),L4(s,e,i,r,d)=ζi(x,t)−ϵe(x,t)−μr(x,t),N1(s,e,i,r,d)=βs(x,t)i(x,t)−βs(x,t)e(x,t)+∇x⋅nνs∇xs,N2(s,e,i,r,d)=βs(x,t)i(x,t)+βs(x,t)e(x,t)+∇x⋅nνe∇xe,N3(s,e,i,r,d)=βs(x,t)i(x,t)+∇x⋅nνi∇xi,N4(s,e,i,r,d)=∇x⋅nνr∇xr,N5(s,e,i,r,d)=0. To solve the system (4.1), we assume that the solutions of (4.1) have the following forms: (4.3) s(x,t)=∑r=0∞sr(x,t),e(x,t)=∑r=0∞er(x,t),i(x,t)=∑r=0∞ir(x,t),r(x,t)=∑r=0∞rr(x,t),d(x,t)=∑r=0∞dr(x,t). Thus from the system (3.14), we obtain (4.4) s0(x,t)=∑i=0k−1tii!si(x,0),e0(x,t)=∑i=0k−1tii!ei(x,0),i0(x,t)=∑j=0k−1tjj!ij(x,0),r0(x,t)=∑i=0k−1tii!ri(x,0),d0(x,t)=∑i=0k−1tii!di(x,0), and (4.5) s1(x,t)=f1(x,t)+L1t(−q(t))+P1t(−q(t))(s0,e0,i0,r0,d0),e1(x,t)=f2(x,t)+L2t(−q(t))+P2t(−q(t))(s0,e0,i0,r0,d0),i1(x,t)=f3(x,t)+L3t(−q(t))+P3t(−q(t))(s0,e0,i0,r0,d0),r1(x,t)=f4(x,t)+L4t(−q(t))+P4t(−q(t))(s0,e0,i0,r0,d0),d1(x,t)=f5(x,t)+L5t(−q(t))(s0,e0,i0,r0,d0), and (4.6) sr′(x,t)=L1t(−q(t))(sr′−1,er′−1,ir′−1,rr′−1,dr′−1)+P1t(−q(t))(∑l=0r′−1sl,∑l=0r′−1el,∑l=0r′−1il,∑l=0r′−1rl,∑l=0r′−1dl),er′(x,t)=L2t(−q(t))(sr′−1,er′−1,ir′−1,rr′−1,dr′−1)+P2t(−q(t))(∑l=0r′−1sl,∑l=0r′−1el,∑l=0r′−1il,∑l=0r′−1rl,∑l=0r′−1dl),ir′(x,t)=L3t(−q(t))(sr′−1,er′−1,ir′−1,rr′−1,dr′−1)+P3t(−q(t))(∑l=0r′−1sl,∑l=0r′−1el,∑l=0r′−1il,∑l=0r′−1rl,∑l=0r′−1dl),rr′(x,t)=L4t(−q(t))(sr′−1,er′−1,ir′−1,rr′−1,dr′−1)+P4t(−q(t))(∑l=0r′−1sl,∑l=0r′−1el,∑l=0r′−1il,∑l=0r′−1rl,∑l=0r′−1dl),dr′(x,t)=L5t(−q(t))(sr′−1,er′−1,ir′−1,rr′−1,dr′−1), where Pjt(−q(t))=It(−q(t))Pj(sj,ej,ij), j=1,2,…,5 are obtained by using Definition 4 and Remark 3.1. By evaluating the components of (4.4)–(4.6) and substituting the results along with the values from (3.2), (4.2) into (4.3), we obtain the approximate analytical solutions for the model (3.1). In particular, we study the model (3.1) with the initial values (4.7) s(x,0)=1+βx−e−βx,r(x,0)=d(x,0)=0,e(x,0)=i(x,0)=−βx+e−βx, and 0<q(t)<1. Therefore, from the systems (4.4), (4.7), we obtain (4.8) s0(x,t)=1+βx−e−βx,e0(x,t)=i0(x,t)=−βx+e−βx,r0(x,t)=d0(x,t)=0, and (4.9) s1(x,t)=tq(t)(α+β(−eβ(−x)(βνs+4βx+2)+2βx(βx+1)+2e−2βx)+μ(β(−x)+eβ(−x)−1))tq(t)Γ(q(t)+1),e1(x,t)=(β(eβ(−x)(βνe+4βx+2)−2βx(βx+1)−2e−2βx)+(eβ(−x)−βx)(−μ+ρ+ϵ))tq(t)Γ(q(t)+1),i1(x,t)=(β(eβ(−x)(βνi+2βx+1)+β(−x)(βx+1)−e−2βx)+(eβ(−x)−βx)(−ζ−η−μ)+ρ(eβ(−x)−βx))tq(t)Γ(q(t)+1),r1(x,t)=tq(t)eβ(−x)βxeβx−1(ϵ−ζ)Γ(q(t)+1),d1(x,t)=ηtq(t)eβ(−x)−βxΓ(q(t)+1), and (4.10) s2(x,t)=(βe−3βx(−7β−e2βx(2α+β(β(βx+1)νe+νi+β2νs2+νs(6β(βx−1)−μ))−ζ−η−4μ+2ρ+βx20β−2ζ−2η−8μ+4ρ+21β2x+2ϵ+3+ϵ)+eβx(β2νe+νi+10νs−ζ−η−4μ+2ρ+β(21βx+10)+ϵ)+βe3βx(4βνs+2αx+x(βx+1)×(−ζ−η−4μ+2ρ+β(7βx+3)+ϵ)))−μ(α+β(−eβ(−x)βνs+4βx+2+2βx(βx+1)+2e−2βx)+μ(β(−x)+eβ(−x)−1)))t2q(t)Γ(2q(t)+1),e2(x,t)=e−3βx(β(β3νe2e2βx+βνee2βx(−μ+ρ+5β(βx−1)+5βsinh(βx)−13βcosh(βx)+ϵ)+β2νieβx(eβx(βx+1)−1)−(βxeβx−1)(7β−2β2νseβx+eβx(ζ+η+4μ−2ρ+eβx(2α+(βx+1)(−ζ−η−4μ+2ρ+β(7βx+3)+ϵ))−2β(7βx+5)−ϵ)))+eβx(μ+ρ+ϵ)(2β−eβx(2β+β2νe−μ+ρ+4β2x+ϵ)+βxe2βx(−μ+ρ+2β(βx+1)+ϵ)))t2q(t)Γ(2q(t)+1), and i2(x,t)=e−3βx(β2ρνee2βx+β4νi2e2βx+β2νie2βx(−2(ζ+η+μ)+ρ+β(3βx−2)+3βsinh(βx)−7βcosh(βx))−(βxeβx−1)(3β2−βeβx(4β−2ζ−2η−3μ+3ρ+β2νs+6β2x)+e2βx(αβ−β(2ζ+2η+3μ−3ρ)−ρ(ζ+η+2μ)+(ζ+η+μ)2+ρ2+3β4x2+4β3x+β2(1−x(2ζ+2η+3μ−3ρ))+ρϵ)))×t2q(t)Γ(2q(t)+1),r2(x,t)=e−2βx((βxeβx−1)(β(ζ−2ϵ)+eβx(β(2ϵ−ζ)+ζ(ζ+η+2μ−ρ)+β2x(2ϵ−ζ)+ϵ2+ϵ(ρ−2μ)))−β2eβx(ϵνe−ζνi+νr+ϵνr))×t2q(t)Γ(2q(t)+1),d2(x,t)=e−2βx(β2νieβx−(βxeβx−1)(eβx(β−ζ−η−μ+ρ+β2x)−β))ηt2q(t)Γ(2q(t)+1), and so on. Therefore, the approximate analytical solutions for the system (3.2) are (4.11) s(x,t)=∑i=03si(x,t),s(x,t)=∑i=03ei(x,t),i(x,t)=∑j=03ij(x,t),r(x,t)=∑i=03ri(x,t),d(x,t)=∑i=03di(x,t). 5 Numerical simulations for the SEIRD-model In this section, we present the numerical study for the host populations (susceptible, exposed, infected, recovered, and deceased). In particular cases, we obtain the numerical solutions for the (3.1) subject to the initial conditions (3.2) with the following values: (5.1) α=μ=0,β=0.0139,ρ=0.0714,ϵ=0.1667,ζ=0.7348,η=0.0218,νs=0.05,νe=0.025,νi=0.04,νr=0.08,q(t)=0.5et−0.36t. The numerical solutions through various values of x and t for the susceptible s(x,t), exposed e(x,t), infected i(x,t), recovered r(x,t), and deceased d(x,t) host populations were obtained in Table 1, Table 2, Table 3, Table 4, and Table 5 respectively. The graphs of solutions for the SEIRD model (3.1) subject to the assumed values of the parameters given by (5.1) are described as follows: Fig. 2 shows the graphs of the susceptible population size s(x,t) and s(t). It is noticed that the susceptible cases for certain groups of the population decrease during the time of the pandemic. Moreover, we see that the susceptible cases are smoothly decreasing. Fig. 3 presents the graphs of the exposed population e(x,t) and e(t). It is observed that the exposed population curve is smoothly decreasing. In addition, the exposed cases are less during the time of the pandemic eventually when x is large. Fig. 4 shows the graphs of the infected population i(x,t) and i(t). However, the rate of the infected population decreases during the time of the pandemic. Also, we noticed that the curve of the infected population smoothly decreases eventually when x is larger. Fig. 5 shows the graphs of recovered population r(x,t) and r(t). The recovered population size increases up to a certain time during the pandemic and starts decreasing. Moreover, we see that the curve of the recovered population is smoothly increasing during the pandemic time. Fig. 6 presents the graphs of the deceased population d(x,t) and d(t). We see that the curve of the deceased population is smoothly increasing when after some time during the pandemic. Fig. 7 presents the graphs of the totall population n(x,t) and n(t). This confirms the accuracy of the solutions where we assumed that the n(x,t)=1. Moreover, Table 6 shows the numerical values of the totall population n(x,t). The susceptible population size is initially reduced due to the distribution of the susceptible cases between the majority of recovered and deceased cases. This confirms the increasing rates of susceptible cases in most countries. Table 1 Numerical values of the of susceptible s(x,t) among different values of x,t when q(t)=0.5et−0.36t with the initial values given by (5.1). x t q(t) s0(x,t) s1(x,t) s2(x,t) s3(x,t) ∑i′=03si′(x,t) 0.25 0.552013 0.00694397 −0.000105354 0.0000199046 −4.10338×10−6 0.00685442 0.25 0.50 0.644361 0.00694397 −0.000143232 0.0000340599 −8.50006×10−6 0.0068263 0.75 0.788500 0.00694397 −0.000172848 0.0000435937 −0.0000108524 0.00680386 0.25 0.552013 0.0138759 −0.00020408 0.0000389249 −8.03176×10−6 0.0137027 0.50 0.50 0.644361 0.0138759 −0.000277453 0.0000666066 −0.0000166363 0.0136484 0.75 0.788500 0.0138759 −0.000334822 0.0000852508 −0.0000212372 0.0136051 0.25 0.552013 0.0207958 −0.00030124 0.0000576174 −0.0000118887 0.0205403 0.75 0.40 0.644361 0.0207958 −0.000409546 0.0000985926 −0.0000246241 0.0204603 0.75 0.788500 0.0207958 −0.000494228 0.00012619 −0.0000314316 0.0203964 Fig. 2 The 2D and 3D graphs of the susceptible population s(x,t) among different values of x and t (on the left when α=0.5) and s(t) (on the right when x=0.5) subject to the initial values given in (5.1) when q(t)=0.5et−0.36t. It is observed that the susceptible cases, for certain groups of the population, decrease during the time of the pandemic. It is interesting to see how the susceptible cases are smoothly decreasing through the fractional order function. Table 2 Numerical values of the exposed e(x,t) among different values of x,t when q(t)=0.5et−0.36t with the initial values given by (5.1). x t q(t) e0(x,t) e1(x,t) e2(x,t) e3(x,t) ∑i′=03ei′(x,t) 0.25 0.552013 0.993056 −0.123628 0.0115908 −0.000898086 0.880121 0.25 0.50 0.644361 0.993056 −0.168077 0.0198337 −0.00185202 0.842961 0.75 0.788500 0.993056 −0.20283 0.0253854 −0.00234502 0.813267 0.25 0.552013 0.986124 −0.122666 0.0114814 −0.000885642 0.874054 0.50 0.50 0.644361 0.986124 −0.166768 0.0196465 −0.00182632 0.837176 0.75 0.788500 0.986124 −0.201251 0.0251458 −0.00231239 0.807707 0.25 0.552013 0.979204 −0.121706 0.0113726 −0.000873317 0.867997 0.75 0.50 0.644361 0.979204 −0.165464 0.0194603 −0.00180087 0.831400 0.75 0.788500 0.979204 −0.199677 0.0249075 −0.00228009 0.802155 Fig. 3 The 2D and 3D graphs of the exposed population e(x,t) among different values of x and t (on the left when α=0.5) and e(t) (on the right when x=0.5) subject to the initial values given in (5.1) when q(t)=0.5et−0.36t. It is expected to have fewer exposed cases during the time of the pandemic eventually when x is large. Also, we can see that the exposed population curve is smoothly decreasing. Table 3 Numerical values of the infected i(x,t) among different values of x and t when q(t)=0.5et−0.36t with the initial values given by (5.1). x t q(t) i0(x,t) i1(x,t) i2(x,t) i3(x,t) ∑i′=03ii′(x,t) 0.25 0.552013 0.993056 −0.356017 0.10275 −0.0251395 0.71465 0.25 0.50 0.644361 0.993056 −0.484017 0.175822 −0.0518433 0.633018 0.75 0.788500 0.993056 −0.584096 0.225037 −0.0656461 0.568351 0.25 0.552013 0.986124 −0.353482 0.102007 −0.0249548 0.709695 0.50 0.50 0.644361 0.986124 −0.48057 0.1745500 −0.0514623 0.628641 0.75 0.788500 0.986124 −0.579937 0.223409 −0.0651637 0.564432 0.25 0.552013 0.979204 −0.350952 0.101265 −0.0247705 0.704747 0.75 0.50 0.644361 0.979204 −0.477131 0.173281 −0.0510823 0.624272 0.75 0.788500 0.979204 −0.575787 0.221785 −0.0646825 0.560520 Fig. 4 The 2D and 3D graphs of the infected population i(x,t) among different values of x and t (on the left when α=0.5) and i(t) (on the right when x=0.5) subject to the initial values given in (5.1) when q(t)=0.5et−0.36t. However, the number of infected cases decreases during the time of the pandemic. Also, the curve of the infected population smoothly decreases eventually when x is large. Table 4 Numerical values of the recovered r(x,t) among different values of x and t when q(t)=0.5et−0.36t with the initial values given by (5.1). x t q(t) r0(x,t) r1(x,t) r2(x,t) r3(x,t) ∑i′=03ri′(x,t) 0.25 0.552013 0.000000 0.295219 −0.111327 0.0253128 0.209205 0.25 0.50 0.644361 0.00000 0.401360 −0.190498 0.052201 0.263063 0.75 0.788500 0.00000 0.484348 −0.243821 0.066099 0.306626 0.25 0.552013 0.00000 0.293158 −0.110542 0.025128 0.207744 0.50 0.50 0.644361 0.000000 0.398558 −0.189154 0.0518201 0.261224 0.75 0.788500 0.00000 0.480967 −0.242101 0.065617 , 0.304483 0.25 0.552013 0.000000 0.291101 −0.109758 0.024944 0.206287 0.75 0.50 0.644361 0.000000 0.395761 −0.187814 0.051441 0.259388 0.75 0.788500 0.00000 0.477592 −0.240385 0.065136 0.302343 Fig. 5 The 2D and 3D graphs of recovered population r(x,t) among different values of x and t (on the left) and r(t) (on the right when x=0.5) subject to the initial values given in (5.1) when q(t)=0.5et−0.36t. The rate of the recovered population can be increased up to a certain time during the pandemic and starts decreasing. However, the curve of the recovered population is smoothly increasing during the time of the pandemic eventually when x is large. Table 5 Numerical values of the deceased d(x,t) among different values of x,t when q(t)=0.5et−0.36t with the initial values given by (5.1). x t q(t) d0(x,t) d1(x,t) d2(x,t) d3(x,t) ∑i′=03di′(x,t) 0.25 0.552013 0.000000 0.011329 −0.003061 0.000782 0.009050 0.25 0.50 0.644361 0.00000 0.015402 −0.005240 0.001613 0.011776 0.75 0.788500 0.000000 0.018586 −0.006705 0.002042 0.013923 0.25 0.552013 0.000000 0.0112495 −0.003040 0.0007763 0.008986 0.50 0.50 0.644361 0.000000 0.015294 −0.005201 0.001601 0.011694 0.75 0.788500 0.000000 0.018456 −0.006657 0.002027 0.0138265 0.25 0.552013 0.000000 0.011171 −0.003018 0.000771 0.008923 0.75 0.50 0.644361 0.000000 0.015187 −0.005164 0.001589 0.011612 0.75 0.788500 0.000000 0.018327 −0.006609 0.002012 0.013730 Fig. 6 The 2D and 3D graphs of the deceased population d(x,t) among different values of x and t (on the left) and d(t) (on the right when x=0.5) subject to the initial values given in (5.1) when q(t)=0.5et−0.36t. However, the curve of the deceased population is smoothly increasing after some time during the pandemic regardless of how large x is. Table 6 Numerical comparison between the susceptible, exposed, infected, recovered, and deceased population with the initial values given by (5.1). x t q(t)=0.5et−0.36t s(x,t) e(x,t) i(x,t) r(x,t) d(x,t) n(x,t) 0.25 0.006854 0.880121 0.714650 0.209205 0.009049 0.939758 0.25 0.50 0.006826 0.842961 0.633018 0.263063 0.011776 , 0.914683 0.75 0.006804 0.813267 0.568351 0.306626 0.013923 0.895704 0.25 0.013703 0.874054 0.709695 0.207744 0.008986 0.940128 0.50 0.50 0.013648 0.837176 0.628641 0.261224 0.011694 0.915207 0.75 0.013605 0.807707 0.564432 0.304483 0.013827 0.896347 0.25 0.020540 0.867997 0.704747 0.206287 0.008923 0.940498 0.75 0.50 0.020460 0.831400 0.624272 0.259388 0.011612 0.915732 0.75 0.020396 0.802155 0.560520 0.302343 0.013730 0.896989 Fig. 7 The 2D and 3D graphs of the total population size n(x,t) among different values of x,t (on the left), and n(x)&n(t) (on the right when t=0.5 and x=0.5) for q(t)=0.5et−0.36t subject to the initial values given by (5.1). 6 Conclusion In the present work, we introduced a fractional mathematical model based on nonlinear Partial Differential Equations (PDEs) of fractional variable-order derivatives which describe the host populations surviving on the transmission and evolution of the SARS-CoV-2 pandemic. Five host population groups have been considered, the susceptible, exposed, infected, recovered, and deceased (SEIRD). Analytical and numerical solutions for the proposed biological model were successfully obtained. The proposed model is new and has not been introduced before in its current formulation, therefore, the proposed technique is not compared with other techniques or real scenarios for the same model. The motivation and support of the used fractional partial derivatives of variable orders that can model the rate of change of subpopulation for the debated model since it relies upon the characteristics of the entire solution functions (on time for all values of the fractional order function q(t)) and not just at the values in the vicinity of particular points, which is helpful to enhance the performance of the dynamics of the transmission and evolution of infectious diseases. As efficient analytical tools, we present an analytical technique based on the homotopy and Adomian decomposition methods for obtaining the solutions for the proposed model. The analytical solutions of the fourth-order terms have been neglected as they almost tend to be zeros. However, the present study is in general and does not depend on having access to people, organizations, data, or documents, and it can be applied to the host population in any country which can help inform the public health interventions. The interesting result of this article is that it yields approximate analytical solutions to the proposed model which can help significantly to predict and control the transmission of infectious diseases. However, several analytical solutions for SEIRD models obtained by different analytical methods are available in the literature. By comparing the proposed model in this paper with those previously introduced (see, for example, [33], [34], [35], [36], [37]), our model is more generalized with a time-fractional variable-order derivative. Further, most of the solutions for epidemic models are numerical (see, for example, [38], [39], [40]), so analytical exact or approximate solutions are rarely available. Therefore, the analytical solution of the proposed epidemic model obtained in this paper is new and potentially helpful to describe the population dynamics of infectious diseases transmission and controlling infectious diseases. Mathematica and Maple software have been used to obtain the analytical and numerical results. CRediT authorship contribution statement Hayman Thabet: Wrote the paper. Subhash Kendre: Developing, evaluating, and editing the paper. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability No data was used for the research described in the article. Acknowledgment The authors acknowledge that a preprint of this work has previously been published [41]. Declarations Ethical approval This research does not contain any studies involving animal or human participants, nor does it take place in any private or protected areas. Funding The work in this paper is partially supported by the Division of Applied Mathematics, Brown University, Providence , RI 02906, United States of America. ==== Refs References 1 Shikongo A. Nuugulu S.M. Elago D. Salom A.T. Owolabi K.M. Fractional derivative operator on quarantine and isolation principle for COVID-19 Advanced Numerical Methods for Differential Equations 2021 CRC Press 205 226 2 Thabet H. Kendre S. Elementary Course in Fractional Calculus 2018 LAP LAMBERT Academic Publishing 3 Thabet H. Kendre S. Peters J. Travelling wave solutions for fractional Korteweg-de Vries equations via an approximate-analytical method AIMS Math. 4 4 2019 1203 1222 10.3934/math.2019.4.1203 4 Thabet H. Kendre S. Peters J. Analytical solutions for nonlinear systems of conformable space-time fractional partial differential equations via generalized fractional differential transform Vietnam J. Math. 47 2019 487 507 5 Jacobs B.A. Harley C. Application of nonlinear time-fractional partial differential equations to image processing via hybrid laplace transform method J. Math. 2018 2018 10.1155/2018/8924547 6 Le U. Pelinovsky D.E. Convergence of Petviashvili’s method near periodic waves in the fractional Korteweg–de Vries equation SIAM J. Math. Anal. 51 4 2019 2850 2883 10.1137/18M1215050 7 Thabet H. Kendre S. Peters J. Advances in solving conformable nonlinear partial differential equations and new exact wave solutions for Oskolkov-type equations Math. Methods Appl. Sci. 45 5 2022 2658 2673 10.1002/mma.7945 8 Thabet H. Kendre S. Peters J. Extended least square homotopy method for solving nonlinear systems of fractional partial differential equations Preprint 2022 0 24 10.21203/rs.3.rs-2223638/v1 9 Organization W.H. Coronavirus disease (COVID-19) outbreak: rights, roles and responsibilities of health workers, including key considerations for occupational safety and health: Interim guidance, 19 march 2020 2020 10 Fauci A.S. Lane H.C. Redfield R.R. COVID-19 — Navigating the uncharted N. Engl. J. Med. 382 13 2020 1268 1269 10.1056/nejme2002387 32109011 11 Yuki K. Fujiogi M. Koutsogiannaki S. COVID-19 pathophysiology: A review Clin. Immunol. 215 April 2020 10.1016/j.clim.2020.108427 12 Khan F.M. Gupta R. ARIMA and NAR based prediction model for time series analysis of COVID-19 cases in India J. Saf. Sci. Resil. 1 1 2020 12 18 10.1016/j.jnlssr.2020.06.007 13 Atangana A. Araz S.I. A novel Covid-19 model with fractional differential operators with singular and non-singular kernels: Analysis and numerical scheme based on Newton polynomial Alex. Eng. J. 60 4 2021 3781 3806 10.1016/j.aej.2021.02.016 14 Thabet H. Kendre S. Conformable mathematical modeling of the COVID-19 transmission dynamics: A more general study Authorea Preprints 2022 15 Ghosh K. Ghosh A.K. Study of COVID-19 epidemiological evolution in India with a multi-wave SIR model Nonlinear Dynam. 109 1 2022 47 55 10.1007/s11071-022-07471-x arXiv:2202.04917 16 Lu Z. Yu Y. Chen Y.Q. Ren G. Xu C. Wang S. Stability analysis of a nonlocal SIHRDP epidemic model with memory effects Nonlinear Dynam. 109 1 2022 121 141 10.1007/s11071-022-07286-w 17 Debbouche N. Ouannas A. Batiha I.M. Grassi G. Chaotic dynamics in a novel COVID-19 pandemic model described by commensurate and incommensurate fractional-order derivatives Nonlinear Dynam. 109 1 2022 33 45 10.1007/s11071-021-06867-5 18 Chu Y.M. Zarin R. Khan A. Murtaza S. A vigorous study of fractional order mathematical model for SARS-CoV-2 epidemic with Mittag-Leffler kernel Alex. Eng. J. 71 2023 565 579 10.1016/j.aej.2023.03.037 19 Bilgil H. Yousef A. Erciyes A. Erdinç Ü. Öztürk Z. A fractional-order mathematical model based on vaccinated and infected compartments of SARS-CoV-2 with a real case study during the last stages of the epidemiological event J. Comput. Appl. Math. 425 2023 115015 10.1016/j.cam.2022.115015 20 DarAssi M.H. Safi M.A. Khan M.A. Beigi A. Aly A.A. Alshahrani M.Y. A mathematical model for SARS-CoV-2 in variable-order fractional derivative Eur. Phys. J.: Spec. Top. 231 10 2022 1905 1914 10.1140/epjs/s11734-022-00458-0 35154580 21 Haq I.U. Yavuz M. Ali N. Akgül A. A SARS-CoV-2 fractional-order mathematical model via the modified Euler method Math. Comput. Appl. 27 5 2022 82 10.3390/mca27050082 22 Aba Oud M.A. Ali A. Alrabaiah H. Ullah S. Khan M.A. Islam S. A fractional order mathematical model for COVID-19 dynamics with quarantine, isolation, and environmental viral load Adv. Difference Equ. 2021 1 2021 10.1186/s13662-021-03265-4 23 Saeed U. Ur Rehman M. Haar wavelet Picard method for fractional nonlinear partial differential equations Appl. Math. Comput. 264 2015 310 322 10.1016/j.amc.2015.04.096 24 Altaf Khan M. Ullah S. Farooq M. A new fractional model for tuberculosis with relapse via Atangana–Baleanu derivative Chaos Solitons Fractals 116 2018 227 238 10.1016/j.chaos.2018.09.039 25 Abu Arqub O. Application of residual power series method for the solution of time-fractional Schrödinger equations in one-dimensional space Fundamenta Informaticae vol. 166 2019 87 110 26 Sinan M. Ahmad H. Ahmad Z. Baili J. Murtaza S. Aiyashi M.A. Botmart T. Fractional mathematical modeling of malaria disease with treatment & insecticides Results Phys. 34 December 2021 2022 10.1016/j.rinp.2022.105220 27 Diethelm K. The Analysis of Fractional Differential Equations: An Application-Oriented Exposition using Differential Operators of Caputo Type 2010 Springer 28 Miller K.S. Ross B. An Introduction to the Fractional Calculus and Fractional Differential Equations 1993 Wiley-Interscience 29 Podlubny I. Fractional Differential Equations: An Introduction to Fractional Derivatives, Fractional Differential Equations, to Methods of their Solution and Some of their Applications, Vol. 198 1998 Academic Press 30 Thabet H. Kendre S. New modification of Adomian decomposition method for solving a system of nonlinear fractional partial differential equations Int. J. Adv. Appl. Math. AndMech. 6 3 2019 1 13 31 Thabet H. Kendre S. Unhale S. Numerical analysis of iterative fractional partial integro-differential equations J. Math. 2022 2022 1 14 10.1155/2022/8781186 32 Thabet H. Kendre S. Chalishajar D. New analytical technique for solving a system of nonlinear fractional partial differential equations Mathematics 5 4 2017 1 15 10.3390/math5040047 33 Viguerie A. Lorenzo G. Auricchio F. Baroli D. Hughes T.J. Patton A. Reali A. Yankeelov T.E. Veneziani A. Simulating the spread of COVID-19 via a spatially-resolved susceptible–exposed–infected–recovered–deceased (SEIRD) model with heterogeneous diffusion Appl. Math. Lett. 111 2021 106617 10.1016/j.aml.2020.106617 arXiv:2005.05320 34 Pérez A.G.C. Oluyori D.A. A model for COVID-19 and bacterial pneumonia coinfection with community- and hospital-acquired infections Math. Model. Numer. Simul. Appl. 2 4 2022 197 210 10.53391/mmnsa.2022.016 35 Arshad S. Siddique I. Nawaz F. Shaheen A. Khurshid H. Dynamics of a fractional order mathematical model for COVID-19 epidemic transmission Physica A 609 2023 128383 10.1016/j.physa.2022.128383 36 Khalaf S.L. Kadhim M.S. Khudair A.R. Studying of COVID-19 fractional model: Stability analysis Partial Differ. Equ. Appl. Math. 7 June 2022 2023 100470 10.1016/j.padiff.2022.100470 37 Shyamsunder S.L. Bhatter S. Jangid K. Abidemi A. Owolabi K.M. Purohit S.D. A new fractional mathematical model to study the impact of vaccination on COVID-19 outbreaks Decision Anal. J. 6 October 2022 2023 100156 10.1016/j.dajour.2022.100156 38 Kumar A. Prakash A. Mehmet Baskonus H. The epidemic COVID-19 model via Caputo–Fabrizio fractional operator Waves Random Complex Media 2022 10.1080/17455030.2022.2075954 39 Ahmad S. Qiu D. ur Rahman M. Dynamics of a fractional-order COVID-19 model under the nonsingular kernel of Caputo-Fabrizio operator Math. Model. Numer. Simul. Appl. 2 4 2022 228 243 10.53391/mmnsa.2022.019 40 El-Sayed A.M. Arafa A. Hagag A. Mathematical model for the novel coronavirus (2019-nCOV) with clinical data using fractional operator Numer. Methods Partial Differential Equations 39 2 2023 1008 1029 41 Thabet H. Kendre S. Fractional mathematical modeling for the transmission dynamics of SARS-CoV-2 infection 2023 Authorea Preprints
PMC010xxxxxx/PMC10205127.txt	==== Front Operations Research Perspectives 2214-7160 2214-7160 The Author(s). Published by Elsevier Ltd. S2214-7160(23)00015-5 10.1016/j.orp.2023.100280 100280 Retraction Notice Retraction notice to “SIR model for the spread of COVID-19: A case study” [Operations Research Perspectives 10 (2023) 100265] Salimipour Ayoob a⁎ Mehraban Toktam a Ghafour Hevi Seerwan b Arshad Noreen Izza c Ebadi M.J. de⁎ a Department of Mathematics, Quchan University of Technology, Quchan, Iran b Department of Biomedical Sciences, Cihan University-Erbil, Kurdistan Region, Iraq c Positive Computing Research Group, Institute of Autonomous Systems, Department of Computer & Information Sciences, Universiti Teknologi Petronas, 32610, Bandar Seri Iskandar, Perak, Malaysia d Department of Mathematics, Chabahar Maritime University, Chabahar, Iran e Section of Mathematics, International Telematic University Uninettuno, Corso Vittorio Emanuele II, 39, 00186, Rome, Italy ⁎ Corresponding author. 24 5 2023 2023 24 5 2023 10 100280100280 © 2023 The Authors 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThis article has been retracted: please see Elsevier Policy on Article Withdrawal (https://www.elsevier.com/about/policies/article-withdrawal). This article has been retracted at the request of the Editor. The Editor of Operations Research Perspectives has retracted the publication based on evidence that the authors have plagiarized parts from another paper. The original publication is entitled ‘Mathematical Modeling of Epidemic Diseases; A Case Study of the COVID-19 Coronavirus’ (Reza Sameni, ArXiV, https://doi.org/10.48550/arXiv.2003.11371). The scientific community takes a very strong view on this matter and apologies are offered to readers of the journal that this was not detected during the submission process.
PMC010xxxxxx/PMC10288205.txt	==== Front IBM spibm International Bulletin of Mission Research 2396-9393 2396-9407 SAGE Publications Sage UK: London, England 10.1177/23969393221139456 10.1177_23969393221139456 Articles Casting a Vision of Communal and Public Life of the Church for the Post-Pandemic Korean-American Churches: Focusing on Veli-Matti Kärkkäinen’s Communion and Public Ecclesiology Shin JongSeock James America Evangelical University, Gardena, CA, USA JongSeock James Shin, America Evangelical University, Gardena, CA, USA. Email: jongseockshin@fuller.edu 22 6 2023 7 2023 22 6 2023 47 3 416429 © The Author(s) 2022 2022 Overseas Ministries Study Center This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. I engage with Veli-Matti Kärkkäinen’s communion and public ecclesiologies in casting a vision of communal and public life of post-pandemic Korean-American churches. For Kärkkäinen, the trinitarian economy of salvation constitutes the ontological ground of a church’s communal life in the wider society. The theologian advances a communal ecclesiology that celebrates irreducible particularities in unity of the Church while upholding a public ecclesiology through creative mutual interaction with other sectors of society in the imitatio Trinitatis. I discuss how his ecclesiological proposals may contribute to Korean-American churches by enhancing their socio-economic accountability, intercultural mutual respect, and intergenerational unity, while remaining deeply rooted in the Gospel of Christ. Veli-Matti Kärkkäinen public and missional ecclesiology Korean-American churches typesetterts1 ==== Body pmcIntroduction The churches in the US and Korea have been experiencing gradual decreases in congregation members participating in worship services. The dropouts mainly comprise the younger generations. In the US, according to the survey conducted by Barna Group 1 in 2019, the dropout rate of church members aged between 18 and 29 had risen from 59 percent to 64 percent since 2011. 2 In 2011, Barna Group published an initial survey investigating why youth and young adults have left their churches. 3 According to the report, many youngsters had left the churches for the following reasons. First, “Churches seem overprotective.” “As Christians, they express the desire for their faith in Christ to connect to the surrounding world. However, much of their experience of Christianity feels stifling, fear-based and risk-averse.” Their churches were defensive against new worldviews rather than trying to express the gospel in newly incarnated manners in response to the demand of contemporary times. Second, the young generations’ “experience of Christianity was shallow.” They responded to the survey that their experiences of church life had not been engaging, as the sermons and the church activities were not relevant to their life circumstances. Third, “Churches come across as antagonistic to science.” Fourth, they responded that “Christians are too confident they know all the answers.” Echoing this opinion, they also responded in the survey that “They wrestle with the exclusive nature of Christianity.” The responders felt that their churches were not in mutual conversations with other members of the wider society, but rather, they were isolated from external discourses and worldviews. These negative responses are not only found in the US. According to a recent article published in Christianity Today, 4 young Christians in their twenties and thirties in Korea call for their church leadership’s deeper understanding of their lives and more openness to the diversity and plurality in the contemporary era. They also demand more room for the marginalized and a more secure and generous space for open and candid debates regarding diverse social issues and challenges. In Korean-American churches, similar issues have been raised by the younger generations. The Korean-American churches have their distinct loci of life in an environment that is culturally more pluralistic in convergence and divergence among diverse cultural, socioeconomic, and generational frames of reference. One of the challenges that the Korean-American churches face is the lack of communication among local churches, their involvement in local communities, and the deficiency of mutual understanding between first-generation and second-generation Korean-Americans. For these reasons, many young members of Korean-American churches are leaving their congregations. Dubbed the “silent exodus” by Helen Lee in Christianity Today in 1996, second-generation Korean-Americans continue “a decades-long trend of leaving their parents’ churches—often to multi-ethnic congregations and, more recently, to non-affiliation.” 5 They seek another church that is more communal, interculturally openly communicative, and socially embedded. 6 Nonetheless, I believe that Korean-American churches have a significant potential given by God for serving their wider society, especially in the post-pandemic era. As transcultural communities, they can tap into their cross-cultural sensibilities and compassion for others and actively change the soil of their churches so that they can fulfill the vision of living out the gospel of Christ by participating in the Trinitarian redemptive mission inside and outside their churches. According to the Barna Group’s report published in May 2020, 7 among the surveyed Christians who stopped going to church, 48 percent responded that they needed heartfelt prayers and emotional support; 24 percent responded that they needed genuine Bible-centered encouragement and hope; 27 percent responded that they needed connection and community. In this article, I argue that the Korean-American churches need to (1) be reoriented to small-group-based organic bodies, (2) reconstruct a missionary sense of calling in reaching out to the broader community in an apostolic yet cooperative/dialogical interaction, (3) create room for intergenerational cooperation for the shared vision of God’s kingdom, (4) adopt both online and offline ministries in a hospitable and creative attitude to technological advancement. To that end, I engage with Veli-Matti Kärkkäinen’s communion and public ecclesiologies for a pluralistic society. His ecclesiological proposals contribute to this work via similar and dissimilar approaches. Kärkkäinen advances a communal ecclesiology from below that respects irreducible particularities in unity of the Church and a public ecclesiology through critical mutual interaction with other sectors of society in the imitatio Trinitatis. Kärkkäinen also envisions the Church as the bearer of the gospel of God’s kingdom that embraces creation as a whole based on cosmic Christology and pneumatology. I believe that Kärkkäinen’s proposals may contribute to Korean-American churches by promoting their holistic health and sustainability. The theologian’s proposals may help them enhance their socioeconomic accountability, intercultural mutual respect, and intergenerational unity, while remaining deeply rooted in the gospel of Christ. Kärkkäinen’s communion and public ecclesiology 1. The Father’s creation via the mediation of the Son and the Spirit: Pluriform creation Kärkkäinen advances a model of the Church as a pluriform and multidimensional communion. His ecclesiology honors the imitatio Trinitatis of the Church as a gathering of the particular in unity while advocating the Christian faith’s public relevance and accountability. For Kärkkäinen, a proper understanding of the Triune life of God is demanded for a theologically healthy ecclesiology. 8 The triune life of God is adequately understood when one studies the Trinitarian history of redemption “from below to above.” 9 God’s redemptive action in the Trinitarian mode is not exclusive to the Being of God because the triune God’s Being is at work in the redemptive history of the triune God. To be more specific, in the biblical and theological traditions, there is an inherent connection between “Spirit Christology” and “Logos Christology.” 10 The concept of the Logos, by whom every creature is created, is inseparable from “Spirit Christology” or “Wisdom Christology.” According to Spirit Christology, the cosmic Spirit plays a significant role in the reconciling and healing ministry of the Son that gives genuine freedom to every sinner. The Spirit enables the Son’s birth, ministry, crucifixion, and resurrection. The Incarnation of the Son is a cosmic event, as one can see in the bodily resurrection of Christ and its universal significance. The Son becomes the locus where all creatures find their finitude in God and eschatological telos for mutual flourishing. That is, whereas the Logos incarnate in creation is the principle of creation, in the Spirit, all creatures in their distinction and interdependence are formed and transformed according to their divinely-granted purposes, which will be fulfilled in the eschaton. 11 The Spirit is the breath of life and the creative ground of existence for creatures’ life-affirming solidarity and mutual recognition. In so doing, the triune God creates diversity and plurality in creation. 12 This Trinitarian framework of creation is fulfilled in the eschatological “boundless temporality of the Trinitarian God.” 13 In that reality, there will be distinctions among finite creatures in time but without separation, while being in harmony through redeemed relations in the interdependence of love and freedom. 2. Undissolved particularities in unity: The Church in the imago Trinitatis This Trinitarian understanding of creation grounds Kärkkäinen’s ecclesiology. The Church is the gathering of particular individuals in unity “instituted by the Son and constituted by the Holy Spirit.” 14 The Church in the image of the Trinitarian God is to be inherently communal by embracing particularities without dissolving their unity in “relationality,” “presence-for-the-other,” “equality,” and “non-dominion.” 15 The Church is the locus of the eschatological Spirit’s ecstatic work through which all individuals and communities are invited to embody the loving community of God’s people in fresh ways while rooted in the gospel. 16 In that vein, as the body of Christ, the “catholicity” and the “holiness” of the Church embrace the principle of a self-transcending love by forgiving each other and embracing particularities. 17 The Church is to live out mutual respect for and forgiveness of others in reconciliation by centering on God’s all-inviting grace revealed on the cross and in the resurrection of Christ. Through the Son and the Spirit, the Father continues to create a genuine unity among individuals and communities within the Church. They are sent into the world where the Spirit is holistically at work towards the eschatological consummation. For Kärkkäinen, the eschatological and cosmic Spirit invites all particular members of human society and creation into loving, caring, and peaceable community that enables mutual flourishing. 18 3. Christian faith in the public sphere: Among sciences and other faiths Kärkkäinen contends that theology provides the disciplines of natural and social sciences with a venue for dialogue. Theology focuses on the metaphysical and philosophical questions for “all that is” in Jesus Christ, the cosmic Redeemer. Natural and social sciences focus on the lower levels of the epistemological hierarchy related to their subject matters. For a fuller understanding of reality, they need to interact mutually and critically with each other. 19 Kärkkäinen suggests that developing a plausibility structure over time through the interplay of faith and testing is not exclusive to religious belief systems. In natural sciences, the design of a hypothetical model is theory-laden and paradigm-laden, as existing theories and paradigms, though fallible, are not only informed by but also influence a scientist’s observation and interpretation. 20 In constructing models and hypotheses based on their continuing investigation of the subject matter, scientists reform their research programs. Likewise, a theological investigation of reality is based on historical data and communal discernments. Christians are to be humble in the face of the truth claims outside the Church in seeking the Truth of God. 21 While sciences explore the operation and mechanism of their subject matters, a Christian theology communally explores and discerns the meaning and the purpose of observed and experienced things through the lens of God’s revelation according to Scripture. By doing so, Christians are open-minded to the recent research in sciences and technological developments while remaining anchored to the gospel of Jesus Christ. Theology seeks both “internal coherence” and “external coherence,” 22 while offering “creative suggestions” to sciences “in the form of questions, topics, and conceptions of nature that scientists might find helpful in their research and as judged by their own professional criteria.” 23 God’s revelation occurs via creaturely media in the surrounding world and is experienced via personal and communal interpretations. Likewise, in dialogue with other denominations of the Christian faith and faith traditions, Kärkkäinen maintains that Christians should engage with the believers of other faiths in hospitality. 24 Understanding the history of the world as God’s trinitarian narrative(s) of creation, all other religious faiths, especially monotheistic faiths, may reflect truths of reality about God, the Creator, and share commonalities with the Christian faith. 25 All humanity has “the common origin” and is on the journey toward the shared destiny “in one God” as all that exists in God’s providence. 26 Therefore, Christians’ proclamation of the gospel is not exclusive to dialogue with another faith tradition, as long as it reflects an “intersubjective” process of shaping and reshaping a community’s experiences and understandings of reality. 27 Christians can cooperate in fulfilling the same vision or goal, such as “common service [for the good], healing, and reconciliation,” when there are overlapping areas of concern. 28 While Christ or the Logos constitutes the principle of creation, the Spirit is the life-giving matrix of God’s creation. All creatures are in the Spirit’s providential care. 29 This continuous divine creative work is characterized by transcendent immanence as the Spirit is inherently an eschatological power at work in all corners of creation. 30 4. The Church in solidarity with the suffering For Kärkkäinen, “the Trinitarian confession of faith, and, derivatively, the Trinitarian basis of the church, is never an abstract, format statement, but rather a particular and specific narrative about the creative, redemptive, and perfecting work of Father, Son, and Spirit in the world.” 31 Therefore, while the communal life of the Church is grounded in the Trinity, Christians are called into the concrete mission that embraces “common witness, healing and restoration, social justice, integrity and flourishing of nature, reconciliation and peace building, and dialogue and interfaith engagement.” 32 By focusing on the cosmic Christology and pneumatology of the New Testament (NT), the Church anticipates and participates in the eschatological fulfillment of “personal transformation and freedom from sin,” “liberation from social and political oppression,” and “liberation from marginalization.” 33 In that sense, both “primitively- and eschatologically-oriented ecclesiology” are important. 34 In the liberating power of the Spirit, the Church co-suffers with the poor and the oppressed, following in the footsteps of Jesus the Son who bears the cross and gives his life for others. All in all, for Kärkkäinen, the Church is invited to genuine fellowship among the particular members in unity without dissolving their distinctiveness. The Trinitarian basis for the life of the Church makes possible individual members’ equal and mutual service for each other and ecumenical dialog among diverse traditions of the Christian faith. Furthermore, the Spirit as the cosmic field of creation enables Christians to conceive of cooperation with the other sectors of the broader society in solidarity with the poor and the marginalized. The importance of ecclesiological notions for Korean-American Churches in the post-pandemic era I believe that Kärkkäinen’s ecclesiological suggestions provide insights for Korean-American Christians’ ecclesial life in the broader society. In the early 1900s, missionaries from diverse nations and denominations sought a way to spread the gospel in Korea more effectively and forged together a “comity” agreement, according to which liturgical regulations and management of local churches were applied identically within the region under the direction by a certain denomination. 35 The uniting force of the agreement substantively contributed to imbuing the liberating spirit of the gospel in the early Korean Christians as they protested against Japanese imperialistic rule. However, as Korean churches grew and matured, the fragmenting of theological unity and denominational conflicts continued to emerge in Korean churches. 36 Likewise, when Korean immigration to the US began in the 1960s, the Korean-American denominations began to vigorously divide into many different sects or subdivisions rather than seeking reconciliation in a mutual dialogue. 37 The Covid-19 pandemic revealed the significance of the genuine communal life of the Church that honors both particularity and unity, as well as social hospitality. Not being able to gather to attend worship services at church buildings, Christians focused on the Church’s real identity as communion instituted by Christ and constituted through the Spirit. Such a communion respects the tradition inherited through the former generations of believers while respecting the particular experience of the new generations and diverse individuals’ charismata. Furthermore, the hospitable cooperation of Christians and other members of society became significant as interconnectivity became necessary to effectively overcome the pandemic’s impacts. Korean-American churches are essentially transcultural since they comprise differing generations of different languages, cultures, and ages. The churches’ lives are inherently transcultural. First-generation immigrant members are faced with new horizons in which to critically engage and integrate their lives while expressing their self-identity in new contexts such as the diaspora churches modeled in the NT. The Korean-American churches have significant accountability, especially in the post-pandemic era. The transcultural congregations of the Korean-American churches could have an effective role in fulfilling the missional roles inside and outside their churches. Yet the reality is that young generations are leaving their mother immigrant churches due to the lack of genuine heart-to-heart communication about their cultural differences, the generational gap, and their real-life issues. In addition, many youth and young adults feel that the Korean-American churches need to engage with the other members of the wider community more actively and respond to their needs and concerns. In the post-pandemic era, many are not returning to their churches for various reasons, such as real-life issues (i.e., physical illness and economic breakdown) resulting from the pandemic, the lack of genuine communion within their churches, and the lack of deeply engaging worship services and heart-to-heart fellowship with other members. Therefore, the Korean-American churches need to orient their lives toward the missional identity in the image of the Trinity. This task can be done by (1) fostering a small-group-based community, (2) serving the wider community through mutual cooperation beyond the wall, (3) forging genuine inter-generational fellowship, and (4) staying connected in offline-online hybridity by creatively engaging with technological advancements. 1. Toward a small-group-based “organic” church Kärkkäinen’s ecclesiology from below provides an essential insight for the post-pandemic Korean-American churches. For Kärkkäinen, in the life of the Trinity, particularity and diversity are honored in the eternal, communal life. According to Cornelius Cole, a reputed organic-church movement activist and expert, churches are placed amidst the exponential growth of economic and political bipolarity, the diversification of world views, and the rapid diversification of lifestyles due to fast technological innovations. 38 In the era of these radical changes, we witness a growing need for dynamic and organic gatherings of people with diverse charismata to respond to these various changes. The body of Christ can multiply and function more effectively and with more agility in the face of the fast-changing world than a big organization to which a structural change is challenging. Furthermore, since the post-pandemic era people have become used to meeting in small groups while staying related to the community online. The benefits of these small gatherings are that participants can have more intimate fellowship with other members. More dynamic and organic interaction can occur in small groups than in large meetings as they enable attention and focus to be brought to each other’s life stories. The composition of a small group may be categorized based on the member’s age, gender, and socio-economic-cultural background. In their shared experience of the joys and sufferings of life, participants can engage more compassionately with each other in prayer and practical help for each other. For Jürgen Moltmann, these gatherings are the living fellowship of the grassroots who stand in solidarity with fellow members and hold each other up in the liberating power of the Spirit. 39 Small-group gatherings may be heterogenous, embracing different generations and socio-economic-cultural backgrounds. The different compositions depend on the purpose and context of meetings. Group members from divergent backgrounds can gather in polyphonic unity without downplaying the distinctive characteristics of each member. What matters most in these groups is to make every effort to attend to the stories of “the others” in sympathy and honor the guiding presence of the Spirit in their lives. In so doing, members of the gathering grow together in faith. Miroslav Volf’s understanding of a person’s identity as a porous self is helpful here. For Volf, a person’s identity is “not self-enclosed,” but “the boundaries of the self are porous and shifting.” 40 In the life of the Trinity, the Father, the Son, and the Spirit are distinct selves while never being separate from each other but united in self-giving love and freedom. As Kärkkäinen points out, this notion describes not only the inner life of the Trinity but also the trinitarian economy in creation. I find a good example of this model in Houston Seoul Church in Houston, Texas. 41 The church is pursuing the small-group-based ecclesial model in the NT that embraces a diversity of age, career, and life experience in small groups. The church’s small groups welcome non-believers into their table fellowship, regardless of their background. They compassionately listen to the newcomers’ life stories, help them when they need assistance, and pray for them as they would their family members. This hospitable table fellowship is where newcomers experience the love of Jesus and are invited to accept the gospel of Jesus through table fellowship and mutual caring. From this cradle of their faith, they begin their journey of discipleship. 2. Staying connected to community and serving others Korean-American churches need to be connected to their wider community. As Kärkkäinen maintains, the churches must be open-minded in cooperating with the other members of society (i.e., social agencies, the believers of other faith traditions, and other Christian denominations) to embody the common good in coherence with the gospel of Christ. If the Spirit is likened to the life-giving and redeeming field for all creatures, as Karl Barth rightly puts it, in Christ is the one Light of God’s Truth. 42 Some people may regard Kärkkäinen as adopting the postmodern tendency of rejecting the absolute truthfulness of the gospel of Christ. However, I believe that such an accusation is misguided in understanding Kärkkäinen. Rather than relativizing the gospel of Christ and adopting an agnostic pluralism, Kärkkäinen takes a “committed pluralism” by focusing on cosmic Christology and pneumatology in dialoging with other faith traditions, the partners of ecumenical talk within the Christian tradition. 43 Kärkkäinen writes that while the apostolicity of the Church means to be sent out to the world, it should start from the Word of God as revealed in Christ. The Church exists where there is a true proclamation of the gospel and a right administration of the sacraments. Yet Kärkkäinen proposes the identity of the Church as dialogical, communitarian, and relational. 44 Cosmic pneumatology does not make the work of the Spirit arbitrary but rather concrete in different historical and social contexts. Christians can learn to enrich their understanding of the gospel of Christ and the kingdom of God by gleaning helpful insights and wisdom from other traditions of the Christian faith and other religious traditions through the light of God’s self-revelation in Christ. 45 Christians’ understanding of the kingdom of God continues to be transformed in the Spirit who continues to work inside and outside the Church as the cosmic Redeemer. In that vein, I agree with Kärkkäinen’s call for the public responsibility of the Church for the world toward communication and cooperation with the broader community. 46 Christ showed genuine self-transcendence toward God, neighbors, and all other human beings for their Shalom in all dimensions: physical, psychological, social, and spiritual. A Christ-like life demands our co-suffering with those who suffer for their liberation in multiple ways. In that sense, I affirm that Korean-American churches need to be open-minded in cooperating with other members of the broader society to participate in God’s ministry for the poor. The Church is called “to participate in the mission of the Spirit of Christ by standing in compassionate solidarity with the poor, the marginalized, and the oppressed, who bear the name of ‘ochlos.’” 47 For Moltmann, “Hope for God’s kingdom and the experience of poverty among the people of the poor, the sad, and the suffering, go together.” 48 Christians can learn from the other members of the wider society and cooperate with them while upholding the gospel of Christ. Shin-Gwang Presbyterian Church in Seoul, Korea, provides a good example of cooperation with other religious bodies to help the poor and the marginalized in society and promote social justice. 49 The church is located in one of the poorest districts of Seoul city. In the community around the church, many residents struggle with physical and financial challenges. The church has been hosting worship services for those physically challenged neighbors while inviting others from the community. The church also provides a support system where neighbors can have heart-to-heart fellowship and exercise to improve their physical and mental health. Furthermore, the church’s members volunteer to distribute food to neighbors in need of assistance. In doing these ministries, the church’s cooperation with the broader community does not exclude other religious bodies such as Buddhist and Won Buddhist organizations. 3. Intergenerational communion Kärkkäinen contends, citing the recent report of the WCC’s “International Network Forum on Multicultural Ministry,” that in humbleness, the members of different cultural and individual particularities are called to participate in other’s lives. 50 The koinonia here does not facilitate “the absolutization of difference, but its sublation into the solidarity of the different precisely for the sake of justice and liberation.” 51 In Korean-American churches, the members’ intergenerational disparity includes language, cultural, and theological differences. Yet the strengths of diaspora churches are the ability to communicate with others as porous and de-centered selves. 52 Therefore, the Korean-American churches need to unite their senior and younger generations in one heart and one Spirit to embody a loving, forgiving, and reconciling community despite their cultural divergences. This unity is to be inclusive and dialogical, rather than exclusive, and befitting to the uniqueness of each other, if the life of the church is to be modeled on the communal life of the triune Creator. I find a good example of this open-minded intergenerational ministry in Open Door Presbyterian Church in Herndon, Virginia. 53 The church fosters a dialogical intergenerational ministry: “1 Vision Two Households and 1 Family.” More specifically, the church has an English ministry (EM) and a Korean ministry (KM) as both independent and interdependent groups. Even though they currently have separate worship services, financial management, and discipleship training, they unite in the same vision for the kingdom of God. They have joint discipleship training groups for both EM and KM leaders. The pastors of both ministries assemble joint retreats and small groups. Furthermore, both KM and EM members join in cooperative diakonia for the broader community and missional outreach. Those events are not merely one-time events. They involve preparatory processes through which the members of both groups participate in shared small groups and prayer meetings. In so doing, they understand each other better and learn from each other to grow together in Christ. For the joint ministries to be more effective, the church forms sub-committees composed of the elders from their KM and EM counterparts. 4. Offline-online hybridity: A creative engagement with technological advancement Kärkkäinen takes a post-foundationalist approach in understanding the relationship between theology and the secular disciplines, especially natural sciences, and between the Church and the world. While celebrating “creative mutual interaction” with other disciplines, theology does not lose its unique locus as its subject matter of study is the trinitarian history of God. 54 Accordingly, for Christians, engaging with secular disciplines helps to more fully understand their surrounding world as God’s creation. As Wolfhart Pannenberg contends, “If the God of the Bible is the creator of the universe, then it is not possible to understand fully or even appropriately the processes of nature without any reference to that God,” while sciences provide theology with a detailed explanation of how God’s creation operates. 55 Accordingly, theological perspectives of the world as God’s creation can be put in mutual dialogue with natural sciences. Human beings’ technological advancement through the use of scientific findings may also reflect God’s good purpose for creation, even though human beings’ decisions are under the influence of sin. The Spirit is the universal Spirit who is the Spirit of Christ, who is the Creator and Redeemer of creation as a whole (Col 1:15–20). God’s ruach is immanent in all creatures, including the bodily, ecological, economic, and political dimensions. 56 The Spirit is “the power of creation and the wellspring of life” (Job 33:4, 13; Ps 104:29). 57 Accordingly, rather than being defensive against technological advancement to preserve their traditional ways of worship, creative stewardship is in demand. Post-pandemic churches need to make the best out of the online technologies outside the Church to continue to multiply in the new cultural environment. According to Barna Group’s survey in 2021, about 32 percent of responders prefer the mixed use of offline and online media for worship services and meetings. 58 Many people have become accustomed to online settings as the medium of their social life. Likewise, many church members have become familiar with online worship services. Even though the non-traditional worship service settings lack in-person interaction among members, developments in online technology enable new horizons of interpersonal fellowship. Therefore, Korean-American churches need to discover new ways to reconnect with those who have left the church, by utilizing online platforms to create a hospitable space for heart-to-heart and dynamic interaction and mutual growth. Saddleback Church in Lake Forest, California, provides an example. The church has developed both online worship services and online small groups. What stands out in the latter is that the church has been ministering to drug and gaming addicts in healing small groups. Through the online healing groups, many participants found a place where they could discuss deep-seated, troubling concerns with other co-suffering participants and compassionate helpers. 59 The groups are diversified based on the church’s needs (i.e., physical, financial, interpersonal, and spiritual) and the talents of church members. 60 The online small-group healing ministry’s participants are in continuous care so that they can grow as members of the church and progress into discipleship training and ministry training groups. In my view, offline and online ministries can work hand in hand. The audiovisual information through online streaming enables interpersonal/communal interactions among church members. Yet worship experiences and diakonia become more dynamic through the togetherness in a shared physical presence. For example, Eberhard Jüngel claims that God’s revelation is “by its definition an aesthetic event” that encompasses our physical, psychological, and affective experiences. 61 For Hans Urs von Balthasar, the aesthetic act of seeing, hearing, and feeling things using various senses is a freely given grace that precedes the intellectual act of “knowing” and “hearing” in a revelation event. 62 I believe online meetings fulfill these desires, but only partly. Being present physically together and having shared experiences never lose their irreducible value. We should remember that the Holy Communion or the Eucharist originates in Christ’s Last Supper with his disciples and the early Christians’ table communion at their houses (Luke 22:20; 1 Cor 11:25). Conclusion In this article, I engaged with Kärkkäinen’s communion and public ecclesiology in making creative suggestions for the Korean-American churches in the post-pandemic era. The missional being and the act of God revealed in the trinitarian economy of salvation provide an ontological ground for the communal and missional life of the Church as it continues to embody Christ’s gospel of the kingdom of God inside the Church and in the public spheres of wider society. I affirm that, especially in the post-pandemic era, Christians are invited to be more open to the narratives of others and enhance their socioeconomic accountability, intercultural mutual respect, and intergenerational unity, while being deeply rooted in the gospel of Christ. Author biography JongSeock James Shin is Assistant Director of Academics at America Evangelical University, Gardena, CA, where he teaches systematic theology. His PhD focuses on trinitarian panentheism and theology-science dialogue. He is author of a chapter in the Handbook of Suffering and the Problem of Evil (T&T Clark, 2023). 1. Barna Group is a Christian polling firm located in Ventura, California. 2. See the report at https://www.barna.com/research/resilient-disciples/. 3. See the report at https://www.barna.com/research/six-reasons-young-christians-leave-church/. 4. See the article at https://www.christiantoday.co.kr/news/325438 [in Korean] 5. For more detailed discussion of this trend, refer to https://www.religionwatch.com/silent-exodus-of-second-generation-korean-americans-accelerates/. 6. I am indebted to James H. Lee, “Lost in Translation: Silent Exodus and the Korean American Church,” Asian American Theological Forum (2014). See the article at https://aatfweb.org/2014/10/31/lost-in-translation-silent-exodus-and-the-korean-american-church/. 7. See the report at https://www.barna.com/research/new-sunday-morning-part-2/. 8. Veli-Matti Kärkkäinen, Hope and Community: A Constructive Christian Theology for the Pluralistic World, vol. 5 (Grand Rapids: Eerdmans, 2017), 317–35. 9. Veli-Matti Kärkkäinen, Christ and Reconciliation: A Constructive Christian Theology for the Pluralistic World, vol. 1 (Grand Rapids: Eerdmans, 2013), 37–51. Also, see his Trinity and Revelation: A Constructive Christian Theology for the Pluralistic World, vol. 2 (Grand Rapids: Eerdmans, 2014), 291–92. 10. Ibid., 204–208. 11. Veli-Matti Kärkkäinen, Creation and Humanity: A Constructive Christian Theology for the Pluralistic World, vol. 3 (Grand Rapids: Eerdmans, 2015), 51–81, 102. 12. Ibid., 51–53. 13. Kärkkäinen, Hope and Community, 95. 14. Ibid., 280. 15. Ibid. 16. Ibid., 312, 334–35. 17. Ibid., 310–14. 18. Ibid., 224–29. 19. Kärkkäinen, Creation and Humanity, 25–31. 20. Kärkkäinen, Creation and Humanity, 4. 21. Ibid., 5–6. 22. Ibid., 29. 23. Ibid., 28. Citing Robert Russell, Cosmology (Minneapolis: Fortress, 2008), 4–24. 24. Kärkkäinen, Hope and Community, 443–55. 25. Ibid., 444. 26. Ibid., 446. 27. Ibid., 445,453. 28. Ibid., 344–58, 453. 29. Kärkkäinen, Creation and Humanity, 57–64. 30. Ibid.,77–82. 31. Kärkkäinen, Hope and Community, 279. 32. Kärkkäinen, Creation and Humanity, 345. 33. Kärkkäinen, Hope and Community, 344. 34. Ibid., 234–35. 35. Insoo Kim, The History of the Korean Protestant Churches (Seoul: Kumran, 2012), vol. 1, 334–36 [my translation]. 36. Kim, History of the Korean Protestant Churches, vol. 2, 342–45. 37. Ibid. The growing denominational separations were also partly attributed to the conventional territorialism in Korean culture. The theological and political differences added force to the denominational separations in the next generations of the Korean churches. Cited in James H. Lee, “Glocality and Covenant: Korean American Interchurch Unity,” Asian American Theological Forum (2015), https://aatfweb.org/2015/02/03/lost-in-translation-silent-exodus-and-the-korean-american-church-part-ii/. 38. Cornelius Cole, Rising Tides: Finding a Future-Proof Faith in an Age of Exponential Change (self-pub., 2018). 39. Jürgen Moltmann, The Spirit of Life: A Universal Affirmation, trans. Margaret Kohl (Minneapolis: Fortress Press, 1992), 240–48. 40. Miroslav Volf, “Trinity is Our Social Programme,” in The Doctrine of God and Theological Ethics, ed. Alan Torrance and Michael Banner (London: T&T Clark, 2006), 112. 41. Refer to the church’s website: http://www.seoulbaptist.org [in Korean]. 42. Karl Barth, The Theology of Calvin, trans. Geoffrey Bromiley (Grand Rapids: Eerdmans, 1995), 164. 43. Kärkkäinen dubs a “committed pluralism” as opposed to an “agnostic pluralism” in evaluating Newbigin’s ecclesiology. I think that Kärkkäinen takes a similar stance on the Church’s missional roles in the public sphere as discussed in this subsection. Veli-Matti Kärkkäinen, “The Church in the Post-Christian Society between Modernity and Late-Modernity,” in Theology in Missionary Perspective, ed. Mark Laing and Paul Weston (Eugene, OR: Pickwick, 2012), 136. 44. Kärkkäinen, Hope and Community, 302–306; his Creation and Humanity, 271–74. 45. Kärkkäinen, Hope and Community, 314–15, 451–55. 46. Ibid., 318–20. 47. JongSeock Shin, “The Church as a Messianic Fellowship in Jürgen Moltmann’s and Wolfhart Pannenberg’s Public Ecclesiology,” Evangelical Review of Theology and Politics 7 (2019): 27. Citing Jürgen Moltmann, The Way of Jesus Christ, trans. Margaret Kohl (Minneapolis: Fortress Press, 1993), 124. For Moltmann, “ochlos” is not limited to the economically poor but includes the politically marginalized, the physically sick, the psychologically and spiritually ill. 48. Moltmann, The Way of Jesus Christ, 126. 49. Refer to a news article about the church’s ministry via ecumenical cooperation at http://www.christianwr.com/news/articleView.html?idxno=27436 [in Korean]. 50. Kärkkäinen, Hope and Community, 335. 51. Ibid. 52. Miroslav Volf, Exclusion and Embrace: A Theological Exploration of Identity, Otherness, and Reconciliation (Nashville, TN: Abingdon Press, 1996), 208. 53. Refer to an interview about the church’s intergenerational ministry at https://sola.network/article/interview-interdependent-church/. 54. Kärkkäinen, Creation and Humanity, 27–29. 55. Wolfhart Pannenberg, Toward a Theology of Nature, ed. Ted Peters (Louisville, KY: Westminster/John Knox Press, 1993), 16. 56. Moltmann, Spirit of Life, 225–28. 57. Ibid., 35. 58. See the report at https://www.barna.com/research/cpw-jethani-deckel/. 59. Sanghoon Lee, The Revolution of Online Ministry (Seoul: Research Center for Church Growth, 2022), 43 [my translation]. 60. Kevin Lee, “Strategies for Online Ministry: Focusing on Saddleback Church,” in The Era of the New Normal, ed. Sanghoon Lee (Seoul: Research Center for Church Growth, 2021), 221 [my translation]. 61. Eberhard Jüngel, “Beauty in the Light of Truth: Theological Observations on the Aesthetic Relation,” in Theological Essays II (Edinburgh: T&T Clark, 1995), 76. 62. Hans Urs Von Balthasar, The Grace of the Lord: A Theological Aesthetics, vol. 1: Seeing the Form (San Francisco: Ignatius Press, 1982), 151.
PMC010xxxxxx/PMC10288206.txt	==== Front Eur Stroke J Eur Stroke J ESO speso European Stroke Journal 2396-9873 2396-9881 SAGE Publications Sage UK: London, England 10.1177/23969873221138046 10.1177_23969873221138046 Erratum Corrigendum to Posterior Reversible Encephalopathy Syndrome (Pres) Secondary To Renal Artery Thrombosis Following Covid-19 Vaccination 23 5 2023 6 2023 23 5 2023 8 2 604604 © European Stroke Organisation 2023 2023 European Stroke Organisation This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. typesetterts1 ==== Body pmcThe following absract has been retracted at the request of the Author, Editor and Publisher: Yeung, J A, Grain R, Koh C, Wang Y Y. Posterior Reversible Encephalopathy Syndrome (Pres) Secondary To Renal Artery Thrombosis Following Covid-19 Vaccination. [Abstract P0913 / 703, pp. 499–500]. In “ESOC Abstracts Supplement 2022.” Eur Stroke J 2022; 7(1S): 3–545. 10.1177/23969873221087559 This abstract has been retracted at the request of the author due to the withdrawal of patient consent.
PMC010xxxxxx/PMC10288209.txt	==== Front Autism Autism AUT spaut Autism 1362-3613 1461-7005 SAGE Publications Sage UK: London, England 37345542 10.1177/13623613231182180 10.1177_13623613231182180 Original Article A longitudinal examination of mental health and marital functioning of mothers and fathers of autistic adolescents during COVID-19 https://orcid.org/0000-0003-3314-9269 Ekas Naomi V 1 Kouros Chrystyna D 2 Rigsby Brock A 3 Madison Sarah 1 Hymel Julianne 1 Filippi Maddy 1 1 Texas Christian University, USA 2 Southern Methodist University, USA 3 Colorado State University, USA Naomi V Ekas, Department of Psychology, Texas Christian University, TCU Box 298920, Fort Worth, TX 76129, USA. Email: naomi.ekas@tcu.edu 22 6 2023 22 6 2023 13623613231182180© The Author(s) 2023 2023 The National Autistic Society, SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Parents, particularly mothers, of autistic children may be especially vulnerable to the negative effects of COVID-19. The current longitudinal study examined changes in psychological distress (anxiety, depression, stress) and marital functioning of mothers and fathers of autistic children across three time points between April and October 2020, and the extent to which pre-COVID factors predicted changes in these outcomes. Participants were 94 mothers and 58 fathers of autistic children drawn from a larger longitudinal study about family relationships and autistic children’s mental health that began prior to the pandemic. Results indicated that mothers reported higher levels of psychological distress compared to fathers in July and October 2020. Although, on average, levels of psychological distress and marital functioning did not significantly change for mothers and fathers, pre-pandemic child functioning and marital satisfaction predicted individual differences in change in marital satisfaction during the pandemic for mothers. Implications of the findings are discussed. Lay abstract Parents of autistic children may be especially vulnerable to the negative effects of COVID-19. The current study examined changes in mental health and marital functioning of mothers and fathers of autistic children across three time points between April and October 2020. The study also explored whether pre-COVID factors could predict outcomes during the pandemic. Participants were 94 mothers and 58 fathers of autistic children drawn from a larger study about family relationships and autistic children’s mental health that began prior to the pandemic. Results indicated that mothers reported higher levels of mental health problems compared to fathers in July and October 2020. Levels of mental health and marital functioning did not change between April and October 2020. Pre-pandemic child functioning and marital satisfaction predicted changes in mother’s ratings of marital satisfaction. The findings have implications for ways to best support families during challenging periods. autism COVID-19 fathers marital functioning mental health mothers Eunice Kennedy Shriver National Institute of Child Health and Human Development https://doi.org/10.13039/100009633 R15 HD094279 edited-statecorrected-proof typesetterts1 ==== Body pmcThe COVID-19 pandemic created unprecedented challenges for families with children in the United States. For example, due to lockdowns, many parents shifted to remote work while also being responsible for managing childcare and serving as their child’s primary educator. Evidence suggests these disruptions to everyday life resulted in increased mental health problems and negatively impacted family relationships (Corbett et al., 2021). Families of children with disabilities, including autism, may be especially vulnerable to the negative impacts of COVID-19 as many of these children rely on daily routines and lost access to critical services (Eshraghi et al., 2020). Building on preliminary studies conducted during the COVID-19 pandemic, the present longitudinal study examined changes in mental health and marital functioning among mothers and fathers of autistic children across the first 7 months of the COVID-19 pandemic. Autism spectrum disorder (ASD) is a neurodevelopmental disorder that affects as many as 1 in 44 children in the United States (Maenner et al., 2021). Compared to parents of non-autistic children, parents of autistic children experience increased levels of stress and depression (e.g. Benson & Karlof, 2009). Parents report feeling more stress as a result of increases in their child’s autism-related behaviors (i.e. social communication difficulties and restricted/repetitive behaviors; Pastor-Cerezuela et al., 2016) and their externalizing behaviors (e.g. aggression, conduct problems, hyperactivity; Lin et al., 2021). Although mothers and fathers are parenting the same child, research suggests potential differences in the impacts on parental mental health. For example, autistic children’s externalizing behaviors predicted stress levels for fathers, whereas dysregulation predicted maternal stress levels (Davis & Carter, 2008). Unfortunately, research including fathers is sparse, and our understanding of how fathers are impacted by raising an autistic child is limited. Research suggests that parents raising an autistic child also may be at greater risk of experiencing marital discord compared to parents of non-autistic children (Sim et al., 2016). Based on a meta-analysis, Sim and colleagues (2016) reported that couples with autistic children reported less relationship satisfaction when compared to couples of children without any disability or even couples raising a child with a different intellectual or developmental disability. Consistent with family systems theory (Cox & Paley, 1997), previous studies found that both parents’ and their child’s psychosocial functioning are associated with the quality of the marital relationship (Langley et al., 2017). For example, Langley and colleagues (2017) found that parental depressive symptoms were significantly associated with less relationship satisfaction among couples raising an autistic child. Furthermore, their marital satisfaction was negatively associated with the behavioral problems of their autistic child. There has been less support linking autistic children’s autism-related behaviors to marital satisfaction (see Sim et al., 2016). Parent mental health and marital functioning during the COVID-19 pandemic The World Health Organization declared COVID-19 a pandemic in March 2020 (Cucinotta & Vanelli, 2020). Families experienced immediate disruptions to their lives as governments mandated lockdowns. In the United States, there was variability in the timing and severity of lockdowns. In the State of Texas, stay-at-home orders were issued on 31 March 2020; however, school closures began on 19 March 2020, and schools shifted to remote learning for the remainder of the school year. Businesses began reopening at the end of April 2020, but most schools did not reopen for in-person instruction until September 2020. Parents were confronted with unprecedented challenges as they attempted to balance their roles as a parent, a worker, a spouse, and perhaps as caretakers of their aging parents. In the United States, parents reported greater stress than non-parents, and their levels of stress at the onset of the pandemic were higher than the 2019 levels (American Psychological Association (APA), 2020). For parents, lines between previously defined roles were blurred as home spaces suddenly served as classrooms and offices. Using ecological momentary assessments, a study conducted in the United States found that parents reported more stress during the week compared to weekends, during working hours, and when they were with their child compared to not (Freisthler et al., 2021). Parents raising children with a developmental disability, such as autism, may be particularly vulnerable to the negative impacts of the COVID-19 pandemic. In addition to the closure of schools, which provide critical special education services to children with disabilities, many specialized service providers were required to terminate in-person services. The disruption to routines may be especially problematic for autistic children as insistence on sameness and routines is a core characteristic of autism. Moreover, autistic individuals often have co-occurring mental and physical health problems that may increase their risk of severe illness if they contract COVID-19, which may exacerbate feelings of stress and anxiety among their caregivers (Eshraghi et al., 2020). Several studies confirm this hypothesis, finding that parents of autistic children reported higher levels of psychological distress compared to parents of non-autistic children during the pandemic (Corbett et al., 2021). In a recent review, Shorey and colleagues (2021) found that parents of children with neurodevelopmental disorders, including autism, were concerned about their child’s behavior and about disruptions to daily routines during the pandemic. Parents in the United States reported feeling stress about these disruptions (Bhat, 2021), and disruptions to services were also associated with poorer well-being for parents in Saudi Arabia (Alhuzimi, 2021). Autistic children’s autism-related behaviors and co-occurring behavior problems also served as sources of psychological distress for parents in France (Miniarikova et al., 2021) and the United States (Manning et al., 2021). The limited studies that examined differences between mothers and fathers of autistic children during the pandemic are mixed. Chinese mothers and fathers did not differ in levels of anxiety, depression, and stress (Wang, 2021), whereas French mothers reported higher levels of anxiety and depression compared to fathers (Miniarikova et al., 2021). To our knowledge, only two published studies examined change in mental health for parents of autistic children since the onset of the pandemic. Corbett and colleagues (2021) assessed parent mental health at the beginning of lockdowns in the United States and again 3 months later and did not find any change in anxiety or stress. Similarly, Toseeb and Asbury (2023) found that levels of parents’ psychological distress remained steady between March and October 2020. The COVID-19 pandemic may also negatively impact romantic relationships. For example, Israeli mothers’ COVID-related fears predicted decreases in marital satisfaction (Reizer et al., 2020). It is possible, however, that the pandemic and accompanying lockdowns were an opportunity for some couples to strengthen or maintain their relationship. In a qualitative study, Goldberg et al. (2020) found that few parents reported declines in their relationship quality, although the amount of intimacy declined. In another study, researchers found decreases in relationship conflict and a slight increase in relationship quality among Spanish adults (Rodríguez-Domínguez et al., 2021). Similarly, a longitudinal study in the United States from December 2019 to April 2020 found no changes in relationship satisfaction (Williamson, 2020), and this pattern was similar for parents and non-parents. Despite the growing body of research examining romantic relationships during COVID-19, we are unaware of any research conducted with parents of autistic children. Given the challenges these parents faced during the pandemic, it is reasonable to hypothesize that the quality of their romantic relationships may also be negatively impacted. The current study The purpose of this study was to examine psychological distress and marital functioning for mothers and fathers of autistic children during the early months of the COVID-19 pandemic. We drew on data collected in the State of Texas, USA in April, July, and October 2020, as well as data collected prior to the pandemic to answer two primary research questions: (1) Did parents’ levels of psychological distress and marital functioning change from April to October 2020? This research question was largely exploratory. Given the uncertainty in the early months of the pandemic, it is plausible that no change would be observed or that levels increased. Conversely, levels of distress may have decreased as parents settled into COVID-related routines. (2) Did pre-pandemic child-level (autism-related behaviors, internalizing and externalizing behavior problems), parent-level (dysphoria), and/or marital-level (marital conflict and satisfaction) vulnerability factors predict the direction and/or rate of change in parents’ outcomes from April to October 2020? We expected that pre-COVID vulnerabilities would predict greater increases in marital conflict and psychological distress, and greater decreases in marital satisfaction. Method Procedure and participants The present research is part of a larger study about family relationships and mental health that started in 2018, before the onset of the COVID-19 pandemic (N = 119 families). To qualify for the larger study, parents must have been married or cohabitating for at least 1 year, living with the target child for at least 50% of the time, and be able to read and speak English. In addition, the child needed to have a community diagnosis of autism (verified with documentation provided by parents), be verbally fluent (i.e. speaking in at least two- to three-word sentences), and have no co-occurring diagnosis of an intellectual disability, bipolar disorder, psychosis, or schizophrenia. The inclusion criteria for children were specified due to the nature of study activities that occurred prior to the onset of the pandemic and that are not the focus of this study. For this study, families were emailed in April, July, and October 2020 to complete an online survey. The number of days between participants’ last pre-COVID assessment and the start of the pandemic ranged from 7 to 511 (M = 199.15, SD = 143.73). Participants were compensated with an electronic gift card to Amazon in increasing amounts due to the additional questions added at each time point (US$15 in April, US$20 in July, and US$25 in October). The study received institutional review board (IRB) approval (IRB# H17-097-KOUC). A total of 94 mothers (Mage = 44.17 years, SD = 6.33) and 58 fathers (Mage = 46.00 years, SD = 6.70) of an autistic child participated after removing three families (two families who separated pre-COVID and, given the focus on comparing mothers and fathers, one family in a same-sex relationship). This resulted in a response rate of 79% and 49% for mothers and fathers, respectively. Most parents were married (90.4%), and the average length of their relationship was 16.58 years (SD = 5.77; range: 1.75–29.50 years). Most parents identified as European American (73.4% mothers, 77.6% fathers); the remaining parents identified as Hispanic/Latina(o) (13.8% mothers, 10.3% fathers), Black/African American (2.1% mothers, 3.4% fathers), Asian/Asian American (5.3% mothers, 3.4% fathers), or selected more than one race (5.3% mothers, 1.7% fathers). Approximately 65% of mothers and 85% of fathers were employed for wages or self-employed at the start of the pandemic. Most families (65.6%) reported a pre-COVID household income above US$80,000. Most of their autistic children were male (78.7%) and were approximately 14 years of age (Mage = 13.98 years, SD = 2.21). Two of the mothers and five of the fathers were stepparents. More than half of the mothers (n = 63) and fathers (n = 30) completed all three waves of data collection during the pandemic. Measures Parent psychological distress Anxiety Mothers and fathers completed the eight-item Emotional Distress/Anxiety Short Form from the National Institutes of Health (NIH) PROMIS toolbox (Pilkonis et al., 2011). Participants were asked to rate how often they felt certain emotions such as nervousness, tension, and fearfulness during the last 7 day on a 5-point Likert-type scale (1 = never to 5 = always). Items were summed and internal consistency was good at all time points (α = 0.93–0.95). Dysphoria Mothers and fathers completed the 10-item dysphoria subscale from the Inventory of Depression and Anxiety Symptoms (IDAS; Watson et al., 2007). Participants rated how much they felt or experienced each item (e.g. had difficulty concentrating, feeling restless) during the previous 2 weeks on a 5-point Likert-type-scale (1 = not at all to 5 = extremely). This scale was completed prior to the COVID-19 pandemic and then at each time point of this study. Items were summed and internal consistency was adequate at all time points (α = 0.89–0.93). Perceived stress The Perceived Stress survey from the NIH PROMIS toolbox (Cella et al., 2010) measures stress over a 7-day period. Participants answered 10 questions on a 5-point Likert-type scale (1 = never to 5 = very often). Sample questions included “How often have you felt stressed or nervous?” and “How often have you been able to control irritations in your life?” Positively worded items were reverse-coded, and items were summed. Internal consistency at all time points was good (α = 0.88–0.90). Marital functioning Marital conflict Mothers and fathers completed the O’Leary Porter Scale (OPS; Porter & O’Leary, 1980) prior to the COVID-19 pandemic and at each time point during the pandemic. The OPS includes nine items that assess the frequency of overt marital conflict that occurred in front of the child during the past 6 months. Example items include “How often do you and your spouse argue over discipline problems in your child’s presence?” and “How often does your spouse complain to you about your personal habits (drinking, nagging, sloppiness, etc.) in front of your child?” Items are rated on a 5-point scale (1 = never to 5 = very often). For surveys administered during the pandemic, the timeframe was modified to “since 13 March, when the United States declared COVID-19 a national emergency,” “since 1 May,” and “since 1 August,” for the April, July, and October surveys, respectively. Items were summed and higher scores reflect higher levels of conflict occurring in front of the child. Internal consistency was good at all time points (α = 0.75–0.89). Marital satisfaction Mothers and fathers provided self-reports of their marital satisfaction on the four-item Couples Satisfaction Index (CSI; Funk & Rogge, 2007) prior to COVID-19 and at each time point of the study. Participants were asked to respond based on their relationship since 13 March 2020 (April 2020 survey), since 1 May 2020 (July 2020 survey), and since 1 August 2020 (October 2020 survey). Items were summed to create a total score, in which higher scores reflect greater marital satisfaction. Internal consistency was good at all time points (α = 0.88–0.95). Children’s pre-COVID functioning Autism-related behaviors The child’s primary caregiver (87.8% mothers) completed the 65-item Social Responsiveness Scale–2 (SRS-2; Constantino et al., 2012) during a pre-COVID study visit. Parents rated their child’s behavior with respect to social awareness, social communication, social cognition, social motivation, and rigid and repetitive mannerisms over the past 6 months on a 3-point Likert-type scale (1 = not true and 3 = almost always true). Items were summed to create two subscales representing children’s level of difficulties with social communication/interaction and rigid/repetitive behavior (i.e. mannerisms); higher scores reflect a higher level of difficulty in that domain. Cronbach’s alpha in the current sample was 0.91 for the social communication subscale and 0.80 for the mannerisms subscale. Children’s adjustment Parents reported on their child’s internalizing (e.g. depression and anxiety symptoms) and externalizing (e.g. conduct problems and hyperactivity) behavioral problems on the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001) at a pre-COVID study visit. Parents rated how true a list of statements was about their child (e.g. argues a lot, destroys his or her own things, worries) over the past 6 months on a 3-point Likert-type scale (0 = not true, 1 = somewhat or sometime true, 2 = very true or often true). Items for each subscale were summed, and higher scores reflect higher levels of internalizing and externalizing behavioral problems. Cronbach’s alpha in the current sample was good for the internalizing (α = 0.88) and externalizing (α = 0.92) scale. Community involvement There was no community involvement in this study. Analysis plan To test our research questions, we conducted a series of hierarchical linear models (HLMs). HLM is appropriate for longitudinal data, in which repeated assessments are nested within person and easily account for missing data. To address Research Question 1, separate HLMs for mothers and fathers and per outcome (psychological distress, marital conflict, marital satisfaction) were conducted using HLM v.8.1 (Raudenbush et al., 2019). First, empty models were run in which there were no predictors. The purpose of these models was to determine whether there was significant variability at Level 1 (i.e. is there within-person variability in the levels of the outcome variable) and Level 2 (i.e. is there between-person variability in the levels of the outcome variable). Next, time (coded 0, 1, 2) was entered as an uncentered Level 1 predictor of the outcome variable Y=b0+b1Time+r In the equation above, the value of b0 is the predicted value of the outcome at Time 0 (i.e. April 2020) and b1 reflects the linear time slope of the outcome variable (i.e. the rate of change every 3 months). Next, to test the second research question, if the slopes in the aforementioned models had significant variability, we proceeded to specify models wherein pre-pandemic child-, parent-, and family-level predictors were entered as grand-mean-centered predictors of the intercept and time slope at Level 2. Each model controlled for the elapsed days between the pre-COVID study visit and the onset of the COVID-19 pandemic. Significant interactions were probed using Preacher and colleague’s (2006) online calculator and plotted at ±1 SD from the mean and at the mean value. A sample equation is Level1:Y=b0+b1Time+e Level2:b0=γ00+γ01ElapsedTime+γ02Predictor(s)+U0 b1=γ10+γ11ElapsedTime+γ11Predictor(s)+U1 The meaning of the Level 1 variables remains unchanged. In the Level 2 model, the intercept and slope are now specified as outcomes. Thus, predictors of the intercept (b0) represent a main effect (i.e. Predictor 1 predicts the outcome variable in April 2020) and predictors of the slope (b1) represent the time × predictor interaction (i.e. Predictor 1 predicts the change in the outcome variable from April to October 2020). Results Attrition and preliminary analyses Analyses were conducted to determine whether attrition was related to demographic or key study variables. First, we examined whether parents who did not complete all time points during the pandemic differed from parents without attrition (Supplemental Table 1). Fathers who completed all time points reported significantly greater marital satisfaction compared to fathers who missed one or more time points, t(56) = 2.10, p = 0.040. Next, we compared parents who were part of the larger study versus those who completed at least one survey during the pandemic (Supplemental Table 2). Mothers who participated in the COVID-19 surveys were older than mothers who did not, t(116) = –2.27, p = 0.025. Fathers with less education were less likely to complete one of the COVID-19 surveys, χ2(5) = 14.63, p = 0.012. Table 1 presents descriptive statistics for all study variables at all time points and Supplemental Table 3 presents correlations between study variables, within and across time, separately for mothers and fathers. Table 1. Descriptive statistics for mothers and fathers. Variable Mothers Fathers Pre-COVID (n = 94) April 2020 (n = 87) July 2020 (n = 76) October 2020 (n = 79) Pre-COVID (n = 93) April 2020 (n = 43) July 2020 (n = 42) October 2020 (n = 42) M (SD) Range M (SD) Range M (SD) Range M (SD) Range M (SD) Range M (SD) Range M (SD) Range M (SD) Range Anxiety (Raw score) – 20.74 (6.88) 8.00–35.00 19.57 (6.49) 8.00–32.00 19.42 (6.73) 8.00–38.00 – 17.67 (6.29) 8.00–30.00 15.93 (6.36) 8.00–31.00 14.69 (6.07) 8.00–30.00 Anxiety (T-score) – 58.46 (8.52) 37.10–74.10 57.37 (7.87) 37.10–78.20 57.23 (8.09) 37.10–78.20 – 54.89 (8.39) 37.10–68.70 52.48 (8.94) 37.10–69.80 50.72 (8.97) 37.10–68.70 Perceived stress (Raw score) – 26.33 (6.33) 12.00–45.00 26.17 (6.64) 10.00–42.00 26.38 (6.75) 10.00–43.00 – 24.79 (6.20) 13.00–42.00 23.76 (7.06) 13.00–44.00 22.64 (6.31) 10.00–36.00 Perceived stress (T-score) – 54.92 (10.97) 28.50–80.40 53.77 (11.04) 22.70–75.90 53.87 (11.01) 22.70–77.40 – 51.14 (9.95) 30.90–75.90 49.75 (11.62) 30.90–78.90 47.95 (11.19) 22.70–68.30 Dysphoria 25.59 (8.11) 12.00–44.00 21.08 (7.15) 10.00–42.00 20.75 (7.33) 10.00–42.00 20.76 (7.06) 10.00–42.00 23.78 (7.39) 12.00–46.00 19.14 (7.38) 10.00–42.00 18.24 (7.73) 10.00–40.00 16.93 (5.34) 10.00–32.00 Marital conflict 19.30 (4.96) 10.00–33.00 17.28 (4.50) 10.00–32.00 17.63 (5.14) 10.00–30.00 17.53 (4.96) 10.00–29.00 19.42 (5.30) 10.00–31.00 16.93 (5.62) 10.00–32.00 16.81 (5.14) 10.00–32.00 16.50 (5.89) 10.00–32.00 Marital satisfaction 14.77 (5.05) 1.00–21.00 13.83 (4.25) 0.00–21.00 13.00 (4.75) 0.00–21.00 13.46 (5.13) 0.00–21.00 15.27 (4.58) 0.00–21.00 14.35 (3.99) 5.00–20.00 14.14 (4.86) 6.00–20.00 15.21 (4.84) 0.00–21.00 To test whether mothers and fathers reported significantly different levels of psychological distress and marital functioning, we conducted a series of paired-samples t-tests. We used the Benjamini–Hochberg False Discovery Rate procedure (Benjamini & Hochberg, 1995) to adjust for multiple tests. As shown in Table 2, in July and October 2020, mothers reported significantly higher levels of anxiety compared to fathers. In October 2020, mothers also reported significantly higher levels of stress and dysphoria compared to fathers. There were no significant differences between mothers and fathers for the marital functioning variables at any time point. Table 2. Results of t-tests comparing mother’s and father’s levels of anxiety, stress, and dysphoria in April, July, and October 2020. April 2020 July 2020 October 2020 t df p Corrected p d t df p Corrected p d t df p Corrected p d Anxiety 3.77 42 0.001 0.005 0.57 3.16 36 0.003 0.009 0.52 3.75 37 0.001 0.005 0.61 Stress 1.45 42 0.154 0.213 0.22 2.17 36 0.037 0.065 0.36 2.83 37 0.008 0.021 0.46 Dysphoria 2.03 42 0.049 0.074 0.31 2.15 36 0.038 0.065 0.35 3.43 37 0.001 0.005 0.56 Conflict 0.18 42 0.862 0.862 0.03 1.08 36 0.289 0.371 0.18 0.73 37 0.469 .528 0.12 Satisfaction 0.25 42 0.807 0.854 0.04 –.95 36 0.346 .415 –0.16 –2.13 37 0.040 0.065 –0.34 Significant values are in bold. Corrected p based on Benjamini–Hochberg False Discovery Rate. To examine the extent to which levels of dysphoria and marital functioning changed from pre-COVID to the current study COVID-19 timeframe (average of April, July, and October 2020 levels), we conducted a series of repeated-measures analysis of covariance tests. We entered the elapsed days between the pre-COVID study visit and the onset of the COVID-19 pandemic as a covariate. We used the Benjamini–Hochberg False Discovery Rate procedure (Benjamini & Hochberg, 1995) to adjust for multiple tests. Mothers’ average levels of dysphoria during the first 7 months of the pandemic were significantly lower compared to their pre-COVID levels, F(1, 92) = 22.03, p < 0.001, corrected p < 0.001, ηp2=0.19 . Fathers’ average levels of dysphoria were also significantly lower during the pandemic, F(1, 56) = 22.37, p < 0.001, corrected p < 0.001, ηp2=0.29 . Mothers’, F(1, 92) = 12.49, p = 0.001, corrected p = 0.002, ηp2=0.12 , and fathers’, F(1, 56) = 11.56, p = 0.001, corrected p = 0.002, ηp2=0.17 , self-reported levels of marital conflict were significantly lower during the pandemic compared to their pre-pandemic levels. Mothers’ self-reported marital satisfaction was significantly lower during the pandemic compared to pre-COVID levels, F(1, 92) = 5.31, p = 0.023, corrected p = 0.023, ηp2=0.06 ; however, for fathers there was no significant difference, F(1, 56) = 0.79, p = 0.378, corrected p = 0.378, ηp2=0.01 . The effect of elapsed days and elapsed days × time was not significant in any of the models. Change in mental health and marital functioning Given the high correlations between parents’ self-reported anxiety, stress, and dysphoria (mother, r range: 0.69–0.81; father, r range: 0.62–0.87; see Supplemental Table 3) and to reduce the number of analyses, we created a composite variable of psychological distress by standardizing each variable and computing the average of the variables. As discussed in the analytic plan, HLMs were conducted. In all the empty models, each dependent variable had significant within-person (Level 1) and between-person (Level 2) variability. Next, time was added to the models; there were no significant systematic linear changes for any of the study variables over time (Tables 3 to 5). Table 3. Results of hierarchical linear models predicting change in maternal (n = 94) and paternal (n = 58) psychological distress. Model Time Child-level Parent-level Marital-level b (SE), p b (SE), p b (SE), p b (SE), p Maternal Intercept –0.05 (0.29), p = 0.873 –0.02 (0.28), p = 0.930 –0.04 (0.23), p = 0.867 –0.05 (0.29), p = 0.876 Time –0.01 (0.14), p = 0.960 –0.02 (0.14), p = 0.898 –0.01 (0.13), p = 0.943 –0.01 (0.14), p = 0.970 Days since pre-COVID study visit – 0.00 (0.00), p = 0.687 0.00 (0.00), p = 0.414 0.00 (0.00), p = 0.698 Child mannerisms – 0.02 (0.07), p = 0.807 – – Child mannerisms × Time – 0.05 (0.04), p = 0.192 – – Child communication – 0.01 (0.02), p = 0.590 – – Child communication × time – –0.02 (0.01), p = 0.109 – – Child internalizing problems – 0.03 (0.04), p = 0.492 – – Child internalizing problems × time – 0.00 (0.02), p = 0.986 – – Child externalizing problems – 0.07 (0.04), p = 0.492 – – Child externalizing problems × time – –0.01 (0.02), p = 0.619 – – Mother dysphoria – – 0.19 (0.03), p < 0.001 – Mother dysphoria × time – – 0.00 (0.02), p = 0.864 – Marital conflict – – – 0.03 (0.07), p = 0.620 Marital conflict × time – – – 0.00 (0.03), p = 0.999 Marital satisfaction – – – 0.00 (0.07), p = 0.973 Marital satisfaction × time – – – –0.04 (0.03), p = 0.280 Paternal Intercept –0.15 (0.37), p = 0.688 –0.07 (0.39), p = 0.853 –0.19 (0.34), p = 0.582 –0.03 (0.34), p = 0.925 Time 0.20 (0.21), p = 0.330 0.13 (0.22), p = 0.570 0.19 (0.21), p = 0.351 0.26 (0.20), p = 0.216 Days since pre-COVID study visit – 0.00 (0.00), p = 0.104 0.00 (0.000), p = 0.230 0.00 (0.00), p = 0.065 Child mannerisms – 0.07 (0.10), p = 0.500 – – Child mannerisms × time – 0.03 (0.05), p = .588 – – Child communication – –0.01 (0.03), p = 0.775 – – Child communication × time – 0.00 (0.02), p = 0.929 – – Paternal Child internalizing problems – –0.01 (0.05), p = 0.903 – – Child internalizing problems × time – 0.04 (0.03), p = 0.185 – – Child externalizing problems – –0.01 (0.06), p = 0.888 – – Child externalizing problems × time – –0.05 (0.03), p = 0.159 – – Father dysphoria – – 0.15 (0.05), p = 0.003 – Father dysphoria × time – – 0.01 (0.03), p=0.681 – Marital conflict – – – 0.09 (0.09), p = 0.296 Marital conflict × time – – – –0.09 (0.05), p = 0.065 Marital satisfaction – – – –0.14 (0.10), p = 0.170 Marital satisfaction × time – – – –0.09 (.006), p = 0.168 Models were tested separately. Bold indicates a significant term. – denotes that the term was not specified in the model. Table 4. Results of hierarchical linear models predicting change in maternal (n = 94) and paternal (n = 58) report of marital conflict. Model Time Child level Parent level Marital level b (SE), p b (SE), p b (SE), p b (SE), p Maternal Intercept 17.44 (.29), p < 0.001 17.46 (.46), p < 0.001 17.45 (.46), p < 0.001 17.45 (.36), p < 0.001 Time 0.10 (.24), p = 0.672 0.12 (0.25), p = 0.640 0.10 (0.24), p = 0.688 0.10 (0.23), p = 0.676 Days since pre-COVID study visit – 0.00 (0.00), p = 0.264 0.01 (0.00), p = 0.074 0.00 (0.00), p = 0.067 Child mannerisms – 0.01 (0.12), p = 0.906 – – Child mannerisms × time – –0.02 (0.06), p = 0.701 – – Child communication – –0.01 (0.04), p = 0.815 – – Child communication × time – –0.01 (0.02), p = 0.557 – – Child internalizing problems – 0.00 (0.07), p = 0.994 – – Child internalizing problems × time – 0.01 (.04), p = .872 – – Child externalizing problems – 0.11 (0.06), p = 0.095 – – Child externalizing problems × time – –0.01 (0.03), p = 0.857 – – Mother dysphoria – – 0.06 (0.06), p = 0.279 – Mother dysphoria × time – – –0.03 (0.03), p = 0.384 – Marital conflict – – – 0.48 (0.08), p < 0.001 Marital conflict × time – – – 0.01 (0.05), p = 0.818 Marital satisfaction – – – –0.12 (0.08), p = 0.150 Marital satisfaction × time – – – –0.10 (0.06), p = 0.070 Paternal Intercept 17.10 (0.74), p < 0.001 – – – Time –0.16 (0.32), p = 0.605 – – – Days since pre-COVID study visit – – – – Child mannerisms – – – – Child mannerisms × time – – – – Child communication – – – – Child communication × time – – – – Paternal Child internalizing problems – – – – Child internalizing problems × time – – – – Child externalizing problems – – – – Child externalizing problems × time – – – – Mother dysphoria – – – – Mother dysphoria × time – – – – Marital conflict – – – – Marital conflict × time – – – – Marital satisfaction – – – – Marital satisfaction × time – – – – Models were tested separately. Bold indicates a significant term. – denotes that the term was not specified in the model. Table 5. Results of hierarchical linear models predicting change in maternal (n = 94) and paternal (n = 58) report of marital satisfaction. Model Time Child-level Parent-level Marital-level b (SE), p b (SE), p b (SE), p b (SE), p Maternal Intercept 13.67 (0.45), p < 0.001 13.62 (0.44), p < 0.001 13.68 (0.45), p < 0.001 13.71 (0.35), p < 0.001 Time –0.26 (0.26), p = 0.318 –0.27 (0.24), p = 0.265 –0.27 (0.26), p = 0.308 –0.26 (0.24), p = 0.299 Days since pre-COVID study visit – 0.00 (0.00), p = 0.451 0.01 (0.00), p = 0.960 0.00 (0.00), p = 0.938 Child mannerisms – –0.05 (0.11), p = 0.674 – – Child mannerisms × time – –0.08 (0.06), p = 0.190 – – Child communication – 0.02 (0.03), p = 0.580 – – Child communication × time – 0.06 (0.02), p = 0.008 – – Child internalizing problems – 0.02 (0.06), p = 0.760 – – Child internalizing problems × time – –0.03 (0.04), p = 0.401 – – Child externalizing problems – –0.17 (0.06), p = 0.006 – – Child externalizing problems × time – 0.09 (0.03), p = 0.007 – – Mother dysphoria – – –0.07 (0.05), p = 0.177 – Mother dysphoria × time – – 0.04 (0.03), p = 0.248 – Marital conflict – – – 0.04 0(.08), p = 0.649 Marital conflict × time – – – 0.03 (0.05), p = 0.534 Marital satisfaction – – – 0.57 (0.08), p < 0.001 Marital satisfaction × time – – – 0.15 (0.06), p = 0.010 Paternal Intercept 14.07 (0.56), p < 0.001 – – – Time 0.23 (0.31), p = 0.464 – – – Days since pre-COVID study visit – – – – Child mannerisms – – – – Child mannerisms × time – – – – Child communication – – – – Child communication × time – – – – Paternal Child internalizing problems – – – – Child internalizing problems × time – – – – Child externalizing problems – – – – Child eternalizing problems × time – – – – Mother dysphoria – – – – Mother dysphoria × time – – – – Marital conflict – – – – Marital conflict × time – – – – Marital satisfaction – – – – Marital satisfaction × time – – – – Models were tested separately. Bold indicates a significant term. – denotes that the term was not specified in the model. For mothers, although the average linear time slope of the outcome variables was non-significant, there was significant individual variability in the slopes of psychological distress and marital satisfaction. Since the variability for marital conflict was marginal (p = 0.06), we proceeded with models testing predictors of change in this variable. For fathers, only psychological distress had significant variability in the time slope. We tested three separate models for each outcome of interest (child-, parent-, and marital-level pre-pandemic predictors). A total of nine models were tested for mothers and three models for fathers. There were no significant Level 2 predictor × time interactions for the outcomes of psychological distress or marital conflict (Tables 3 and 4). However, children’s social communication difficulties and externalizing problems were significant predictors of the linear time slope in marital satisfaction from April to October 2020 (Table 5). As shown in Figure 1 (Panel A), marital satisfaction linearly increased over time across all levels of child social communication difficulties. The least amount of change, however, was in mothers of children with below-average levels of social communication difficulties, b = 3.16, SE = 1.27, p = 0.015, followed by children with average levels, b = 4.24, SE = 1.66, p = 0.013. The steepest increase in marital satisfaction was for mothers of children with higher levels of social communication difficulties, b = 5.32, SE = 2.06, p = 0.011. A similar pattern was found for the effect of child externalizing problems on change in mothers’ marital satisfaction (Figure 1, Panel B). There was no significant change in marital satisfaction among mothers of children with average, b = 0.72, SE = 0.45, p = 0.11, and lower, b = –0.11, SE = 0.25, p = 0.68, levels of externalizing problems. However, mothers of children with higher levels of externalizing problems, b = 1.55, SE = 0.72, p = 0.033, reported a significant linear increase in marital satisfaction during the early months of the pandemic. Figure 1. Changes from April to October 2020 in mother’s marital satisfaction as a function of pre-pandemic levels of child social communication difficulties (Panel A) and externalizing behavior problems (Panel B). SRS: levels of child social communication difficulties. In the marital-level model, pre-COVID levels of marital satisfaction predicted linear change in levels of marital satisfaction during the early months of the pandemic (Table 5). Mothers with higher levels of satisfaction pre-COVID reported the greatest increase in marital satisfaction, b = 2.72, SE = 1.16, p = 0.021, followed by mothers with average levels of satisfaction, b = 1.96, SE = 0.88, p = 0.028. Mothers with lower levels of pre-COVID marital satisfaction did not report significant change in the marital satisfaction during the early months of COVID-19, b = 1.21, SE = 0.61, p = 0.050 (Figure 2). There were no significant findings for fathers (Tables 3 to 5). Figure 2. Changes from April to October 2020 in mother’s marital satisfaction as a function of pre-pandemic levels of marital satisfaction. CSI: Couples Satisfaction Index. Discussion The goal of this longitudinal study was to examine the experiences of mothers and fathers of autistic children during the early months of the COVID-19 pandemic. In some cases, the findings of this study support emerging research on parent mental health and marital functioning during the pandemic; however, there were some discrepant findings as well. Consistent with previous research, mothers’ psychological well-being was more adversely impacted during the pandemic compared to fathers; however, both mothers’ and fathers’ levels of dysphoria were lower during the pandemic compared to their pre-pandemic levels. Although, on average, we did not find evidence of systematic change in marital functioning or psychological distress for mothers or fathers, there were significant individual differences. Furthermore, for mothers, pre-pandemic factors differentiated patterns of change in marital satisfaction, but in ways that were counter to the study hypotheses. Consistent with emerging COVID-19 research on parent mental health (e.g. Patrick et al., 2020) and study hypotheses, mothers reported higher levels of anxiety compared to fathers in both July and October 2020. These findings may be explained by the role overload that many mothers experienced during the first months of the pandemic. In addition to engaging in the same amount of childcare (Shafer et al., 2020), many mothers were also tasked with serving as their child’s educator during working hours. In a sample that was primarily female (>90%), Freisthler and colleagues (2021) found that parents reported higher levels of stress during the week and during working hours. When raising an autistic child, the mothers in this study may have stepped in to provide therapy-like services, which added additional responsibilities. Mothers also reported higher levels of stress and dysphoria compared to fathers, but only in October 2020. In the State of Texas, schools had only reopened for in-person learning a few weeks before the October 2020 data collection and had been delayed by lawsuits and protests. It is possible that mothers reported this increase in October because of continued uncertainty surrounding their child’s return to school. There is perhaps one silver lining with respect to parent’s mental health. When compared to their pre-pandemic levels, mothers’ and fathers’ levels of dysphoria were significantly lower, on average, during the pandemic. This is counter to some published research regarding mental health during the pandemic. One possible explanation is that the assessment of pre-pandemic dysphoria did not occur immediately before the onset of the pandemic and levels of dysphoria may have already decreased before the start of the pandemic. For example, a longitudinal study of European adults over the age of 50 found declines in feelings of depression between 2017 and 2020 (Van Winkle et al., 2021). Similarly, Breslau et al. (2021) found considerable individual variability in responses to the pandemic as 52.1% of an adult sample in the United States reported declines in psychological distress from pre-pandemic assessments. Recent reviews have also found associations between exercise and decreased depression during the pandemic (e.g. Luo et al., 2022; Marconcin et al., 2022), and this may provide an explanation for our findings. That is, during the pandemic parents may have regularly taken their children out for walks or to the park to ensure that their child was getting enough activity and to prevent boredom. Indeed, walks around the neighborhood may have been one of the few activities available to these families during the lockdowns. Because the current study sample was primarily from a higher socioeconomic background, they may have had better ability to be involved in exercise near their home (e.g. large yards, workout and sports equipment, safe neighborhoods to walk, neighborhoods with playgrounds), thereby contributing to their lower dysphoria. Of course, this explanation is speculative, and more longitudinal research is needed that charts trajectories of mental health from before the onset of the pandemic through the course of the pandemic, as well as potential mechanisms accounting for these changes. Another possible explanation for the decrease in dysphoria symptoms during the pandemic period is that the daily family leisure experiences of parents of autistic and non-autistic children may have become more similar given lockdown restrictions. Prior to the pandemic, parents of autistic children reported experiencing higher levels of parenting burden (Picardi et al., 2018) and, thus, were less likely to engage in the same family activities enjoyed by parents of non-autistic children. The knowledge that they were unable to enjoy the same activities as parents of non-autistic children may have led to feelings of disappointment and the “fear of missing out (FOMO),” which have been associated with decreased general mood and overall life satisfaction (Przybylski et al., 2013). Since most families were not able to enjoy public social outings during the lockdowns, parents of autistic children may have benefited from feeling as if their family experiences did not differ from other families. This study also examined two aspects of marital functioning (conflict and satisfaction) during the pandemic. Mothers and fathers of autistic children reported similar levels of conflict and satisfaction during the pandemic. However, when compared to their pre-pandemic levels, fathers reported lower levels of conflict, and mothers reported lower levels of satisfaction. Previous research shows that fathers may be less accurate in identifying conflict compared to mothers (Papp et al., 2002), which could explain the decrease in conflict reported by fathers, but not mothers, in this study. Pandemic-era research on attributions about couple conflict has also shown that, during the early stages of the pandemic when our data were collected, people became less likely to attribute conflicts to their partner’s internal characteristics (Williamson, 2020). Taken with Papp and colleagues’ (2002) findings, it is possible that these changes in conflict attributions also led fathers to be less likely than mothers to classify conflict as marital or romantic in nature. Alternatively, mothers may have taken on more of the parenting responsibilities without the couple necessarily discussing these changes. Whereas this shift of responsibilities may have registered as a lack of conflict for fathers, mothers may have felt less appreciated by their partner, resulting in lower marital satisfaction. Another possible explanation for the discrepancy between fathers reporting less conflict but mothers reporting less marital satisfaction may be rooted in differences in division of one’s own time within couples. Vigil et al. (2022) found that both time for oneself and time spent outside of the home during the pandemic were significantly and positively associated with relationship satisfaction. Since mothers likely took on new responsibilities, in addition to maintaining their existing levels of childcare responsibilities (Shafer et al., 2020), it is also possible that they took less time for themselves compared to fathers. Thus, consistent with Vigil et al. (2022), the lack of time for themselves relative to their partner may have negatively impacted mothers’ satisfaction with their spouse. More research investigating relationship dynamics during the pandemic is needed, particularly among vulnerable families. The longitudinal nature of the data allowed us to examine changes during the early months of the pandemic. On average, we did not find evidence of within-person change in levels of the outcomes. While counter to hypotheses, these results are consistent with longitudinal research of parents of autistic children in the United States (Corbett et al., 2021) and the United Kingdom (Toseeb & Asbury, 2023). Although caution is needed when interpreting null findings, there are several possible explanations for this lack of change. Although the State of Texas saw rapid reductions and elimination of mandates during the study period, it is possible that these were not sufficient to improve psychological distress. There was still a considerable lack of knowledge about COVID-19 and vaccines were not available. Some parents may have felt comfortable with the state’s decision to reopen businesses, whereas others experienced distress. Daily life for many families remained relatively consistent during the study period. For example, children remained home until just weeks before the October 2020 survey, and many workplaces continued remote work policies. Together, these could explain the lack of change in outcomes. Mothers of autistic children with the highest levels of social communication difficulties and externalizing problems pre-pandemic reported the greatest improvement in their marital satisfaction. These findings were the opposite of our predictions, as pre-pandemic research generally showed that parents of children with greater developmental difficulties reported worse outcomes (e.g. Lin et al., 2021). At the same time, there has been less support for the effects of child behaviors on marital functioning in pre-pandemic studies (see Sim et al., 2016). Prior work has found that the degree of parental role specialization, such that mothers provide more childcare and fathers provide increased household income, increases as children’s autism-related behaviors increase (Hartley et al., 2014). For families of children with greater autism-related behaviors, fathers who typically provide less childcare and greater family income may have stayed at home and engaged in more childcare than usual during the pandemic. Notably, bidirectional links have been reported between parenting satisfaction and marital satisfaction (Rogers & White, 1998); thus, increased satisfaction with paternal childcare and involvement may have spilled over to increase mothers’ marital satisfaction. It is also possible that children required more help from parents to adjust to the pandemic, and the time spent together at home allowed parents to work together to meet their child’s needs, thereby strengthening their relationship. Consistent with our hypotheses, mothers who reported higher levels of pre-pandemic marital satisfaction also reported the greatest increase in satisfaction during the pandemic. A strong relationship may have shaped how mothers viewed the effects of the pandemic and they may have felt more supported by their spouse. Instead of thinking of the time together as stressful, these mothers may have seen pandemic lockdowns as an opportunity to spend more time together, to engage in more in-depth discussions, or to reflect and make readjustments within their relationship. Research on romantic relationships during the pandemic, particularly among parents of autistic children, is limited, and this study only captured snapshots at three time points during the pandemic. Daily measures could help to shed light on couples’ behaviors, cognitions, and emotions as they relate to the marital relationship. Limitations and future directions The limitations of our study provide directions for future research. First, we used a convenience sample of families who were part of an ongoing longitudinal study. Although this allowed us to incorporate pre-pandemic levels of functioning, the families in this study may not be representative of all families raising an autistic child or those with lower resources and time to participate in a research study. Notably, due to the demands of the original, pre-pandemic study, children with a co-occurring intellectual disability were not included. Thus, our findings only generalize to families raising children with higher levels of intellectual functioning and with verbal skills. Although this study adds to the growing number of studies conducted in a single state, there was considerable variability in how states responded to the pandemic; thus, our findings also may not generalize to families living in other parts of the United States. Nonetheless, the findings of this study will provide critical information for policy within the State of Texas. Furthermore, the sample was predominately White and additional research with more diverse samples is needed. Second, the sample size was small, relative to the larger study and particularly among fathers. Thus, our statistical power for testing trajectories was likely low. Future research with larger samples and with more fathers is needed to fully understand the effects of the pandemic on families. The longitudinal assessments of families during the pandemic ended in October 2020. The effects of the pandemic during the early months may not necessarily reflect long-term effects of the pandemic. Further research, therefore, is needed to understand the extent to which initial effects of the pandemic strengthened and which dissipated over time, and what effect novel coronavirus strains and surges (e.g. delta, omicron), as well as mitigation measures (i.e. vaccines, masking), had on families’ abilities to cope with the stresses of the pandemic. Finally, there are omitted variables, that were not measured in this study, that could also explain changes in parents’ psychological well-being. For example, closure of schools and service providers may have resulted in a loss of services for autistic children and their families. This loss of services could contribute to increased stress and anxiety for parents (Furar et al., 2022). Thus, research is needed that examines the extent to which disruptions to children’s services, and the way the disruption occurred (e.g. loss of service vs service switched from in-person to remote), predict parent psychological well-being. Conclusions and implications Overall, the results of this study suggest that mothers of autistic children were more adversely impacted during the early months of the pandemic compared to fathers. Although this supports emerging pandemic-related research, it begs the question of how society can best support mothers of autistic children during the next wave of this pandemic or future crises that impact schools or workplaces. Initial responses to the pandemic were predominantly focused on transitioning children to remote educational and therapeutic settings. Parents’ emotional and psychological needs, however, were rarely addressed or met (Latzer et al., 2021). Finding ways to intentionally connect parents of autistic children with one another might have been helpful. For example, virtual respite care, in which mothers could have some time alone in the home, could also be facilitated online. Indeed, during the pandemic, organizations began offering virtual respite care where children engaged in activities and played games remotely with a trained respite care provider (e.g. Families as Allies, 2020). Finally, this study also highlights that the pandemic experiences of families may differ depending on pre-pandemic factors. Thus, it is critical to identify families of autistic children who may be especially vulnerable during times of uncertainty. Supplemental Material sj-docx-1-aut-10.1177_13623613231182180 – Supplemental material for A longitudinal examination of mental health and marital functioning of mothers and fathers of autistic adolescents during COVID-19 Click here for additional data file. Supplemental material, sj-docx-1-aut-10.1177_13623613231182180 for A longitudinal examination of mental health and marital functioning of mothers and fathers of autistic adolescents during COVID-19 by Naomi V Ekas, Chrystyna D Kouros, Brock A Rigsby, Sarah Madison, Julianne Hymel and Maddy Filippi in Autism We wish to thank our research laboratory staff for data collection and preparation, as well as the children and parents who participated in this study. We also thank all of the individuals enrolled in SPARK and the SPARK Research Match staff who assisted in the recruitment of participants. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the National Institute of Child and Human Development (R15 HD094279) awarded to PIs Ekas and Kouros. ORCID iD: Naomi V Ekas https://orcid.org/0000-0003-3314-9269 Supplemental material: Supplemental material for this article is available online. ==== Refs References Achenbach T. M. Rescorla L. A. (2001). Manual for the ASEBA school-age forms and profiles. University of Vermont, Research Center for Children, Youth, & Families. Alhuzimi T. (2021). Stress and emotional wellbeing of parents due to change in routine for children with autism spectrum disorder (ASD) at home during COVID-19 pandemic in Saudi Arabia. Research in Developmental Disabilities, 108 , Article 103822. 10.1016/j.ridd.2020.103822 American Psychological Association. (2020). Stress in America 2020 (Vol. 1 ). https://www.apa.org/news/press/releases/stress/2020/stress-in-america-covid.pdf Benjamini Y. Hochberg Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological), 57 , 289–300. 10.1111/j.2517-6161.1995.tb02031.x Benson P. R. Karlof K. L. (2009). Anger, stress proliferation, and depressed mood among parents of children with ASD: A longitudinal replication. Journal of Autism and Developmental Disorders, 39 (2 ), 350–362. 10.1007/s10803-008-0632-0 18709548 Bhat A. (2021). Analysis of the SPARK study COVID-19 parent survey: Early impact of the pandemic on access to services, child/parent mental health, and benefits of online services. Autism Research, 14 (11 ), 2454–2470. 10.1002/aur.2618 34591364 Breslau J. Finucane M. L. Locker A. R. Baird M. D. Roth E. A. Collins R. L. (2021). A longitudinal study of psychological distress in the United States before and during the COVID-19 pandemic. Preventative Medicine, 143 , Article 106362. 10.1016/j.ypmed.2020.106362 Cella D. Riley W. Stone A. Rothrock N. Reeve B. Yount S. Amtmann D. Bode R. Buysse D. Choi S. Cook K. Devellis R. DeWalt D. Fries J. F. Gershon R. Hahn E. A. Lai J.-S. Pilkonis P. Revicki D. , & PROMIS Cooperative Group. (2010). The Patient-Reported Outcomes Measurement Information System (PROMIS) developed and tested its first wave of adult self-reported health outcome item banks: 2005–2008. Journal of Clinical Epidemiology, 63 (11 ), 1179–1194. 10.1016/j.jclinepi.2010.04.011 20685078 Constantino J. N. Davis S. A. Todd R. D. Schindler M. K. Gross M. M. Brophy S. L. Metzger L. M. Shoushtari C. S. Splinter R. Reich W. (2012). Validation of a brief quantitative measure of autistic traits: Comparison of the social responsiveness scale with the autism diagnostic interview-revised. Journal of Autism and Developmental Disorders, 33 , 427–433. 10.1023/A:1025014929212 Corbett B. A. Muscatello R. A. Klemencic M. E. Schwartzman J. M. (2021). The impact of COVID-19 on stress, anxiety, and coping in youth with and without autism and their parents. Autism Research, 14 (7 ), 1496–1511. 10.1002/aur.2521 33913261 Cox M. J. Paley B. (1997). Families as systems. Annual Review of Psychology, 48 (1 ), 243–267. 10.1146/annurev.psych.48.1.243 Cucinotta D. Vanelli M. (2020). WHO declares COVID-19 a pandemic. Acta Biomedica, 91 , 157–160. 10.23750/abm.v91i1.9397 32191675 Davis N. O. Carter A. S. (2008). Parenting stress in mothers and fathers of toddlers with autism spectrum disorders: Associations with child characteristics. Journal of Autism and Developmental Disorders, 38 (7 ), 1278–1291. 10.1007/s10803-007-0512-z 18240012 Eshraghi A. A. Li C. Alessandri M. Messinger D. S. Eshraghi R. S. Mittal R. Armstrong F. D. (2020). COVID-19: Overcoming the challenges faced by individuals with autism and their families. Lancet Psychiatry, 7 (6 ), 481–483. 10.1016/S2215-0366(20)30197-8 Families as Allies. (2020, June 1). “Virtual Respite” offers creative activities for kids, a break for caregivers. https://www.faams.org/virtual-respite-offers-creative-activities-for-kids-a-break-for-caregivers/ Freisthler B. Gruenewald P. J. Tebben E. McCarthy K. S. Wolf J. P. (2021). Understanding at-the-moment stress for parents during COVID-19 stay-at-home restrictions. Social Science & Medicine, 279 , Article 114025. 10.1016/j.socscimed.2021.114025 Funk J. L. Rogge R. D. (2007). Testing the ruler with item response theory: Increasing precision of measurement for relationship satisfaction with the Couples Satisfaction Index. Journal of Family Psychology, 21 , 572–583. 10.1037/0893-3200.21.4.572 18179329 Furar E. Wang F. Durocher J. S. Ahn Y. A. Memis I. Cavalcante L. Klahr L. Samson A. C. Van Herwegen J. Dukes D. Alessandri M. Mittal R. Eshraghi A. A. (2022). The impact of COVID-19 on individuals with ASD in the US: Parent perspectives on social and support concerns. PLOS ONE, 178 (8 ), Article e0270845. 10.1371/journal.pone.0270845 Goldberg A. E. McCormick N. Virginia H. (2020). Parenting in a pandemic: Work-family arrangements, well-being, and intimate relationships among adoptive parents. Family Relations, 70 (1 ), 7–25. 10.1111/fare.12528 Hartley S. L. Mihaila I. Otalora-Fadner H. S. Bussanich P. M. (2014). Division of labor in families of children and adolescents with autism spectrum disorder. Family Relations, 63 (5 ), 627–638. 10.1111/fare.12093 25484479 Langley E. Totsika V. Hastings R. P. (2017). Parental relationship satisfaction in families of children with autism spectrum disorder (ASD): A multilevel analysis. Autism Research, 10 (7 ), 1259–1268. 10.1002/aur.1773 28339168 Latzer I. T. Leitner Y. Karnieli-Miller O. (2021). Core experiences of parents of children with autism during the COVID-19 pandemic lockdown. Autism, 25 (4 ), 1047–1059. 10.1177/1362361320984317 33435701 Lin Y.-N. Iao L.-S. Lee Y.-H. Wu C.-C. (2021). Parenting stress and child behavior problems in young children with autism spectrum disorder: Transactional relations across time. Journal of Autism and Developmental Disorders, 51 (7 ), 2381–2391. 10.1007/s10803-020-04720-z 32965625 Luo Q. Zhang P. Liu Y. Ma X. Jennings G. (2022). Intervention of physical activity for university students with anxiety and depression during the COVID-19 pandemic prevention and control period: A systematic review and meta-analysis. International Journal of Environmental Research and Public Health, 19 (22 ), Article 15338. 10.3390/ijerph192215338 Maenner M. J. Shaw K. A. Bakian A. V. Bilder D. A. Durkin M. S. Esler A. Furnier S. M. Hallas L. Hall-Lande J. Hudson A. Hughes M. M. Patrick M. Pierce K. Poynter J. N. Salinas A. Shenouda J. Vehorn A. Warren Z. Constantino J. N. Cogswell M. E. (2021). Prevalence and characteristics of autism spectrum disorder among children aged 8 years: Autism and Developmental Disabilities Monitoring Network, 11 sites, United States, 2018. MMWR Surveillance Summaries, 70 (11 ), 1–16. 10.15585/mmwr.ss7011a1 Manning J. Billian J. Matson J. Allen C. Soares N. (2021). Perceptions of families of individuals with autism spectrum disorder during the COVID-19 crisis. Journal of Autism and Developmental Disorders, 51 , 2920–2928. 10.1007/s10803-020-04760-5 33090358 Marconcin P. Werneck A. O. Peralta M. Ihle A. Gouveia É. R. Ferrari G. Sarmento H. Marques A. (2022). The association between physical activity and mental health during the first year of the COVID-19 pandemic: A systematic review. BMC Public Health, 22 (1 ), Article 209. 10.1186/s12889-022-12590-6 Miniarikova E. Vernhet C. Peries M. Loubersac J. Picot M. C. Munir K. Baghdadli A. (2021). Anxiety and depression in parents of children with autism spectrum disorder during the first COVID-19 lockdown: Report from the ELENA cohort. Journal of Psychiatric Research, 149 , 344–351. 10.1016/j.jpsychires.2021.11.022 34776249 Papp L. M. Cummings E. M. Goeke-Morey M. C. (2002). Marital conflicts in the home when children are present versus absent. Developmental Psychology, 38 (5 ), 774–783. 10.1037/0012-1649.38.5.774 12220054 Pastor-Cerezuela G. Fernández-Andrés M. I. Tárraga-Mínguez R. Navarro-Peña J. M. (2016). Parental stress and ASD: Relationship with autism symptom severity, IQ, and resilience. Focus on Autism and Other Developmental Disabilities, 31 (4 ), 300–311. 10.1177/1088357615583471 Patrick S. W. Henkhaus L. E. Zickafoose J. S. Lovell K. Halvorson A. Loch S. Letterie M. Davis M. M. (2020). Well-being of parents and children during the COVID-19 pandemic: A national survey. Pediatrics, 146 (4 ), Article e016824. 10.1542/peds.2020-016824 Picardi A. Gigantesco A. Tarolla E. Stoppioni V. Cerbo R. Cremonte M. Alessandri G. Lega I. Nardocci F. (2018). Parental burden and its correlates in families of children with autism spectrum disorder: A multicentre study with two comparison groups. Clinical Practice and Epidemiology Mental Health: CP & EMH, 14 , 143–176. 10.2174/1745017901814010143 Pilkonis P. A. Choi S. W. Reise S. P. Stover A. M. Riley W. T. Cella D. , & PROMIS Cooperative Group. (2011). Item banks for measuring emotional distress from the Patient-Reported Outcomes Measurement Information System (PROMIS®): Dysphoria, anxiety, and anger. Assessment, 183 (3 ), 263–283. 10.1177/1073191111411667 Porter B. O’Leary K. D. (1980). Marital discord and childhood behavioral problems. Journal of Abnormal Child Psychology, 8 (3 ), 287–295. 10.1007/BF00916376 7410730 Preacher K. J. Curran P. J. Bauer D. J. (2006). Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31 (4 ), 437–448. 10.3102/10769986031004437 Przybylski A. K. Murayama K. DeHaan C. R. Gladwell V. (2013). Motivational, emotional, and behavioral correlates of fear of missing out. Computers in Human Behavior, 29 (4 ), 1841–1848. 10.1016/j.chb.2013.02.014 Raudenbush S. W. Bryk A. S. Cheong Y. F. Congdon R. (2019). HLM 8 for Windows (Version 8) [Computer software]. Scientific Software International. Reizer A. Koslowsky M. Geffen L. (2020). Living in fear: The relationship between fear of COVID-19, distress, health, and marital satisfaction among Israeli women. Health Care for Women International, 41 (11–12 ), 1273–1293. 10.1080/07399332.2020.1829626 33084536 Rodríguez-Domínguez C. Carrascal-Caputto B. Durán M. (2021). Anxiety and intimate relationships in times of lockdown due to COVID-19. Psychological Trauma: Theory, Research, Practice, and Policy, 14 (2 ), 237–246. 10.1037/tra0001094 34472945 Rogers S. J. White L. K. (1998). Satisfaction with parenting: The role of marital happiness, family structure, and parents’ gender. Journal of Marriage and Family, 602 (2 ), 293–308. 10.2307/353849 Shafer K. Scheibling C. Milkie M. A. (2020). The division of domestic labor before and during the COVID-19 pandemic in Canada: Stagnation versus shifts in fathers’ contributions. Canadian Review of Sociology, 57 (4 ), 523–549. 10.1111/cars.12315 33151625 Shorey S. Lau L. Tan J. X. Ng E. D. Aishworiya R. (2021). Families with children with neurodevelopmental disorders during COVID-19: A scoping review. Journal of Pediatric Psychology, 46 (5 ), 514–525. 10.1093/jpepsy/jsab029 33758930 Sim A. Cordier R. Vaz S. Falkmer T. (2016). Relationship satisfaction in couples raising a child with autism spectrum disorder: A systematic review of the literature. Research in Autism Spectrum Disorders, 31 , 30–52. 10.1016/j.rasd.2016.07.004 Toseeb U. Asbury K. (2023). A longitudinal study of the mental health of autistic children and adolescents and their parents during COVID-19: Part 1—Quantitative findings. Autism, 27 (1 ), 105–116. 10.1177/13623613221082715 35669991 Van Winkle Z. Ferragina E. Recchi E. (2021). The unexpected decline in feelings of depression among adults ages 50 and older in 11 European countries amid the COVID-19 pandemic. Socius: Sociological Research for a Dynamic World, 7 , 1–11. 10.1177/2F23780231211032741 Vigil J. Strauss H. Talamini F. Zentner M. (2022). Relationship satisfaction in the early stages of the COVID-19 pandemic: A cross-national examination of situational, dispositional, and relationship factors. PLOS ONE, 17 (3 ), Article e0264511. 10.1371/journal.pone.0264511 Wang C. (2021). Mental health and social support of caregivers of children and adolescents with ASD and other developmental disorders during COVID-19 pandemic. Journal of Affective Disorders Reports. 6 , Article 100242. 10.1016/j.jadr.2021.100242 Watson D. O’Hara M. W. Simms L. J. Kotov R. Chmielewski M. McDade-Montez E. A. Gamez W. Stuart S. (2007). Development and validation of the Inventory of Dysphoria and Anxiety Symptoms (IDAS). Psychological Assessment, 19 (3 ), 253–268. 10.1037/1040-3590.19.3.253 17845118 Williamson H. C. (2020). Early effects of the COVID-19 pandemic on relationship satisfaction attributions. Psychological Science, 3 (12 ), 1479–1487. 10.1177/0956797620972688
PMC010xxxxxx/PMC10290924.txt	==== Front Glob Soc Policy Glob Soc Policy GSP spgsp Global Social Policy 1468-0181 1741-2803 SAGE Publications Sage UK: London, England 10.1177/14680181231180532 10.1177_14680181231180532 GSP Forum Costa Rica’s social policy response to the pandemic: From potential for universalism to austerity-as-usual Voorend Koen Abarca Daniel Alvarado University of Costa Rica, Costa Rica Koen Voorend, Instituto de Investigaciones Sociales, Universidad de Costa Rica, Apartado Postal 11501-2060, San José, Costa Rica. Email: koen.voorend@ucr.ac.cr 16 6 2023 16 6 2023 14680181231180532© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. edited-statecorrected-proof typesetterts1 ==== Body pmcCosta Rica is internationally recognized for its long-standing and strong universal social policy regime. It performs extremely well in output welfare indicators, especially compared with other countries in Latin America. For example, Costa Rica has the second highest life expectancy in Latin America and the Caribbean (80.6 years), at par with countries in Northern Europe, it has a low infant mortality rate of 9 per 1000 births and general high levels of social security coverage (around 90%) (National Institute of Statistics and Censuses (INEC), 2023; World Bank, 2022). Notably, it achieved this relatively early in its history, by the 1940s. However, this social policy regime has come under increasing pressure since the 1980s, when austerity measures followed the 1980s debt crisis (Martínez-Franzoni and Sánchez-Ancochea, 2013). Recently in 2018, new austerity measures were proposed by government authorities (Presidencia de la República de Costa Rica, 2018), and there had been a slow-but-steady erosion of the country’s social protection system (Martínez-Franzoni and Sánchez-Ancochea, 2023). Initially, we argued that the pandemic represented an opportunity to strengthen Costa Rica’s commitment to universalism in social protection (Voorend and Alvarado, 2021). In reaction to COVID, during 2020, the country quickly took several measures in different social policy sectors: health care, pensions, and social assistance. Now, mid-2023, the question we explore is whether this commitment to universalism has progressed or whether it is business-as-usual, that is, back to austerity and more pressure on Costa Rica’s once-so-impressive social policy regime? The COVID-19 social policy legacy The COVID-19 pandemic implied a contraction of Costa Rica’s gross domestic product (GDP) by −4.1%, while unemployment increased (24% in 2020, now lower at 11.6% in 2023), as did poverty (around 23% since 2020). Likewise, informality grew to 44% of the working population (INEC, 2023; Programa del Estado de la Nación (PEN), 2022). In response to these implications, the Costa Rican State took several social policy measures (Voorend and Alvarado, 2021). However, the social policy investments made during the pandemic have not meant structural change of Costa Rica’s social policy regime. The additional healthcare infrastructure created to meet the peaks of COVID-19 infections was temporary and did not result in an increase of hospital capacity. Today, COVID cases are treated in small units within the pre-existing hospitals (Elpais, 2022). During the pandemic, health coverage was temporarily extended to any person with COVID, irrespective of insurance status, nationality, or migratory status (Arce, 2020). This universal measure still applies to date and has been extended to the vaccination campaigns (Gómez, 2021), but it is an explicit measure for COVID only and does not encompass other health treatments. Another important policy applied in 2020 was to reduce the minimum contribution base by 75% for employers and by 25% for independent workers (Presidencia de la República de Costa Rica, 2020), to ensure that public health insurance coverage would not drop in the face of the pandemic. After December 2020, however, this measure was discontinued. Concerning pensions, measures made it possible to opt for early pension payment advances and to withdraw special savings funds. Such facilities granted during the pandemic were aimed at maintaining the existing coverage and the services provided to members, rather than at growth of universal coverage (Voorend and Alvarado, 2021). All measures were discontinued, and currently, there are sustained public concerns over the financial strength of the pensions system, as well as possible imminent bankruptcy. Finally, the star program during the pandemic in Costa Rica was a special conditional cash transfer program called Plan Proteger (Protection Plan), conceived as a temporary basic income scheme for people who had lost all or part of their employment and source of income (Ministry of Labor and Social Security (MTSS), 2020). The program ran for 6 months during 2020 with an investment of approximately 0.9% of GDP and was generally well received and evaluated (Ministry of Planification and Economic Policy (MIDEPLAN), 2022). However, plans for its continuation after 2020 were not approved by Congress due to scandals over resource management (Arrieta, 2020). Much of the discussion around the decision to discontinue this program centered around austerity in the face of the public deficit and public debt, perceived to be unsustainable (Martínez-Franzoni and Sánchez-Ancochea, 2023). Back to austerity-as-usual? While the social policy measures during the pandemic were important to deal with the adverse effects; in 2023, it seems none resulted in sustained and structural improvements to advance any dimension of universalism in Costa Rica neither in terms of coverage, quality of services, nor their financing and payment. Instead, under the banner of austerity, the country seems to be re-taking a worrisome turn to austerity, which compromises its commitment to universalism. Austerity measures had previously been implemented by right-of-center governments (Alvarado’s presidency, 2018–2022) and are currently pursued by the right-wing populist government of Rodrigo Chaves. Costa Rica’s fiscal deficit was 8.03% in 2020, 5.18% in 2021, and in part following austerity measures that cut public costs; in 2022, it dropped to 1.9% (EFE, 2022; Murillo, 2021). However, no fiscal adjustments were passed to improve tax recollection. A case in point is the 2021 agreement between the Costa Rican government and the International Monetary Fund (IMF) which allows the country to access US $1.775 million in financing in exchange for structural and legislative reforms aimed at lowering the country’s public deficit and debt (IMF, 2021). This agreement is often cited in public debate as a justification for further austerity measures and has had a direct impact on social spending. Where Costa Rica was historically among the countries in Latin America with high public social spending, it has lost ground over the last decade. Initially, during the pandemic, public social spending increased from 11.75% of GDP in 2019 to 12.3% in 2020, but it has since declined to a historic low of 9.8% in 2023 (Bermúdez, 2022). This decline has affected policy areas that were initially addressed during the pandemic, as the year 2023 represented the lowest investment in health over the last decade, with only 0.7% of GDP (with a drop of 8.4% in nominal terms from the previous year). While social assistance spending did grow in nominal terms (by 1.3%), but as a percentage of GDP, it also reached its lowest level in the last decade: 3.4% of GDP (Bermúdez, 2022). Another encompassing measure is the Global Wage (Salario Global) that was passed in March of 2022 with the Public Employment Framework Law and came into effect in March 2023. In a nutshell, this law forces autonomous state institutions to define wage categories for occupations, eliminating all surplus payments, bonuses and incentives and freezing public wages for indefinite periods. For social policy, concretely, it is feared that the policy will imply a brain drain from the public service sector. With stagnating or decreasing real wages, the well-trained and educated public servants (doctors, nurses, university professors, etc.) might be tempted to find jobs in the private sector. Therefore, while the pandemic presented an opportunity to strengthen Costa Rica’s long-standing commitment to universalism, and some of the introduced measures initially had the potential to expand the country’s social policy regime, this ‘repositioning’ of the state was short-lived. The dominant discourses around austerity, and policy measures accompanying it, seem to have gained the upper hand (Martínez-Franzoni and Sánchez-Ancochea, 2023). This ‘austerity-as-usual’ implies further erosion of Costa Rica’s universal social policy regime, increasingly chipping away from the remnants of what once was a truly unique, solidary, and universal social protection scheme. Author biographies Koen Voorend is Professor of Development Studies and Director of the Institute of Social Research, at the University of Costa Rica. He holds an MA and a PhD in Development Studies from the International Institute of Social Studies, The Hague, and an MA in International Economics Studies from the Maastricht University, the Netherlands. His work centers on social policy and migration, refuge, and more recently on living income. His recent publications featured in Studies in Social Justice (2023); Journal of International Migration and Integration (2022); and World Development (2021). His most recent book ‘A Welfare Magnet in the South? Migration and Social Policy in Costa Rica’ was published in Spanish (San José: Editorial UCR, 2019) and English (Rotterdam: EUR, 2016). E-mail: koen.voorend@ucr.ac.cr/koenvoorend@gmail.com. Daniel Alvarado Abarca is researcher at the Institute of Social Research and a lecturer of the School of Political Science at the University of Costa Rica. He holds a degree in Political Science and Sociology from the University of Costa Rica. His work centers on migration, refuge, social policy, and more recently on living income. His recent publications featured in Studies in Social Justice (2023), Journal of International Migration and Integration (2022), and the Social Sciences Journal of the University of Costa Rica (2022). E-mail: daniel.alvaradoabarca@ucr.ac.cr/daniel.alvarado.abarca18@gmail.com. Funding: The author(s) received no financial support for the research, authorship and/or publication of this article. ==== Refs References Arce A (2020) Ministro de Salud insiste a población sin seguro social acercarse a centros médicos en caso de tener síntomas de Covid-19. SINART, 10 July. Available at: https://costaricamedios.cr/2020/07/10/ministro-de-salud-insiste-a-poblacion-sin-seguro-social-acercarse-a-centros-medicos-en-caso-de-tener-sintomas-de-covid-19/ Arrieta E (2020) 200 mil bonos ‘Proteger’ en riesgo por pulso entre Carlos Alvarado y oposición. La República, 11 June. Available at: https://www.larepublica.net/noticia/200-mil-bonos-proteger-en-riesgo-por-pulso-entre-carlos-alvarado-y-oposicion Bermúdez M (2022) Costa Rica se queda atrás respecto a líderes en gasto social de América Latina. Semanario Universidad, 2 November. Available at: https://semanariouniversidad.com/pais/costa-rica-se-queda-atras-respecto-a-lideres-en-gasto-social-de-america-latina/#:~:text=El%20pa%C3%ADs%20pas%C3%B3%20de%20estar,de%20presentaci%C3%B3n%20en%20el%20mundo EFE. (2022) Costa Rica registra déficit fiscal de 1,8% del PIB al mes de octubre. Swissinfo.ch, 2 December. Available at: https://www.swissinfo.ch/spa/costa-rica-econom%C3%ADa_costa-rica-registra-d%C3%A9ficit-fiscal-de-1-8—del-pib-al-mes-de-octubre/48106000 Elpais (2022) Ministra de Salud niega saturación hospitalaria por Covid-19 en Costa Rica. Elpaís.cr, 25 May. Available at: https://www.elpais.cr/2022/05/25/ministra-de-salud-niega-saturacion-hospitalaria-por-covid-19-en-costa-rica/ Gómez A (2021) Vacunación de migrantes indocumentados divide opiniones en Costa Rica. Voz de América, 28 July. Available at: https://www.vozdeamerica.com/a/centroamerica_vacunacion-migrantes-indocumentados-divide-opiniones-entre-costarricenses/6075528.html International Monetary Fund (IMF) (2021) IMF reaches staff-level agreement with Costa Rica on a three-year extended fund facility and completes 2021 Article IV discussions. Available at: https://www.imf.org/en/News/Articles/2021/01/22/pr2120-costa-rica-imf-reaches-staff-level-agreement-3year-eff-and-completes-21-aiv-discussions Martínez-Franzoni J Sánchez-Ancochea D (2013) Good Jobs and Social Services: How Costa Rica Achieved the Elusive Double Incorporation. Basingstoke: Palgrave Macmillan. Available at: https://www.kerwa.ucr.ac.cr/handle/10669/76419?locale-attribute=es Martínez-Franzoni J Sánchez-Ancochea D (2023) La pandemia de enfermedad por coronavirus (COVID-19) y las narrativas de la política social en Costa Rica: historia de una (breve) oportunidad. Revista de la CEPAL 139. Available at: https://repositorio.cepal.org/bitstream/handle/11362/48798/RVE139_Martinez.pdf?sequence=1&isAllowed=y Ministry of Labor and Social Security (MTSS) (2020) Programa proteger. Available at: https://www.mtss.go.cr/elministerio/despacho/covid-19-mtss/plan_proteger/bono_proteger.html Ministry of Planification and Economic Policy (MIDEPLAN) (2022) Evaluación de impacto del Bono Proteger determinó la efectividad del programa. Available at: https://www.mideplan.go.cr/evaluacion-de-impacto-del-bono-proteger-determino-la-efectividad-del-programa Murillo Á (2021) Gobierno decreta más medidas de austeridad para el período 2021-2025. Semanario Universidad, 11 January. Available at: https://semanariouniversidad.com/pais/gobierno-decreta-mas-medidas-de-austeridad-para-el-periodo-2021-2025/ National Institute of Statistics and Censuses (INEC) (2023) Estadísticas y fuentes. Available at: https://inec.cr/estadisticas-fuentes Presidencia de la República de Costa Rica (2018) Gobierno anuncia medidas de contención del gasto. Available at: https://www.presidencia.go.cr/comunicados/2018/05/gobierno-anuncia-medidas-de-contencion-del-gasto/ Presidencia de la República de Costa Rica (2020) CCSS ajusta medidas financieras de apoyo durante 2020 ante emergencia nacional. Available at: https://www.presidencia.go.cr/comunicados/2020/06/ccss-ajusta-medidas-financieras-de-apoyo-durante-2020-ante-emergencia-nacional/ Programa del Estado de la Nación (PEN) (2022) ¿Cuáles son los principals dsafíos económicos para el próximo gobierno de Costa Rica?. Available at: https://estadonacion.or.cr/cuales-son-los-principales-desafios-economicos-para-el-proximo-gobierno-de-costa-rica/ Voorend K Alvarado D (2021) Costa Rica’s social policy response to Covid-19: Strengthening universalism during the pandemic? CRC 1342 Covid-19 Social Policy Response Series, 6. Bremen: CRC 1342. Available at: https://www.socialpolicydynamics.de/f/62eb122c1b.pdf World Bank (2022) Costa Rica: Overview. Available at: https://www.worldbank.org/en/country/costarica/overview
PMC010xxxxxx/PMC10290926.txt	==== Front Glob Soc Policy Glob Soc Policy GSP spgsp Global Social Policy 1468-0181 1741-2803 SAGE Publications Sage UK: London, England 10.1177/14680181231180518 10.1177_14680181231180518 Article COVID-19 and welfare state trajectory in South Korea: Stagnation, consolidation, or transformation? Shim Jaemin Hong Kong Baptist University, Hong Kong Jaemin Shim, Hong Kong Baptist University, AAB 1129, 11/F, Academic and Administration Building, Baptist University Road Campus, Kowloon Tong 999-077, Hong Kong. Email: jamesshim83@gmail.com Jaemin Shim is also affiliated to German Institute for Global and Area Studies, Hamburg, Germany 21 6 2023 21 6 2023 14680181231180518© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. edited-statecorrected-proof typesetterts1 ==== Body pmcIn order to protect health and minimize financial damage to the public, South Korea responded to the initial phase of the Covid-19 pandemic with various social policy tools. In the first 9 months of 2020, nearly three dozen pieces of social policy legislation were passed along with four rounds of supplementary budget approval, many of which explicitly focused on coping with the pandemic (Shim, 2021). Specific social policy measures adopted ranged from conventional ones such as health care and social assistance supports to more unconventional means such as tax breaks and university tuition fee reductions. In view of the speed and the level of political consensus pertinent to social policy decisions during this period, the social policy responses to the pandemic were swift and bipartisan in South Korea (Shim, 2021). Furthermore, in light of the country’s previous experiences of turning ‘a crisis’ into ‘an opportunity’ for welfare expansion, the pandemic could potentially have a significant effect on the nature of South Korea’s welfare state. More than 3 years have passed since the outbreak of the pandemic. In response to the latest call to update government’s social policy responses to the pandemic from a global perspective (Dorlach, 2023), the goal here is to track key social policy changes (or lack thereof) in South Korea and examine what effect they had on the overall structure of the welfare state. One of the key social policy initiatives proposed by the president Moon Jae-in during the pandemic was the expansion of employment and work injury insurance schemes to theretofore uncovered segments of the population. This includes various forms of self-employed or agency workers such as artists, delivery workers, door-to-door salespersons, and replacement drivers. The president’s ambition to bolster the employment security net was made clear in his ‘Korean New Deal’ road map in July 2020; a year later, the president reconfirmed his intention with the ‘Korean New Deal 2.0’ road map. Upon examining concrete legislative measures taken, the president has filled many blind spots in the employment and work injury insurance schemes. Step-by-step, employment insurance increased its scope and took legal effect, covering various types of artists (December 2020), then workers in twelve different occupation categories (July 2021), and finally, covering messenger service workers and replacement drivers (January 2022). As a result of expanded coverage, nearly a million new workers were newly insured (Korea Workers' Compensation and Welfare Service, 2021). Similarly, the work injury insurance scheme also saw an increase in the number of insured by adding new occupation categories to the eligibility conditions and, simultaneously, making opting out more difficult. The key impetus behind the expansion resembles that of the late 1990s in South Korea – a crisis exposed the blind spots of the existing social security net and prompted the government to execute pro-welfare reforms swiftly. South Korea was hit hard by the 1997 Asian Financial Crisis and saw a nearly five-fold increase in the number of unemployed, reaching 1.8 million (Shin, 2003). However, only a small fraction of employees met the eligibility conditions required to be insured under the employment insurance scheme in place; consequently, poverty soared, and income inequality widened. This crisis necessitated the government to extend insurance coverage to uncovered social groups. In the first half of 1998, employment insurance added employees in businesses containing between 5 and 10 regular employees. Subsequently, in the second half of 1998, employees in businesses featuring between one and four regular employees were added, except those in agriculture, forestry, fishing, and hunting. In 2020, the pandemic revealed the flaws of regular employee-oriented employment insurance. Social distancing and the government’s near-lockdown policies were particularly damaging to the self-employed, who make up a quarter of South Korea’s employment. Moreover, the skyrocketing demand for delivery services led to the corresponding increase in the employment of related jobs, such as agency workers. Yet, these groups were not covered by the existing employment insurance, which prompted the government to intervene. Another landmark social welfare scheme implemented by the government in the early stage of the pandemic was the Emergency Relief Allowance (ERA) offered to all households in April 2020, for example, four-person households received 1 million KRW (EUR 750). This national-level universal cash transfer was offered with the issuance of a national bond and was unprecedented in South Korean social welfare history. Moreover, at the local level, several sub-national-level governments offered various types of social assistance universalistic in nature. For instance, local governments in Seoul, Daegu, and Busan introduced regional vouchers or pre-paid cards to their residents (Soon et al., 2021). Among others, the most noteworthy social welfare provision was installed in 2020 in the Gyeonggi Province, where universal cash transfer was offered to every household under the leadership of Governor Lee Jae-Myung. Had any of the key universal social assistance schemes continued beyond the pandemic period and been institutionalized, it could be said that Covid-19 had had a transformative influence on the nature of the South Korean welfare state. However, over time, universalistic social assistance turned out to be temporary, mostly one-off offers. For instance, six more rounds of ERA were provided between 2020 and 2022. However, unlike the first round, all six ERAs were offered in a targeted manner to those who were hit hard by the pandemic, such as smaller businesses and precarious workers. Phasing out of universalistic cash transfers is reflected in the government’s limited fiscal commitment. Until September 2021, beyond the existing automatic stabilizer, South Korea additionally spent around 3 percent of its GDP in response to Covid-19, a substantially lower proportion than the Organisation for Economic Co-operation and Development [OECD] average of 6 percent (International Monetary Fund [IMF], 2021). Upon examining the political dynamics surrounding the introduction of the first round of ERA, perhaps the chance of genuine welfare state transformation was not high to begin with. Initially, it was a means-tested allowance designed for the lowest 70 percent of income households. However, in the run-up to the general election scheduled for April 2020, there had been race-to-the-top bidding between ruling and opposition party politicians; this eventually extended its coverage to the entire population (Yang, 2020). However, after securing a landslide electoral victory, the ruling party backpedaled and made the ERA more selective (Yang, 2020). This example demonstrates that the multi-party competition incentivizes politicians to prioritize universal social welfare; however, it can also increase election-driven short-termism treating social policies as no more than a voter-attracting tool (Shim, 2019, 2022). Without long-term political commitment and a serious debate on how to secure the necessary funding, a true transformation into a more inclusive welfare state is infeasible, and the South Korean experience during the pandemic years is a testament to that. All in all, it can be said that South Korea’s social policy responses to the pandemic played an important role in consolidating the country’s welfare state. In the existing literature, welfare state consolidation is regarded as the elaboration and extension of existing welfare schemes (e.g. Esping-Andersen, 1996). In this sense, the pandemic revealed critical blind spots of existing unemployment and work injury insurance and catalyzed their beefing up. However, the pandemic did not have a transformative effect of raising the existing welfare state to another level. Universalistic social assistance offered at the beginning of the pandemic was discontinued, and beyond politically flirting with catchy ideas such as ‘universal basic income’, no serious political attempt was made by any major political parties. Author biography Jaemin Shim is an assistant professor at the Department of Government and International Studies at Hong Kong Baptist University. His primary research interests lie in democratic representation, comparative welfare states, women and politics, and legislative politics. He is currently leading the Global Mass-Elite Discrepancy (GMED) Project and Gender and Policy-Vote Trade-Offs Project. ==== Refs References Dorlach T (2023) Social policy responses to Covid-19 in the global south: Evidence from 36 countries. Social Policy and Society 22 (1 ): 94–105. Esping-Andersen G (1996) After the golden age? Welfare state dilemmas in a global economy. In: Esping-Andersen G (ed.) Welfare States in Transition: National Adaptations in Global Economies. Thousand Oaks, CA: Sage, pp. 1–31. International Monetary Fund (IMF) (2021) Fiscal monitor database of country fiscal measures in response to the COVID-19 pandemic. Available at: https://www.imf.org/en/Topics/imf-and-covid19/Fiscal-Policies-Database-in-Response-to-COVID-19 Korea Workers' Compensation and Welfare Service (2021) Status of insured artists and labor providers [Yesul-in Mich Nomujegongja Piboheomja Hyeonhwang]. Shim J (2019) The legislature and agenda politics of social welfare: A comparative analysis of authoritarian and democratic regimes in South Korea. Democratization 26 (7 ): 1235–1255. Shim J (2021) South Korea’s Social Policy Response to Covid-19: Swift and Bipartisan Attempts (Covid-19 Social Policy Response Series). Bremen: Collaborative Research Centre. Shim J (2022) Overpromising social welfare benefits? Electoral competition and welfare politics in Taiwan. Journal of East Asian Studies 22 (1 ): 99–123. Shin DM (2003) Social and Economic Policies in Korea: Ideas, Networks, and Linkages. New York: Routledge. Soon S Chou CC Shi SJ (2021) Withstanding the plague: Institutional resilience of the East Asian welfare state. Social Policy & Administration 55 (2 ): 374–387.33821059 Yang J-J (2020) Covid-19 exposes gaps in South Korea's social security system. East Asia forum. Available at: https://www.eastasiaforum.org/2020/09/23/covid-19-exposes-gaps-in-south-koreas-social-securitysystem/ (accessed 21 April 21 2023).
PMC010xxxxxx/PMC10290928.txt	==== Front Trop Doct Trop Doct TDO sptdo Tropical Doctor 0049-4755 1758-1133 SAGE Publications Sage UK: London, England 37340738 10.1177/00494755231181155 10.1177_00494755231181155 Editor's Forum Harnessing artificial intelligence in the post-COVID-19 era: A global health imperative https://orcid.org/0000-0003-0642-6902 Gulumbe Bashar Haruna 1 https://orcid.org/0000-0002-1074-7920 Yusuf Zaharadeen Muhammad 2 Hashim Abubakar Muhammad 3 1 Lecturer II, Department of Microbiology, Faculty of Science, 487325 Federal University Birnin Kebbi , Birnin Kebbi, Nigeria 2 Assistant Lecturer, Department of Biochemistry, College of Natural and Applied and Sciences, 518070 Al-Qalam University Katsina , Katsina, Nigeria 3 Lecturer II, Department of Computer Science, Faculty of Science, 487325 Federal University Birnin Kebbi , Birnin Kebbi, Nigeria Bashar Haruna Gulumbe, Department of Microbiology, Faculty of Science, Federal University Birnin Kebbi, Birnin Kebbi, Nigeria. Email: bashar.haruna@fubk.edu.ng 20 6 2023 20 6 2023 00494755231181155© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Despite the World Health Organization's declaration that the COVID-19 global emergency has ended, the threat of future pandemics remains a significant concern. This paper highlights the potential role of Artificial Intelligence (AI) in strengthening global health systems and mitigating future health crises. We discuss AI's proven utility throughout the COVID-19 pandemic, including disease surveillance, diagnostics, and drug discovery. AI's ability to rapidly analyze vast amounts of data to derive accurate trends and predictions underscores its superiority over traditional computer technology. However, the effective and ethical implementation of AI encounters significant challenges, including a pronounced digital divide, with applications mainly concentrated in high-income countries, thus exacerbating health inequities. We argue for international cooperation to enhance digital infrastructure in low- and middle-income countries, tailoring AI solutions to local needs, and addressing ethical and regulatory issues. The importance of maintaining evidence-based practice, rigorous evaluation of AI's impact, and investment in AI education and innovation are stressed. Ultimately, the potential of AI in global health systems is clear, and tackling these challenges will ensure its robust contribution to global health equity and resilience against future health crises. Post-COVID-19 AI global health artificial intelligence pandemic edited-statecorrected-proof typesetterts19 ==== Body pmcSir, The World Health Organization's recent declaration that the COVID-19 global emergency has ended symbolizes a significant milestone in combating the pandemic. 1 However, this does not mean the outbreak is over, thus the spectre of future health crises looms large. In navigating this complex landscape, the integration of Artificial Intelligence (AI), a technology that simulates human intelligence processes by machines, particularly computer, 2 into global health systems emerges as an innovative and sustainable strategy.3,4 AI, as it stands, encompasses learning (the acquisition of information and rules for using the information), reasoning (using the rules to reach approximate or definite conclusions), and self-correction.2,5 It is important to note that AI's ability to process and analyze data accurately depends on the quality and veracity of the data it receives. 2 Consequently, when the input data is erroneous, the output or decisions from the AI could also be faulty. AI has demonstrated remarkable utility throughout the pandemic across various applications. 4 In disease surveillance and prediction, AI tools like Canada's BlueDot and Boston Children's Hospital's HealthMap leveraged diverse data sources in real-time, enabling early detection and tracking of COVID-19 cases.3,4 Diagnostic and prognostic applications of AI, such as the model developed by the University of Oxford for interpreting chest radiographs, enhanced clinical decision-making processes. AI also played a significant role in accelerating drug discovery and vaccine development processes. For instance, UK-based company BenevolentAI identified existing drugs with potential efficacy against COVID-19, and Moderna deployed AI in the design of its mRNA vaccine. 6 These AI tools showed superiority over ordinary computer technology through their ability to analyze and draw insights from vast amounts of data quickly, identifying trends and predictions more accurately. 7 Yet, the broader implementation of AI encounters substantial challenges. Notably, the digital divide, with AI applications primarily concentrated in high-income countries, undermines collective security, and exacerbates health inequities. 6 Addressing this divide requires international cooperation and investment to enhance digital infrastructure in low- and middle-income countries. Partnerships between these countries, high-income countries, and tech companies could facilitate technology transfer and capacity building. Moreover, tailoring AI solution development and deployment to different regions’ unique needs and contexts will enhance their relevance and effectiveness. Another critical area of focus is the ethical and regulatory landscape associated with AI. Protecting patient data privacy, ensuring transparency in algorithmic decision-making, and avoiding bias in AI models are paramount. A collective approach, integrating diverse perspectives from health professionals, AI researchers, ethicists, and the public, can foster the development of robust ethical guidelines and regulatory frameworks. 6 The maintenance of evidence-based practice is fundamental as AI continues to integrate into health systems. Rigorous evaluation of AI's impact on health outcomes, including their cost-effectiveness and potential unintended consequences, should be carried out through well-structured research designs. Such evaluations will facilitate understanding the effectiveness of AI applications and their broader implications in the healthcare landscape. Stimulating innovation in AI and investing in education and training is crucial. Encouraging the development of AI tools that can manage the long-term health effects of COVID-19 and predict and respond to future pandemics will be vital. Training healthcare professionals in the use of AI tools and educating the public about these technologies will promote acceptance and appropriate use. Additionally, investing in AI research training will ensure a steady pipeline of talent to drive innovation in this field. The potential of AI as a robust tool to strengthen global health systems is clear. Confronting the challenges of digital equity, ethical and regulatory issues, and rigorous evaluation of AI tools is crucial as the world navigates the post-emergency phase of the COVID-19 pandemic. Undertaking these efforts will ensure AI's role as a formidable ally in global health, bolstering health equity, augmenting health outcomes, and fostering resilience in the face of future health crises. Author contribution: We declare that all listed authors have made substantial contributions to all of the following parts of the manuscript: - Drafting the paper or revising it critically. - Approving the submitted version. We also declare that no one who qualifies for authorship has been excluded from the list of authors. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iDs: Bashar Haruna Gulumbe https://orcid.org/0000-0003-0642-6902 Zaharadeen Muhammad Yusuf https://orcid.org/0000-0002-1074-7920 ==== Refs References 1 Lenharo M . WHO declares end to COVID-19’s emergency phase. Nature 2023. doi:10.1038/d41586-023-01559-z 2 Gillath O Ai T Branicky MS , et al. Attachment and trust in artificial intelligence. Comput Hum Behav 2021; 115 : 106607. 3 McCall B . COVID-19 and artificial intelligence: protecting health-care workers and curbing the spread. Lancet Digit Health 2020; 2 : e166–e167.32289116 4 Gaur L Singh G Agarwal V . Leveraging artificial intelligence tools to combat the COVID-19 crisis. In: Singh PK Veselov G Vyatkin V Pljonkin A Dodero JM Kumar Y (eds) Futur. Trends netw. Commun. Technol, Singapore: Springer, 2021, pp. 321–328. 10.1007/978-981-16-1480-4_28. 5 Artificial Intelligence (AI) \| SpringerLink. https://link.springer.com/chapter/10.1007/978-981-13-3384-2_8 (accessed May 18, 2023). n.d. 6 Chamola V Hassija V Gupta V , et al. A comprehensive review of the COVID-19 pandemic and the role of IoT, drones, AI, blockchain, and 5G in managing its impact. IEEE Access 2020; 8 : 90225–90265. 7 MdM H Islam MU Sadeq MJ , et al. Review on the evaluation and development of artificial intelligence for COVID-19 containment. Sensors 2023; 23 : 527.36617124
PMC010xxxxxx/PMC10290930.txt	==== Front Sociol Race Ethn (Thousand Oaks) Sociol Race Ethn (Thousand Oaks) SRE spsre Sociology of Race and Ethnicity (Thousand Oaks, Calif.) 2332-6492 2332-6506 SAGE Publications Sage CA: Los Angeles, CA 10.1177/23326492231177639 10.1177_23326492231177639 Original Research Article “American Tales of Heroes and Villains”: Donald Trump’s Framing of Latinos During COVID-19 Times https://orcid.org/0000-0003-4422-7429 Viladrich Anahí 1 1 Queens College & The Graduate Center – The City University of New York, CUNY, USA Anahí Viladrich, Department of Sociology, Queens College & The Graduate Center – The City University of New York, 65-30 Kissena Boulevard, Powdermaker Hall 233 A, Queens, NY 11367-1597, USA. Email: anahi.viladrich@qc.cuny.edu 16 6 2023 16 6 2023 23326492231177639© American Sociological Association 2023 2023 American Sociological Association This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Based on a qualitative analysis of Donald Trump’s speeches and public documents from 2020, this article examines the role of xenophobia in constructing oppositional divisions within Latino groups in the United States. Rather than pitting minority ethnic/racial groups against the White majority, xenophobia frames unauthorized populations against legal, albeit subordinated, ones. Five main Latino categories are identified in this study. First, the “illegal immigrant” is portrayed as the criminal border crosser that targets other Latinos—the latter embodied by the “Hispanic victim.” Next, is the “Hispanic border patrol” agent who safeguards the United States by actively detaining and expelling undocumented immigrants. Third, the “Hispanic supporter” is welcomed into the American Dream by ascribing to meritocratic values of hard work and family values. A final actor is represented by foreign allies (e.g., Mexico’s President) who crack down their own citizens to protect the United States border. Furthermore, this article discusses Trump’s xenophobic camouflage of race (and racism) by highlighting undocumented Latinos’ alleged immoral and criminal nature rather than their physical characteristics. Concomitant to this narrative is the conditional inclusion of a subset of Hispanics into the American dream. In the conclusions, the article compares the study findings with the results of the 2020 presidential election to shed light on the growth of Trump’s Latino base. This research piece ultimately provides a contribution to our understanding of the conceptual power of xenophobia in galvanizing divergent interests within racial and ethnic minorities, in this case Latinos in the United States. Trump Hispanics Latinos xenophobia racism Rousseau Gift 78202-00-03 PSCCUNY 65367-00 53 edited-statecorrected-proof typesetterts1 ==== Body pmcIntroduction In September 2020, I signed up to volunteer for a phone bank organized by a handful of progressive unions in the New York metropolitan area. My colleagues and I were enlisted to reach out to Latinos in battleground states, mostly Texas and Florida, to convince them to cast their votes for Joe Biden in the upcoming Presidential election. As part of a 101 virtual training, we were tasked to remind them of then-President Trump’s appalling handling of the COVID-19 pandemic, along with the public policies that had failed Latinos amid his draconian immigration agenda. Less than an hour into our phone marathon, I realized that the contagious optimism I had initially shared with the other volunteers was not reciprocated by a sizable number of those on the other end of the line. The social scientist in me wanted to know more about them so, for the next two and a half hours, I inquired about their motives for supporting Trump. Despite the wide variety of reasons they gave, including Trump’s purportedly conservative family values and his anti-socialist agenda, a common thread in their responses was his stand on immigration issues along with his commitment to rebounding the economy during the pandemic. They were also concerned about Biden’s potential lockdowns if he won and his alleged “open door” migration policy, which would presumably allow undocumented immigrants to continue “jumping the line” to unfairly take advantage of government handouts. I left the phone bank convinced that there was something critical to be learned from Trump’s “rhetorical efficacy” (Schaefer 2020) in attracting disparate Latino groups during an out-of-the ordinary pandemic year. A month and a half later, the results of the 2020 Presidential Election reported an increase in Latino support for Trump at the national level. Despite his overt anti-immigrant views, he scored a record number of votes among Latino populations, particularly those without a college degree and in red states, especially Florida and Texas (Equis 2021; Garza 2021; Igielnik, Keeter, and Hartig 2021). Contrary to the portrayal of Latinos as a monolithic aggregate, recent scholarship has showed their enormous diversity, a large segment of which favored Trump in the United States Presidential elections of 2016 and 2020 (Alamillo 2019; Corral and Leal 2020; Galbraith and Callister 2020; Gonzalez-Sobrino 2020, 2021). The trend in traditional media to frame immigration policy as Latinos’ main concern has failed to recognize not only their heterogeneity but also the array of divergent issues that set them apart (Cadava 2020; Dávila 2012). Research on Hispanic Republicans has shed light on how conservative discourses, including anti-Black Latino racism (Haywood 2017), have aligned with their paths to incorporation (Alamillo 2019; Cadava 2020; Garza 2021). With Donald Trump rising as a Presidential candidate in 2016, the United States witnessed the gradual marriage of right-wing media with the executive branch which, in due course, contributed to the polarization of the electorate (Stelter 2020; Yang and Bennett 2022). As the leading media channel, Fox News (Rupert Murdoch’s multibillion-dollar empire) exerted a significant influence on Trump and the Republican party generally (Stelter 2020). As a tour de force of unpaid advertising, Trump paired up with Fox News to forge a feedback loop of interactive propaganda through which they generated repetitively broadcast news—to the point that it eventually became unclear who was following whose lead (Ott and Dickinson 2019; Stelter 2020; Yang and Bennett 2021). At the time, reports of undocumented immigrants killing American citizens were a recurrent feature of Fox programs such as Fox and Friends. These stories were designed to instill fear and hate, particularly toward foreigners. In April 2020, at the height of the COVID-19 pandemic, Trump’s announcement that he would “suspend immigration,” followed by the rapid expulsion of asylum seekers under Title 42, was triggered by host Tucker Carlson’s on-air call for Trump to take a tougher stance on immigration. Since the 2016 United States Presidential election, scholars and the general public have been fascinated by Trump’s relatively quick rise to political stardom: from media spectacle to spokesperson for a new wave of neopopulism (al Gharbi 2018; Kellner 2017). The sociological literature has been prolific in addressing the complex, even contradictory forces behind Trump’s political ascent among White and minority groups—mostly Christian and conservative. A recent body of sociological research has focused on the “Trump phenomenon” in relationship to the rise of global authoritarianism, racism, and right-wing populism (Baker, Cañarte, and Day 2018; Massey 2021; Smith 2019a). This literature has shown how the Trump administration strengthened nativism and xenophobia in the United States, greatly supported by the mainstream and social media’s amplification of his anti-immigrant agenda (Baker et al. 2018; Canizales and Vallejo 2021; Cervantes and Menjívar 2018; Louie and Viladrich 2021; Silber Mohamed and Farris 2020a). Much of this prolific body of research has focused on Trump’s derogative images of Latino immigrants as lawless criminals (Cervantes and Menjívar 2018; Menjívar 2016; Sanchez and Gomez-Aguinaga 2017). Scholars have particularly addressed Trump’s obsession with the United States-Mexico border and his promises to finish building the wall—a physical and symbolic manifestation of apartheid and a prime symbol of ethnoracial exclusion (Heuman and González 2018; Massey 2021). Conversely, a gap still exists in the literature regarding the complex reasons behind Latino support for conservative and anti-immigrant agendas, along with the type of discourses deployed by Republican representatives to attract them. In addressing the latter, this study provides an original contribution to the sociological field of race and ethnicity by studying how xenophobic discourses that exclude “criminal Latino aliens” also grant entry to subsets of the Latino population. By examining Presidential speeches and official documents issued in 2020, a pandemic and national election year in the United States, this article shows how xenophobia works by labeling difference: Hispanic patriots versus illegal border crossers, and defenders of national integrity versus spreaders of foreign pathogens. The choice of the year 2020 for the analysis signals the confluence of two critical phenomena: the onset of a global pandemic and a United States Presidential election year that featured the largest number of voters generally, and of Latinos particularly (Bergad and Miranda 2021). The convergence of these events offers a unique opportunity to witness Trump’s rhetorics regarding Latinos amid his divisional politics. While the pandemic prompted Trump’s refurbished xenophobic tropes (e.g., his obsession with building the wall to keep “illegals” away along with his anti-Asian rants), the impeding election increased his efforts to attract the Latino vote by relying on both inclusionary and exclusionary stances. It is precisely here, in the production of discourses, where we began this research journey. We next turn to the notion of xenophobia, identified here as the main conceptual construct for analyzing and interpreting the study results. The methods section will then be presented, followed by the research findings that, both in their descriptive and qualitative form, reveal Trump’s conditional inclusion of legal Latino groups (i.e., called Hispanics by Trump) vis-à-vis the exclusion of undocumented Latino immigrants—including asylum seekers. The study results will highlight the main Latino actors identified in Trump’s public speeches and documents, in conversation with central dimensions of his xenophobic construct. Against the image of the “illegal criminal,” Trump’s preferred Latinos are portrayed as industrious legal immigrants fully committed to defending the nation against undocumented trespassers. In Trump’s narrative, the former are represented by four distinct but intertwining categories: the Border Patrol agent, the Hispanic supporter, the victim of crimes perpetrated by “illegal aliens,” and Latin American heads of state aligned with Trump’s national security doctrine. The discussion section will review these findings in dialogue with the scholarly literature and, in the conclusions, this piece will provide a comparison between the study results and the outcomes of the 2020 Presidential election toward shedding light on the growth of Trump’s Latino base. Conceptual Framework: Deconstructing Xenophobia During Trumpist Times Xenophobia (etymologically xénos, meaning “stranger” and phóbos, meaning “fear”) is generally conceptualized as apprehension and hatred directed to those considered alien to one’s cultural identity or nationality. Although this term tends to be used and juxtaposed with others such as nativism and racism, an understanding of their differences is key to better conceptualizing their respective reach and impact (Kim and Sundstrom 2014; Saito 2021; Sundstrom 2013). Despite being semantically related, racism and xenophobia are not the same, as the latter entails the racialization of those excluded from the national polity—even if they belong to it (Saito 2021). More pervasively complex than binary racist dimensions (i.e., White and Black) xenophobia requires the exclusion of the “other” in an in-group versus out-group dichotomy (Baker et al. 2018; Wimmer 1997). The differences between racism and xenophobia can also be understood in terms of hierarchization as racism implies the tiering of out-groups, while xenophobia identifies an out-group response without rankings (Wimmer 1997). While open racism is not generally accepted, xenophobic rejection of the other tends to galvanize the nation against the divisions promoted by internal racism (Saito 2021). Unlike nativism, xenophobia is not linked to the nation-state since racial/ethnic groups may express xenophobic claims without conveying nativist connotations. An example would be the case of nationless, nomadic populations that may act upon out-group rivalries against other tribes—while xenophobic, these reactions do not involve nativist claims (Baker et al. 2018; Sundstrom 2013). Furthermore, nativism denotes the preference for, and privileges granted to, the native-born population, while xenophobia conveys the rejection of the foreign born that are presumably antagonistic to the nation’s moral principles (Fernandez 2013). Recent scholarship has highlighted the contribution of German political theorist Carl Schmitt to the conceptualization of the “other” in the development of modern xenophobia amid national populism, islamophobia, White nationalism, and its sister movements in Europe (Henley and Warren 2020; Mannarini and Salvatore 2020; Marin 2016; Sundstrom 2013). As noted by L. Marin (2016), even if Schmitt’s theories do not directly translate into racism or chauvinism, his community-based politics is reliant on the exclusion of the other(s) and “leads directly to xenophobia as an unavoidable result of the political” (p. 312). In The Concept of the Political (Schmitt 2007), originally written in 1927, Schmitt argues that state sovereignty is dependent on the identification of public enemies: strangers who are not necessarily foreign and can be anyone, such as a religious group, a political party, a sexual or racial minority. What becomes politically relevant for Schmitt is the public construction of the enmity itself, which finds its counterpoint in a homogeneous nation. In this view, the opposition between “friend and enemy” is crucial in mobilizing mass support and a sine qua non condition for the existence of a united nation. As will be shown in this article, the xenophobic construction of the “ally/enemy” involves a more complex toolbox than Schmitt’s binary distinctions, in which racism is masked as foreignness of both newcomers and racialized minorities (i.e., the perennial foreigner). As a twofold process, “othering” is directly rooted in racism (i.e., perceptions of White racial superiority) and xenophobia (i.e., fear/hatred of foreigners). Rhetorical constructions of “othering,” leading to stigmatizing discourses and practices, have largely informed Latinos’ experiences of exclusion (Cervantes and Menjívar 2018; Chavez 2013; Lamont, Park, and Ayala-Hurtado 2017; Menjívar 2016). Likewise, the distinction between xenophobia and racism should not preclude us from assessing their interdependence. Racist xenophobia is rampant, particularly in the developed world, as it is mostly people of color who are excluded and demonized. In a related framework, racist nativism—or the racialized framing of noncitizens as threats—is also key to understanding United States racial hierarchies (Huber et al. 2008; Louie and Viladrich 2021). Xenophobia and racism also share common ground in the reproduction of exclusion and dominance. Whereas racism focuses on physical characteristics to target purportedly inferior groups, xenophobia rejects those perceived as foreigners—even if members of the same community (Wistrich 2013). Rooted in racist xenophobia, the “perennial foreigner” and “alien citizen” stigma are applied to those who, although legally belonging to the nation-state, are perceived as strangers who live among us (Kim and Sundstrom 2014; Ngai 2014). Likewise, the notion of alien citizenship directly speaks to a judicial field that both grants and removes legal rights. Two examples of this phenomenon are the expulsion of thousands of ethnic Mexicans during the Great Depression, and the internment of United States citizens of Japanese origin during the Second World War (Ngai 2014). Similarly, concerns about “illegal aliens” during Trump’s presidency led to lawmaking efforts to strip birthright citizenship from immigrants seen as unworthy of it. Trump’s diatribe against “anchor babies” was directed at removing citizenship rights from the children of noncitizens born in the United States, under the assumption that their mothers had given birth in this country for personal gain and at the expense of American taxpayers (Foster 2017). During the last year of his presidency, Trump’s xenophobic narrative was clearly seen in his vilification of Asian communities, especially in his framing of COVID-19 as the “Chinese virus” and the “Kung fu virus.” Research has shown how such xenophobic rhetoric led to racial and ethnic hate as well as verbal and physical violence against Asian populations in the United States and overseas (Gover, Harper, and Langton 2020; Louie and Viladrich 2021; Viladrich 2021). Rather than rejecting the “other” on the basis of skin color or other phenotypical characteristics, xenophobia focuses on foreigners’ nonvisible traits—including their alleged (im)moral and deviant features—as the rationale for expelling them from the nation-state. Hence, xenophobia offers a channel for covert racism to express itself under the guise of protecting the public from immigrants’ unlawful and unethical mores. For instance, the scorn directed against Muslims by Islamophobes, rather than being rooted in race, is justified on the basis of Islam’s alleged incompatibility with Western cultural values and democratic principles (Baker et al. 2018; Sundstrom 2013). N. T. Saito (2021) argues that xenophobia—in its shaping of a foreign population as a threat—achieves what racism cannot: it allows the latter to spread its wings by camouflaging racial contempt as fear of those who allegedly bring deadly pathogens and moral turpitude. As will be discussed in this article, using taxpayer dollars to build an astronomically expensive border wall is justified by invoking the hordes of Latin American invaders that must be stopped. Neither weak nor innocent, the latter are portrayed as morally inferior and tenacious adversaries always ready to attack. Research Design and Methods This article is based on the qualitative content analysis of former President Trump’s speeches, press briefings, and official documents released during the year 2020. This period, known as “the great coronavirus pandemic year” (Gostin 2020) offers a unique opportunity to understand how the 45th United States President navigated a public health crisis of unprecedented proportions by labeling particular immigrant groups as either allies or “virus vectors.” An analysis of Trump and his public statements allows us to tap into a first-order production of public messages (Petroff, Viladrich, and Parella 2021). Trump’s speeches, official statements, and press releases were culled from two main repositories: the Whitehouse archive (https://trumpwhitehouse.archives.gov/) which is the official website of the United States presidency, and the American Presidency Project (https://www.presidency.ucsb.edu/), a free online university-based database for public Presidential documents hosted by the University of California at Santa Barbara. A third source, NEXIS Uni (and academic database engine), was accessed through the University’s online library to further identify any of Trump’s public appearances (e.g., roundtables) that might have been missed by the other web resources. The following Boolean terms, in alphabetical order, were used in conducting the searches (separately and combined): “Hispanic/s,” “illegal/s,” “immigrant/s,” “Latino/s,” “Latin-American migrants,” “migrants,” “President Trump,” “Trump,” “undocumented,” and “US/United States President.” The main selection criteria for gathering the documents were that former President Trump was the prime speaker at official, government-sponsored events, or the author and/or signatory of official documents, as in the case of White House press releases. This project did not consider entries from his social media feeds, about which much research has already been conducted. For the sake of consistency, this study only included transcripts and records that explicitly addressed Hispanics/Latinos in the United States, either by referring to them directly (i.e., mentions of Hispanic Americans) or indirectly, as when Trumps used the terms “illegals” or “Mexicans,” which implicitly referred to undocumented Latin American immigrants. An initial search led to 112 records divided among speeches, official White House statements, and press releases. After eliminating duplicate transcripts, 89 items were left in the final corpus. Three types of in-person events were included: rallies (i.e., mass demonstrations of Trump supporters), spoken addresses (i.e., formal speeches and press briefings), and roundtables. The latter refers to semi-informal settings during which Trump sat down with other speakers (e.g., Latinos for Trump roundtables) to share his experiences and hear the attendees’ personal and political trajectory. Each document was uploaded to Dedoose, a web-based application utilized for qualitative and mixed-methods analysis (Dedoose 2021). The analysis followed a constructivist and inductive approach, which, in agreement with grounded theory principles, required the texts to be read several times for the purpose of identifying themes and common narratives throughout the corpus (Glaser and Strauss 2017). In a second stage, codes (i.e., empirical indicators of conceptual categories) were created and applied during text analysis—a process called “code mapping” (Gonzalez 2016). For instance, references to Latinos’ role in Border Patrol units were classified under the code “Patriotic Duty,” and tropes indicating the alleged criminality of “illegal trespassers” were labeled under the “Criminal Aliens” code. A codebook was created to categorize the selected excerpts (i.e., phrases, quotes, and passages from speeches), which were then entered in Dedoose and linked to each code. The analysis resulted in 22 codes and 314 coded excerpts across five distinct thematic categories: Hispanic Supporters (98), Latin American Allies (45), Border Security/National Security (78), and Illegal Aliens (93). Identifiers for each of the recorded materials were then entered and classified according to type of document (i.e., rally, roundtable, press release), target audience (e.g., Hispanic supporters), place where the speech/document took place or was released (e.g., the White House), and date (i.e., month and day in the year 2020). Research Findings Descriptive Results Table 1 disaggregates the documents according to their specific type. Trump’s preference for in-person events is reflected in the fact that this category accounted for approximately 65 percent of the materials (N = 58) versus 35 percent of his official written documents (N = 31). Table 1. Documents Analyzed by Type. Document/Event Type Frequency Briefing and spoken Addresses 29 Press release and White House statements 31 Rally 24 Roundtable 5 Total 89 Graph 1 illustrates the distribution of the population categories that appeared in the coded excerpts across the documents (i.e., Hispanic Supporters, Conditional Allies, and Illegal Alliens). As shown in this figure, Trump’s rants about “illegal alliens” were most apparent during his rallies (i.e., MAGA events) when he typically stirred his base against “criminals” crossing the United States-Mexican border. Conversely, his references to Hispanic supporters were more marked during the roundtables, which were mostly organized by his Latino followers (i.e., Latinos for Trump), and spoken addresses (i.e., campaign events), in which his target audience included the general Latino population. The third population category (Latin American heads of state as conditional allies) appears consistently across all documents, signaling these countries’ support for Trump’s national security doctrine. Graph 1. Population categories across excerpts by document type. Graph 2 presents the number of excerpts associated with each population category that appeared in the corpus during 2020. The figure of the “illegal alien” was most prominent during the first three-month period of 2020. Meanwhile, Trump’s calls to his “Hispanic supporters” peaked in September and October, close to the end of the campaign period, and decreased noticeably after the United States Presidential election in November 2020. Graph 2. Distribution of population categories by month. We compared Graph 2 with poll data presented in Graph 3 (taken from Equis 2021) that shows Latinos’ approval of Trump on major issue areas and found striking correlations between the two. Graph 3. Average marginal effects on the probability of voting for Trump (vs. Biden) on major issue areas, among all Hispanics. As per Graph 3, in early March 2020, Trump’s immigration agenda—led by his calls to close the border and prevent “illegals” from crossing it—earned the highest approval among his Latino sympathizers. As the pandemic progressed, Trump’s Latino support increased with regard to his handling of the COVID-19 pandemic and the economy (i.e., his efforts to reopen businesses and raise employment levels, along with his government’s financial stimulus). According to post-election results, Trump emphasis on the latter explained his uptick among Latinos in the 2020 Presidential Election (Equis 2021). As will be discussed later, Trump asked his Hispanic supporters to keep the economy open during COVID-19, as he pledged to protect both their financial stability and, ultimately, their entry into the American Dream. Finally, Graph 4 brings together the conceptualization of the categories presented above by introducing the study’s framework: “Xenophobic Conceptualization of Latinos in the United States” On the right of the graph, the term “Hispanics” (i.e., legal Latino immigrants and their families) represents the “included” population and is divided into three categories: the “Hispanic Border Patrol agent”; the “Hispanic supporter,” who is given access to the American Dream; and the “Hispanic victim” of crimes perpetrated by “illegal aliens.” Trump’s conditional allies, in the center of the graph, are epitomized by Latin American countries (and their representatives) who supported Trump in his fight against “illegal immigration.” Finally, the category of “excluded” Latin Americans, on the left of the graph, is divided into two groups: “Border crossers” (i.e., the MS-13 gang, drug traffickers) and “domestic enemies”—undocumented immigrants protected by sanctuary city laws. We now turn to analyzing the qualitative meaning of the findings drawn from these main population categories. Graph 4. Xenophobic conceptualization of Latinos in the United States. Hispanics at the Border: Exposé of the Unfailing Surveillant During the final year of his presidency, Trump’s promise to complete the United States-Mexican border wall became one of his main battle cries, and he consistently encouraged his base to engage with the “build the wall” chant—all amid the largest public health crisis in the United States in more than a century. Meanwhile, he oscillated between minimizing the impact of the COVID-19 pandemic and denying it altogether, such as when calling it a Democratic “hoax,” or blaming China and Latino border crossers for its spread. Trump’s keen maneuvering of divisional politics achieved its maximum expression when framing the southern wall as his own persona, while praising Hispanic Americans for being vigilant against the nation’s enemies—namely Latino “trespassers.” During a “Latinos for Trump Roundtable” in Las Vegas (on September 25, 2020) he stated,I’m a wall between the American Dream and—and chaos and—and a horror show, a horror show. It would be very bad. It would be very bad. Many Latino and Hispanic Americans came here to pursue the American Dream, having left countries that really had a very, very unruly group of people. You know, we’re throwing out MS-13 all the time . . . But they’re rough. They’re bad. And they’re bad people . . . They’re bad people, and we throw thousands and thousands out of our country. And if we didn’t have ICE and these people [Hispanic American Border Patrol officers] you would be living in real fear and real problems. A few key findings transpire from the excerpt above. First, Trump’s “wall” is continuously invoked as reassurance of the physical and symbolic barrier that will stop “illegal” Latin Americans from entering the United States. The wall as a protective fortification of the nation-state is a classic example of xenophobic language that claims to be in favor of racial integration while simultaneously denigrating specific racial/ethnic groups—not based on their phenotypic markers but on their alleged incompatibility with American values. Trump’s bombastic manner of speech was more palpable during his public appearances than in his written documents and was aimed to rile up his base—both White people and members of ethnic minorities. As noted by Hirschfeld David and Baker (2019), Trump’s advisors ingeniously produced the “build the wall” chant to enthuse his supporters whenever they seemed disengaged. Trump’s wall also speaks to the notion of the sovereign ruler who will not hesitate to use an iron fist to save his nation, as described by C. Schmitt (2007) early on. Trump’s praise of Hispanic Border Patrol agents—who keep the “bad hombre” away—is also consistent across all his in-person appearances. Public assertions of rising immigrant criminality are a sine qua non condition for framing the Hispanic protector, whose main duty is to defend the physical and symbolic boundaries of the nation. Trump repeatedly reminded his followers of the large number of Hispanics serving in the Border Patrol and military and claimed that they understood and wanted to protect the southern border more than anyone. During a rally in Yuma, Arizona (August 18, 2020), he referred to them as “heroes that protect our nation in uniform.” That includes the millions of incredible Hispanic Americans who follow our laws, uplift our community [applause] and protect our nation, in uniform. Half of all Border Patrol agents are Hispanic Americans; I was just with them. And today I salute each and every one of those true American heroes and that’s what they are. [Applause]. And you know nobody understands the border better than Hispanics. They know what’s well, what’s bad. They don’t want bad people coming into our country, taking their jobs, taking their homes, causing crime. Hispanic Americans are the people that are most in favor of what we’re doing on the border because they understand it. They understand it better than anybody. With the help of these patriots, we’ve stopped the rampant asylum fraud, shut down the human smugglers and we’re finding the drug dealers, traffickers, predators, and we’re throwing them the hell in jail or sending them back home. [Applause]. The recruitment of Border Patrol agents among people of color has been a longstanding tactic to assure compliance on the part of subaltern groups to help build a unified political façade. In fact, minorities have historically been the first to be called to the front during armed conflicts (Buckley 2002; Saito 2021). The reiterated convocation of a powerful enemy is also needed to feed an expensive technocratic military apparatus able to defend the country against it. Finally, the construction of the “illegal other” keeps Hispanics’ demands in check by diverting attention from domestic issues, including racial and social inequality, all while devising a common enemy that is blamed for the nation’s woes (Saito 2021). Promises of Whiteness: Hispanics Joining the American Dream I’ve achieved more for the Hispanic Americans. And think of it, I’ve achieved more for Hispanic Americans in 47 months than Joe Biden has achieved in 47 years. I’ve been here for 47 months. Donald Trump, Latinos for Trump Roundtable, Florida, September 25, 2020. In Trump’s view, Hispanics are committed stakeholders that have prospered under his leadership and will continue to do so. Due to their proven Americanness, they are welcomed into the American Dream—a reward for their commitment to protecting and investing in the nation-state. During a White House event honoring veterans of the Bay of Pigs Invasion (September 23, 2020), the former President remarked,Hispanic Americans embody the American Dream. And my administration is delivering for you that American Dream like nobody has ever delivered for the Hispanic Americans and, hopefully, for everybody else. We implemented the historic tax cuts, regulation cuts, and I recently created the Hispanic Prosperity Initiative to expand economic opportunity. At the height of the pandemic and closer to the 2020 election, Trump’s speeches increasingly switched from his anti-immigrant stand to prioritizing his rebounding of the United States economy. This was paired with his praising Hispanics for their individual enterprise and hard work as their entry ticket to the American Dream. Promising the latter has been a political strategy of the Republican Party since Reagan (Sosa 2009). Trump’s extension of the American Dream to eligible Hispanics is also in tune with the latter’s aspiration for whiteness supported by their meritocratic achievements (Filindra and Kolbe 2022; Hochschild 1995). Research has consistently pointed out the association between becoming American and achieving whiteness, on one hand, and seeking inclusion into the American Dream and embracing the larger White majority, on the other (Basler 2008; Menjívar 2016). Basler’s study (2008) revealed how Mexican American respondents aligned themselves with Whites (against Blacks) for the sake of gaining the latter’s social and political capital. By identifying as White, Basler’ participants channeled their demand for validation and protective inclusion into an imagined American community. Trump’s promises to improve Latinos’ financial prospects during COVID-19, while taking credit for their pre-pandemic success, became a staple of his public speeches. On July 9, 2020, he signed an executive order, the Hispanic Prosperity Initiative, which promised to expand jobs and educational opportunities for Latinos. Once again, he deflected blame for the United States health crisis by referring to the pandemic as the “Chinese virus” and promised to reopen the economy as soon as possible to return Hispanics to their pre-pandemic employment levels. The Hispanic Americans and the Hispanic American community is a treasure. Thank you. . . . Before the plague from China came in—you know what that is; that’s the China virus—before it came in and hit us, we achieved the lowest Hispanic American unemployment rate and the lowest poverty rate ever recorded—history of our country—ever recorded. And we’re getting back to it very quickly. We achieved the highest ever incomes for Hispanic Americans and many other American groups and communities. At a rally held in Henderson, Nevada (September 13, 2020), the former President celebrated his economic achievements on behalf of his Hispanic supporters, while reminding them of their role in keeping “criminals from coming across” the border. For the last four years, I’ve been delivering for our incredible Hispanic community; I’m fighting for school choice, safe neighborhoods, low taxes, low regulations on all Hispanic-owned small business, and they are great businesspeople. And they understood that. . . . Our Hispanic population knows our southern border better than anybody else and they don’t want criminals coming across. They want people to come across, but they want them to come across legally. And we have the strongest southern border now that we’ve ever had. [Cheers and applause]. Hispanics, as constructed in Trump’s narrative, are fierce individual entrepreneurs, defenders of the Second Amendment and advocates of heterosexual, religious, and anti-abortion family values—as an extension of the ideal Trumpist worker, they could be labeled as “rugged meritocrats” (Cech 2017). Still, Trump’s rhetorical investment in his Hispanic supporters was dependent on their proven support and patriotism in the fight against “illegal trespassers.” Framing the Internal Enemy: Hispanic Victims Versus Illegal Perpetrators During his Presidential campaign, Trump made explicit references to the horrific killings of American citizens at the hands of “criminal aliens” and blamed sanctuary cities—municipalities that protect undocumented immigrants from Homeland Security—for providing cover to illegal outsiders. These stories typically highlighted the murders perpetrated by those unlawfully residing in United States territory. Contrary to empirical evidence (Collingwood and O’Brien 2019), Trump framed sanctuary cities, such as San Francisco and New York, as magnets for undocumented immigrants who were allegedly responsible for these cities’ high crime rates. At an event organized by the Members of the National Border Patrol Council (February 14, 2020), Trump detailed the horrific rape and murder of a 92-year-old Hispanic woman, Maria Fuentes, at the hands of an “illegal criminal”:In my State of the Union Address, I shared the tragic story of Maria Fuertes, a 92-year-old great-grandmother who was allegedly raped, beaten, and murdered by a criminal illegal alien in New York City. I think I can—no, I don’t think—I’ll take the word “allegedly” out. She was raped, beaten, and murdered. Only five weeks before the murder, the criminal alien had been arrested for assault. ICE asked for the criminal to be turned over but, instead, he was released under New York City’s sanctuary laws. If New York had simply honored ICE’s detainer request—very simple thing to do—Maria Fuertes could be with her family right now. And we’re deeply moved to be joined this afternoon by Maria’s grieving granddaughter, Daria. Where is Daria? Daria Ortiz. Daria. Please come up, Daria. Please. Daria is joined by her beautiful five-year-old son. Thank you, Daria. It’s a great honor! During his first Presidential campaign, Trump began calling mothers and families whose children had lost their lives to undocumented immigrants “Angel moms” and “Angel families.” After his election, this group became an informal “think tank” that supported his claims that unauthorized immigrants were on a killing spree of innocent Americans. Fox and other pro-Trump think tanks, such as America First Policies, helped amplify the plight of Angel moms along with Trump’s crusade to build the wall and keep “illegals” at bay (Fearnow 2019). On October 20, just a few weeks before the United States Presidential election, the Trump Administration released a document titled “National Day of Remembrance for Americans Killed by Illegal Aliens,” which commemorated those whose lives had been “egregiously taken from us by criminal illegal aliens.” Trump promised to prevent similar crimes and seek justice on behalf of the victims. In 2014, Mary Ann Mendoza’s son, Hispanic Sergeant Brandon Mendoza, died in a car collision with an undocumented immigrant. During a “Students for Trump Rally” in Phoenix, Arizona (June 23, 2020) Trump referred to Ms. Mendoza, who later became a prime spokesperson for the Angel Moms, as follows:I also want to recognize a truly courageous woman, “angel mom”—she’s a friend of mine too, by the way. One of the first people I met when I announced I was going to run. What she has gone through with illegal immigration, nobody will ever go through. Hopefully, nobody ever has to go through. And she lost a magnificent child to an illegal immigrant. And it’s just a horrible thing. And she spent her life—she spent the last long period of years fighting and fighting and fighting. And people have gotten to respect her greatly: Mary Ann Mendoza, wherever you might be. A lot of guts. She’s got a lot of guts. The discursive efficacy of anti-immigrant narratives achieves new heights when the foreign invader makes other Latinos a prime target. Not only is xenophobia here deployed to camouflage racism by endowing “the other” with essential immoral traits, but it also offers the tools for its internalization among specific Hispanic groups. The trope of the “enemy within” highlights the wrongdoings of outsiders who, rather than being detained at the border as they should have been, get a free ride to sanctuary cities to “kill our children.” Finally, the dual-frame illegal criminal-Hispanic victim puts into evidence how the notions of the external and internal enemy become entangled into one single entity: a border trespasser that metamorphoses into a sanctuary city’s illegal refugee. Conditional Allies in Defense of the Nation-state Devised to recruit support for his reelection campaign, Trump’s initial promise that Mexico would pay for the southern border wall continued throughout 2020. Meanwhile, he did succeed in erecting what it has been called an “invisible legal wall” (Smith 2019b) represented by the Third Country Transit Bar—his version of an anti-asylum agreement that he signed with Central America’s Northern Triangle countries (El Salvador, Honduras, and Guatemala). Never using the term asylum petitioners, Trump consistently called them “illegal criminals” and “gang members” as at a rally in Milwaukee, Wisconsin (January 14, 2020):We are removing these illegal criminals and gang members by the tens of thousands, and we will not let them back. And we made a deal with Guatemala, Honduras, El Salvador, Mexico, where they accept them back. Under the Obama administration, they’d get them, they’d bring ’em back. And Honduras and Guatemala, would say “go to hell—we’re not taking them back.” But with us, they take ’em back. They take ’em back. [Applause]. Very great! In the meantime, the Mexican Government had agreed to deploy their troops to help stop migrants from entering the United States, in return for preventing sanctions that would otherwise force Mexico to pay additional tariffs on its exports and remittances. Mexican President López Obrador became a prime actor in this discursive repertoire by endorsing additional bilateral agreements, including the Migrant Protection Protocols (MPP) known as the “Remain in Mexico” program. Under the MPP, asylum seekers were ordered to wait in Mexico until their court date was set in the United States Lack of legal counsel, plus the financial and physical insecurity of waiting at border towns, made it almost impossible for asylum seekers to win their cases (American Immigration Council 2021). At a Rally in Yuma, Arizona (August 18, 2020), Trump praised the Mexican President for his backing of the MPP as follows:I entered into a historic partnership with Mexico, known as the migrant protection protocols, to safely return asylum seekers to Mexico, while awaiting hearings in the United States. You know about that [in the past] we had to have them in the United States, and we captured them, we had to keep him here. I said: “No, no, we don’t want them here, we want them outside.” We got sued all over the place and we won. So now they don’t come into the United States, they can wait outside.” Trump’s references to his “love affair” with Mexico was evident in his frequent praising of López Obrador for having joined forces to protect the United States-Mexico border, as reported at a “Keep America Great Again Rally” in Iowa (January 30, 2020):Right now, we have a love affair with Mexico because the Democrats, the Democrats wouldn’t give us what we needed and I got Mexico, they’re great. They put up 27,000 soldiers on our southern border and the numbers are plummeting, 87% down [cheers and applause]. So, the President of Mexico is doing a great job. Trump’s ability to turn enemies into allies, and vice versa, fits well into tropes that forge conditional coalitions toward protecting the nation-state. The narrative of Mexicans as enemies, which predominated during Trump’s first Presidential run, was gradually replaced by the portrayal of “Mexico as an ally,” particularly during his reelection campaign. Close to the end of his mandate, he needed to reinforce the reach and impact of his early pledges (i.e., “illegals will be sent back to where they came from”). To that end, he used the same xenophobic framework of inclusion/exclusion, with which he had skillfully contraposed the Hispanic versus the undocumented Latino immigrant, for his Mexican and Central American associates. Still, while the latter became his conditional partners, their citizens and their countries’ asylum seekers continued to be portrayed as enemies of the United States. Discussion: The Coupled Effect Of Xenophobia This article inquired about former President Trump’s rhetorics of inclusion/exclusion with respect to subordinate groups, Latinos in this case. By relying on xenophobia as the main conceptual framework (see Graph 4), this study examined Trump’s public messages designed to attract the Latino vote during an extraordinary pandemic and election year. To the best of our knowledge, this is the first study to examine how Trump’s xenophobia served his aim to embrace his Latino sympathizers (i.e., the deserver of the American Dream, the defender of the nation-state, and the victim of the illegal immigrant) for the sake of upholding to United States racial and ethnic hierarchies. These findings have shown how xenophobia simultaneously ostracizes certain groups while embracing others—even if conditionally. Building on Schmitt’s “friend versus enemy” distinction, xenophobia draws an imaginary “us” against the “other” and, under Trump’s presidency, it galvanized a sector of his Latino electorate for a common crusade against an external enemy—the “illegal trespasser” who settles among us—which simultaneously includes carriers of the “Kung Fu virus,” the immigrant freeloader, and the gang member/drug dealer. A discussion of our main findings regarding Trump’s use of xenophobia in engaging the Latino electorate is in order. First, Trump’s skillful camouflaging of race, the enemy is both colorless and shapeless, is systematically deployed in his refusal to name and personalize undocumented immigrants. In all his documents and public appearances, those crossing the border from Mexico are “faceless,” with no other identity than that associated with their criminal record—typically they are drug traffickers/human smugglers and/or gang members. Trump continuously reminded his audiences of the nonvisible characteristics that make “the other” ineligible for citizenship, including the deviant features of the MS-13 border crosser and the illegal felon protected by sanctuary city laws. In line with our findings, the literature shows that while racism rests on specific phenotypical features to justify domination, xenophobia subjugates those with alleged essential immoral traits (Chiricos, Welch, and Gertz 2004; Saito 2021; Wistrich 2013). And while open racism is not generally accepted, rejection of the other (framed as a threat) is celebrated (Saito 2021). Trump’s “cloaked racism” is conspicuously illustrated in his popular self-laudatory claim: “I am the least racist person there is anywhere in the world,” as he railed against Mexican immigrants and Muslims—not on the basis of their phenotypical characteristics but on their presumed criminal nature and deviant behaviors (Kelly 2020; McIntosh and Mendoza-Denton 2020). By stressing cultural and moral difference rather than racial characteristics, xenophobia provides a screen for White supremacy as United States citizens are now contrasted against those who are deemed essentially and perennially foreign. While this phenomenon is nothing new, this was the first time that a President made the nation-state and the Republican Party openly subservient to the principles of White supremacy (Massey 2021). Second, the figure of the “illegal alien” is depicted as a powerful threat to both the White-majority and subaltern groups (i.e., the image of the Hispanic victim). Protected by sanctuary cities, the illegal immigrant not only invades the national polity but also remains among us. Therefore, the full investment of the nation-state is required to keep him in check—in Trump’s narratives, “border invaders” are consistently male. In line with our findings, studies on the Latino threat narrative point to the public construction of “the brown threat,” shared with other groups such as Middle Eastern immigrants, who allegedly bring crime and moral turpitude and are therefore underserving of the American Dream (Chavez 2013; Rivera 2014). This rhetoric has been fueled by claims that the greatness of the United States has been threatened by the invasion of illegal workers, from Mexico and other Latin American countries, who steal jobs from legitimate American citizens and commit crimes against them (Hinojosa-Ojeda and Telles 2021). Third, to be effective, xenophobia must grant the right to inclusion to particular subsets of subaltern groups. Rather than counterposing specific ethnic/racial groups against the White majority, xenophobia frames an illegitimate minority against a legal, albeit subordinated, one. Trump constructed a homogeneous Hispanic minority that is welcomed into the American Dream and respected for its hard work and family values. Contrary to unidentified “illegal immigrants,” Trump typically called his Hispanic supporters by their first and last names (e.g., Jorge or Martinez), and told poignant stories about their struggles and paths to success. This concords with the media literature that shows that naming individuals and recounting their personal trajectories is an effective tool to forge empathy and identification with public audiences (Petroff et al. 2021; Viladrich 2019). Still, as a reminder of Latinos’ perennial foreigner status, Trump made sure to frame a distance effect between a “we/us” (i.e., White patriots, Americans) and “them” (i.e., Hispanics). In this view, the plurals “they” and “them” work to stress the discrepancies between the White majority and a seemingly homogeneous Latino aggregate (Schneider and McClure 2020). Fourth, although Trump’s speeches made quite explicit his political support for Hispanics, their status as included had to be earned. As discussed in this article, the figure of the “Hispanic Border Patrol agent” is subjected to a patriotic scrutiny that grants Latinos provisional inclusion in exchange for their unconditional commitment to protecting the nation. Every time Trump praised Hispanics for their value and contributions to America’s grandeur, he reminded them of their inescapable duty to reject “illegals” and defend the homeland—a symbolic entry fee for joining the American Dream and becoming “honorary whites” (Bonilla-Silva 2004). Finally, Trump’s welcoming of representatives from some Latin American countries, as conditional foreign allies, was dependent on their endorsement of American’s racial-ethnic status quo: Mexicans paying for the wall, and Guatemalans and Salvadorans persecuting their own kind and bringing their deported nationals back home. Trump’s speeches ultimately conveyed a binary distinction between those non-Whites that are conditionally welcomed into the national polity (i.e., the Hispanic supporter and the Border Patrol agent) and those excluded from it (i.e., the illegal immigrant) due to their cultural and moral incongruency with United States values. Last, Trump’s keen maneuvering of divisional politics achieved its maximum expression when praising his allies, internal and external, for being vigilant against the nation’s enemies—namely, Latino trespassers. In its next and final section, this article draws some parallels between the study findings and Trump’s growth among the Latino electorate in 2020 toward envisioning new areas of sociological research. Conclusion: The Day After The 2020 Presidential Election This article began with a vignette recounting the author’s personal realization of Trump’s inroads among Latinos just before the United States 2020 Presidential Election. Fast forward a few years, and there is now little doubt that one of the major surprises of that election was his receiving roughly 38 percent of the Latino electorate—one of the highest shares of the Latino vote for the Republican Party in history (Igielnik et al. 2021). The former President outperformed himself, with a 10-point gain among Latinos from 2016 to 2020 (Equis 2021; Igielnik et al. 2021). Latinos’ support for Trump grew particularly in states like Texas (where he received 41 percent of the Latino vote) and Florida (45 percent), the latter not only due to the endorsement of Cubans but of LatAms, particularly in South Florida, a group that includes people from a variety of Latin American countries, such as Venezuela (Equis 2021). Scholars and policy analysts are still debating the reasons for Trump’s Latino uptick during the 2020 election. Post-election research suggests that Trump and the GOP actively campaigned for the Latino vote and made gains that cut across geography and nationality, crafting messages that resonated well with specific segments of the electorate—even if for different reasons (Equis 2021; Ocampo, Garcia-Rios, and Gutierrez 2021; Robb 2022). This phenomenon has also been credited to the relentless pro-Trump propaganda broadcast by right-wing outlets (i.e., Fox News) along with the active social media messages (i.e., WhatsApp, YouTube, and Twitter) featuring Trump, his followers, and think tanks such as “Latinos for Trump” (Louie and Viladrich 2021; Robb 2022). Although this study focused on the production of Trump’s discourses, and not its effects, some key parallels can be drawn between our findings and the results of the 2020 election. As suggested in the framing of the “Hispanic supporter,” studies have shown that Trump attracted Hispanic entrepreneurs (i.e., the “Goya type”) who ascribed to pro-family and meritocratic ideals of upward mobility, along with their deep distrust in government intervention and taxes (Garza 2021; Robb 2022). Almost counterintuitively, the economic downturn during COVID-19 allowed Trump to highlight his promise to keep the economy open—a message that resonated strongly among his Latino voters—even as immigration issues became less important. Furthermore, Biden’s cautious approach toward the pandemic left room for Trump to court the Latino electorate (Equis 2021; Ocampo et al. 2021). Trump’s Latinos, many of whom had experienced a pre-pandemic bonanza, were fearful that electing Biden would lead to more shutdowns and economic recession during an enduring health crisis (Garza 2021; Equis 2021). Despite the ongoing high unemployment in 2020, Trump continued to be seen as a successful entrepreneur committed to guaranteeing jobs and protecting traditional family values, religion, and free enterprise. His trajectory as a businessman made him particularly appealing to first-time, younger, and swing Latino voters, who saw him as best equipped to drive the country out of a pandemic-driven recession (Equis 2021; Garza 2021; Ocampo et al. 2021). In line with our findings, the literature has shown how the divisive power of Trump’s rhetoric, including his call for patriotic duty, helped split the Latino vote (Garza 2021; Robb 2022). Both the Democratic Party and the media overstated Latinos’ monolithic interests, particularly regarding progressive immigration issues, including the plight of undocumented Latinos (Gonzalez-Sobrino 2021; Ocampo et al. 2021; Robb 2022). Furthermore, Trump’s calls for militarization and national safety found a sounding board among his supporters. For instance, the “defund the police” slogan of the BLM (Black Lives Matter) movement, a direct threat to Trump’s “law and order” motto, helped move the Latino needle toward Trump among those who feared an increase in crime (Garza 2021). A final note about the scope of our findings is in order, particularly regarding the contemporary role of xenophobia in cementing ethnoracial nationalism. If, until now, xenophobia was mostly known for its exclusionary discourses and practices, this study revealed how it grants conditional inclusion to particular racial and ethnic groups that, irrespective of their skin color, pledge allegiance to the nation-state—even at the price of persecuting their own kind. Finally, by stressing the moral inferiority of the “other,” rather than their racial/ethnic characteristics, xenophobia succeeds in camouflaging racism while merging the nation’s external and internal enemy into one single entity: the undocumented Latino immigrant. Future studies should consider the emotional and intellectual resonance (i.e., framing effects) of Trump’s speeches on different Latino audiences, particularly regarding the different dimensions of the xenophobic construct (i.e., the framing of the “other” as a threat to minority populations), which may have been effective in attracting specific Latino groups. In sum, we need more data on the diverse—and even contradictory—effects of Trump’s tropes on Latino voters, including the complex ways through which xenophobic narratives help enlist minorities into neopopulist, anti-immigrant political stands. I would like to acknowledge my colleague Vivian S. Louie, with whom I began shaping my research notions on Trump’s rhetorics with respect to minority groups generally and Latinos particularly. Sara E. Grummert’s theoretical ideas and data management expertise were invaluable in helping me refine my research findings. I am also thankful to Sandra Gil Araujo and Carolina Rosas, along with the members of their research group at the Gino Germani Institute (University of Buenos Aires, Argentina), for their helpful comments to an earlier version of this paper. As always, Slava Faybysh provided superb editorial assistance. I am particularly grateful to the Journal’s co-editor B. Brian Foster for being in constant communication with me (and for his kindness) during the submission and revision process, and to managing editor Donald R. Guillory for his timely copyediting assistance. The final version of this paper owes a debt of gratitude to the anonymous reviewers whose comments greatly helped improve it. Author Biography Anahí Viladrich is an interdisciplinary social science scholar and public health specialist whose work focuses on international migration, Latinos in the United States, health disparities, gender, and culture. The author of more than 60 peer-reviewed publications, this article builds upon her current research trajectory, grants, and publications, geared toward understanding the role of public discourse in shaping mainstream representations of immigrants in the United States and overseas. She is currently a full professor in the Department of Sociology and is affiliated with the Department of Anthropology at Queens College, the Graduate Center (Sociology) and the Graduate School of Public Health and Health Policy of the City University of New York (CUNY). Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and publication of this article: This project was financed by a PSC-CUNY Award (65367-00 53). ORCID iD: Anahí Viladrich https://orcid.org/0000-0003-4422-7429 ==== Refs References al Gharbi M . 2018. “Race and the Race for the White House: On Social Research in the Age of Trump.” The American Sociologist 49 (4 ):496–519. Alamillo R. 2019. “Hispanics Para Trump?: Denial of Racism and Hispanic Support for Trump.” Du Bois Review 16 (2 ):457–87. American Immigration Council. 2021. “The Migrant Protection Protocols.” Factsheet, October 6. Retrieved November 6 (https://www.americanimmigrationcouncil.org/research/migrant-protection-protocols). Baker J. O. Cañarte D. Day L. E. 2018. “Race, Xenophobia, and Punitiveness Among the American Public.” The Sociological Quarterly 59 (3 ):363–83. Basler C. 2008. “White Dreams and Red Votes: Mexican Americans and the Lure of Inclusion in the Republican Party.” Ethnic and Racial Studies 31 (1 ):123–66. Bergad L. W. Miranda L. A. Jr. 2021. “Latino Voter Registration and Participation Rates in the 2020 Presidential Election.” Latino Data Project, Report No. 94, NYC: CLACS. Bonilla-Silva E. 2004. “From Bi-racial to Tri-racial: Towards a New System of Racial Stratification in the USA.” Ethnic and Racial Studies 27 (6 ):931–50. Buckley G. L. 2002. American Patriots: The Story of Blacks in the Military from the Revolution to Desert Storm. New York: Random House. Cadava G. 2020. The Hispanic Republican: The Shaping of an American Political Identity, from Nixon to Trump. New York: HarperCollins. Canizales S. L. Vallejo J. A. 2021. “Latinos & Racism in the Trump Era.” Daedalus 150 (2 ):150–64. Cech E. A. 2017. “Rugged Meritocratists: The Role of Overt Bias and the Meritocratic Ideology in Trump Supporters’ Opposition to Social Justice Efforts.” Socius 3 :1–20. Cervantes A. G. Menjívar C. 2018. “‘Bad Hombres’: The Effects of Criminalizing Latino Immigrants Through Law and Media in the Rural Midwest.” Migration Letters 15 (2 ):182–96. Chavez L. R. 2013. The Latino Threat: Constructing Immigrants, Citizens, and the Nation, Second Edition. Stanford, CA: Stanford University Press. Chiricos T. Welch K. Gertz M. 2004. “Racial Typification of Crime and Support for Punitive Measures.” Criminology 42 (2 ):358–90. Collingwood L. O’Brien B. G. 2019. Sanctuary Cities: The Politics of Refuge. Oxford, England: Oxford University Press. Corral Á. J. Leal D. L. 2020. “Latinos Por Trump? Latinos and the 2016 Presidential Election.” Social Science Quarterly 101 (3 ):1115–31. Dávila A. 2012. Latinos, Inc.: The Marketing and Making of a People. Berkeley, CA: University of California Press. Dedoose. 2021. Web Application for Managing, Analyzing, and Presenting Qualitative and Mixed Method Research Data. Version 9.0.17. Los Angeles, CA: Sociocultural Research Consultants, LLC. Equis. 2021. “Post-mortem: Part 2.” The American Dream Voter, December 14. (https://equisresearch.medium.com/post-mortem-part-two-the-american-dream-voter-66dd6f673d1e). Fearnow B. 2019. “Who Are ‘Angel Moms?’ Trump Jr, RNC Tout Victims’s Families to Promote Border Wall.” Newsweek. January 14. (https://www.newsweek.com/angel-moms-illegal-immigrants-killed-donald-trump-jr-border-wall-fox-news-1290461). Fernandez L. 2013. “Nativism and Xenophobia.” Pp. 1–7 in The Encyclopedia of Global Human Migration, edited by Ness Immanuel Bellwood Peter . Chichester, England: Wiley-Blackwell. Filindra A. Kolbe M. 2022. “Latinx Identification with Whiteness: What Drives It, and What Effects Does It Have on Political Preferences?” Social Science Quarterly 103 (6 ):1424–39. Foster C. H. 2017. “Anchor Babies and Welfare Queens: An Essay on Political Rhetoric, Gendered Racism, and Marginalization.” Women, Gender, and Families of Color 5 (1 ):50–72. Galbraith Q. Callister A. 2020. “Why Would Hispanics Vote for Trump? Explaining the Controversy of the 2016 Election.” Hispanic Journal of Behavioral Sciences 42 :77–94. Garza D. 2021. “Policies, Not Personality: The Defining Characteristic of the 2020 Latino Vote.” Harvard Journal of Hispanic Policy 33 :16–26. Glaser B. G. Strauss A. L. 2017. Discovery of Grounded Theory: Strategies for Qualitative Research. New York: Routledge. Gonzalez-Sobrino B. 2020. “Searching for the Sleeping Giant: Racialized News Coverage of Latinos Pre-2020 Elections.” Sociological Forum 35 (S1):1019–39. Gonzalez-Sobrino B. 2021. “The Nuances of the “Latino Vote”: Toward the Unpacking of the Panethnic.” Sociological Forum 36 (2 ):548–49. Gonzalez M. M. 2016. “The Coding Manual for Qualitative Research: A Review.” Qualitative Report 21 (8 ):1546–48. Gostin L. O. 2020. “The Great Coronavirus Pandemic of 2020—7 Critical Lessons.” JAMA 324 (18 ):1816–17. Gover A. R. Harper S. B. Langton L. 2020. “Anti-Asian Hate Crime During the COVID-19 Pandemic: Exploring the Reproduction of Inequality.” American Journal of Criminal Justice 45 (4 ):647–67. Haywood J. M. 2017. “Anti-Black Latino Racism in an Era of Trumpismo.” International Journal of Qualitative Studies in Education 30 (10 ):957–64. Henley K. Warren P. 2020. “Populism, American Nationalism and Representative Democracy.” Pp. 193–207 in Democracy, Populism, and Truth (AMINTAPHIL: The Philosophical Foundations of Law and Justice Book 9) edited by Navin M. C. Nunan R. Switzerland, Springer. Heuman A. González A. 2018. “Trump’s New Essentializing Rhetoric: Racial Identities and Dangerous Liminalities.” Journal of Intercultural Communication Research 47 (4 ):326–42. Hinojosa-Ojeda R. Telles E. 2021. The Trump Paradox. Berkeley, CA: University of California Press. Hirschfeld David J. Baker P. 2019. “How the Border Wall Is Boxing Trump in.” The New York Times. Retrieved October 10, 2021. (https://www.nytimes.com/2019/01/05/us/politics/donald-trump-border-wall.html). Hochschild J. L. 1995. Facing Up to the American Dream: Race, Class and the Soul of the Nation. Princeton, NJ: Princeton University Press. Huber L. P. Lopez C. B. Malagon M. C. Velez V. Solorzano D. G. 2008. “Getting Beyond the Symptom, Acknowledging the Disease: Theorizing Racist Nativism.” Contemporary Justice Review 11 (1 ):39–51. Igielnik R. Keeter S. Hartig H. 2021. “Behind Biden’s 2020 Victory.” Pew Research Center Report, June 20. Retrieved October 23, 2021 (https://www.pewresearch.org/politics/2021/06/30/behind-bidens-2020-victory/). Kellner D. 2017. American Horror Show: Election 2016 and the Ascent of Donald J. Trump. New York: Springer. Kelly C. R. 2020. “Donald J. Trump and the Rhetoric of White Ambivalence.” Rhetoric and Public Affairs 23 (2 ):195–223. Kim D. H. Sundstrom R. R. 2014. “Xenophobia and Racism.” Critical Philosophy of Race 2 (1 ):20–45. Lamont M. Park B. Y. Ayala-Hurtado E. 2017. “Trump’s Electoral Speeches and His Appeal to the American White Working Class.” The British Journal of Sociology 68 :S153–80. Louie V. Viladrich A. 2021. “Divide, Divert, and Conquer: Deconstructing the Presidential Framing of White Supremacy in the COVID-19 Era.” Social Sciences 10 (8 ):280. Mannarini T. Salvatore S. 2020. “The Politicization of Otherness and the Privatization of the Enemy: Cultural Hindrances and Assets for Active Citizenship.” Human Affairs 30 (1 ):86–95. Marin L. 2016. “Something to Die For. The Individual as Interruption of the Political in Carl Schmitt’s the Concept of the Political.” Revue Roumaine de Philosophie 60 (2 ):311–25. Massey D. S. 2021. “What Were the Paradoxical Consequences of Militarizing the Border with Mexico?” Pp. 32-46 in The Trump Paradox, edited by Hinojosa-ojeda R. Tells E. Berkeley, CA: University of California Press. McIntosh J. eds. 2020. Language in the Trump Era: Scandals and Emergencies. Cambridge, England: Cambridge University Press. Menjívar C. 2016. “Immigrant Criminalization in Law and the Media: Effects on Latino Immigrant Workers’ Identities in Arizona.” American Behavioral Scientist 60 (5–6 ):597–616. Ngai M. M. 2014. Impossible Subjects: Illegal Aliens and the Making of Modern America. Princeton, NJ: Princeton University Press. Ocampo A. X. Garcia-Rios S. I. Gutierrez A. E. 2021. “Háblame De Tí: Latino Mobilization, Group Dynamics and Issue Prioritization in the 2020 Election.” The Forum 18 (4 ):531–58. Ott B. L. Dickinson G. 2019. The Twitter Presidency: Donald J. Trump and the Politics of White Rage. New York: Routledge. Petroff A. Viladrich A. Parella S. 2021. “Framing Inclusion: The Media Treatment of Irregular Immigrants’ Right to Health Care in Spain.” International Journal of Intercultural Relations 82 :135–44. Rivera C. 2014. “The Brown Threat: Post-9/11 Conflations of Latina/os and Middle Eastern Muslims in the US American Imagination.” Latino Studies 12 (1 ):44–64. Robb J. 2022. Political Migrants: Hispanic Voters on the Move—How America’s Largest Minority Is Flipping Conventional Wisdom on Its Head. Arlington, VA: NumbersUSA Education & Research Foundation. Saito N. T. 2021. “Why Xenophobia?” La Raza Law Journal 31 :1–23. Sanchez G. R. Gomez-Aguinaga B. 2017. “Latino Rejection of the Trump Campaign: How Trump’s Racialized Rhetoric Mobilized the Latino Electorate as Never Before.” Aztlan 42 (2 ):165–81. Schaefer D. O. 2020. “Whiteness and Civilization: Shame, Race, and the Rhetoric of Donald Trump.” Communication and Critical/Cultural Studies 17 (1 ):1–18. Schmitt C. 2007. The Concept of the Political. Expanded ed. Chicago, IL: University of Chicago Press. Schneider U. McClure K. K. 2020. “I’m Doing Great with the Hispanics. Nobody Knows It.” Pp. 130–152 in Linguistic Inquiries into Donald Trump’s Language: From “Fake News” to “Tremendous Success,” edited by Schneider U. Eitelmann M. New York: Bloomsbury Academic. Silber Mohamed H. Farris E. M. 2020. “‘Bad Hombres’? An Examination of Identities in US Media Coverage of Immigration.” Journal of Ethnic and Migration Studies 46 (1 ):158–76. Smith D. N. 2019a. “Authoritarianism Reimagined: The Riddle of Trump’s Base.” The Sociological Quarterly 60 (2 ):210–23. Smith S. M. 2019b. “The Trump Administration’s Third Country Transit Bar.” Georgetown Immigration Law Journal 34 :539–43. Sosa L. 2009. “Politics and the Latino Future: A Republican Dream.” Pp. 115–124 in Latinos and the Nation’s Future, edited by Cisneros H. Rosales J. Houston, Texas: Arte Público Press. Stelter B. 2020. Hoax: Donald Trump, Fox News, and the Dangerous Distortion of Truth. New York: Simon and Schuster. Sundstrom R. R. 2013. “Sheltering Xenophobia.” Critical Philosophy of Race 1 (1 ):68–85. Viladrich A. 2019. “We Cannot Let Them Die”: Framing Compassion Towards the Health Needs of Unauthorized Immigrants in the United States (United States).” Qualitative Health Research 29 (10 ):1447–60. Viladrich A. 2021. “Sinophobic Stigma Going Viral: Addressing the Social Impact of COVID-19 in a Globalized World.” American Journal of Public Health 111 (5 ):876–80. Wimmer A. 1997. “Explaining Xenophobia and Racism: A Critical Review of Current Research Approaches.” Ethnic and Racial Studies 20 (1 ):17–41. Wistrich R. S. , ed. 2013. Demonizing the Other: Antisemitism, Racism and Xenophobia. New York: Routledge. Yang Y. Bennett L. 2022. “Interactive Propaganda: How Fox News and Donald Trump Co-produced False Narratives About the COVID-19 Crisis.” Pp. 83–100 in Political Communication in the Time of Coronavirus, edited by Van Aelst P. Brumler J. G. New York: Routledge.
PMC010xxxxxx/PMC10290933.txt	==== Front Clin Pediatr (Phila) Clin Pediatr (Phila) CPJ spcpj Clinical Pediatrics 0009-9228 1938-2707 SAGE Publications Sage CA: Los Angeles, CA 37326038 10.1177/00099228231177792 10.1177_00099228231177792 Resident Rounds A Rapidly Expanding Chest Wall Mass in an Adolescent With COVID-19 Klair Nate MD 1 Ireland Malia DVM, MPH 2 https://orcid.org/0000-0002-2108-9433 Schleiss Mark R. MD 3 1 Department of Pediatrics, Residency Training Program, University of Minnesota, Minneapolis, MN, USA 2 Zoonotic Diseases Unit, Minnesota Department of Health, St. Paul, MN, USA 3 Division of Pediatric Infectious Diseases, Department of Pediatrics, University of Minnesota, Minneapolis, MN, USA Mark R. Schleiss, Division of Pediatric Infectious Diseases, Department of Pediatrics, University of Minnesota, 2001 6th Street SE, Minneapolis, MN 55455, USA. Email: schleiss@umn.edu 16 6 2023 16 6 2023 00099228231177792© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. edited-statecorrected-proof typesetterts1 ==== Body pmcEducational Objectives Recognize the microbiological differential diagnosis of empyema necessitans, a clinical entity in which an empyema extends out of the pleural space and into the chest wall, producing infection and concomitant swelling in the surrounding soft tissues. Be aware of the immunomodulatory effects of acute SARS-CoV-2 disease that have been recognized during the COVID pandemic, and the potential role of such virally-induced immune modulation in predisposing children to severe infections with the endemic mycoses. Case Report A previously healthy 15-year-old boy presented to the emergency department with a 1-week history of an expanding, tender chest mass near the upper left sternal border and 1 to 2 days of mild, dry cough. He had a recent household exposure to family members with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. He had received his usual childhood vaccines and 2 SARS-CoV-2 mRNA vaccines. On examination, he was afebrile with normal vital signs. His oral dentition was noted to be poor, with extensive dental caries present, particularly prominent in the posterior mandibular and maxillary molars. His breath sounds were diminished over the left hemithorax. A chest wall mass was noted over the anterior left chest with a raised area of induration that had a diameter of approximately 5 cm. The mass was firm and rubbery in consistency, but without frank fluctuance. In addition to a positive SARS-CoV-2 polymerase chain reaction (PCR) nasal swab, a leukocytosis of 14.3 e3/mL was noted. The differential leukocyte count demonstrated 71% neutrophils; 16% lymphocytes (absolute lymphocyte count, 2.28 e3/mL); 11% monocytes; and 1% eosinophils. His C-reactive protein (CRP) was elevated at a level of 74.9 mg/dL. The erythrocyte sedimentation rate was 64 mm/h (normal range, 0-15 mm/h). Computed tomography (CT) imaging showed a dense consolidation in the left upper lobe extending into the pleural cavity and the chest wall, with 2 well-defined intramuscular pectoralis major abscesses identified both by chest radiograph (Figure 1A) and by chest CT (Figure 1B, C). Figure 1. Radiographic findings in patient with empyema necessitans secondary to Blastomyces gilchristii infection. (A) Dense consolidation in the left upper lobe noted on chest radiograph (arrow, a). Right-ward deviation of trachea is noted. (B) Demonstration of transpleural extension of disease (arrow, a) with a well-defined intramuscular pectoralis major abscess noted by CT scan (sagittal view; arrow, b). (C) Transverse CT view similarly demonstrates pectoralis major abscess with extension of infection into chest wall (arrow). Abbreviation: CT, computed tomography. Infectious diseases consultation was obtained. His medical history indicated no previous hospitalizations, no history of unusual infections, and no family history of primary immunodeficiencies. Additional history revealed that breeding colonies of turtles and several snakes were present in the household, although the patient had minimal exposure to these reptiles. There were also 2 fish, 2 dogs and 1 cat in the household. Based on the imaging and physical examination findings, a diagnosis of empyema necessitans was made. In light of his history of household exposure to reptiles and the finding of poor oral dentition on examination, initially a broad differential diagnosis was considered (Table 1). Table 1. Microbiology Etiologies Associated With Empyema Necessitans.1-8 Reported etiologies of empyema necessitans Bacterial Streptococcus pneumoniae Staphylococcus aureus Mycobacterium tuberculosis Nocardia spp Actinomycosis spp Fusobacterium spp Salmonella spp Aggregatibacter actinomycetemcomitans Porphyromonas gingivalis Fungal Histoplasmosis Aspergillosis Blastomycosis Mucormycosis Coccidioidomycosis Other Entamoeba histolytica Paragonimiasis Strongyloides spp Toxocariasis Echinococcus spp Discussion Hospital Course The infectious diseases consultative service offered the opinion that the patient had pneumonia and accompanying parapneumonic effusion that had progressed to frank empyema necessitans. Surgical intervention was recommended. After consultation with the pediatric surgery service, the decision was made to drain the chest wall abscess under the guidance of the interventional radiology service. A 10F left upper anterior chest wall subcutaneous drain was placed, concomitant with drainage of ~20 mL of thick, purulent, white fluid. No organisms were identified by the gram stain, and he was treated empirically with ceftriaxone. Flow cytometry analysis of the fluid revealed a preponderance of T cells, with no evidence of aberrant immunophenotype. Cytology demonstrated no evidence for malignant cells, with Grocott methenamine silver and Periodic acid-Schiff stains positive for budding yeast forms. Fungal antigen and antibody tests recommended by the infectious disease consultation service were positive for Blastomycosis dermatitidis. The urine antigen was positive at a level of 1.44 ng/mL, and the serum antigen was positive at a value of 0.22 ng/mL. Serum antibody studies demonstrated a positive enzyme immunoassay (EIA) test, with an index value of 8.1. Immunodiffusion and serum precipitin antibodies for B dermatitidis were also positive. Upon further questioning, the patient was noted to have gone camping in the woods in the Detroit Lakes region of Minnesota during the summer months. Approximately 2 weeks after collection, a fungal culture of abscess fluid was reported positive for a Blastomyces species. Subsequent DNA sequence analysis of this isolate, performed at the Centers for Disease Control and Prevention, demonstrated that his infection was due to an organism highly related to B dermatitidis, specifically, Blastomyces gilchristii. 9 After commencement of antifungal therapy, the patient’s symptoms improved. Additionally, immunological studies were recommended by the infectious diseases consultative service but were declined by the primary care hospitalist team. Although amphotericin B therapy is recommended by the Infectious Diseases Society of America for “severe or moderately severe disease” due to blastomycosis, 10 with his limited symptoms, the primary care team elected, after consultation with the family, to discharge the patient on oral itraconazole therapy. A therapeutic level of itraconazole (trough level, 1.4 mg/mL) was documented on outpatient follow-up. Following 6 months of therapy, the patient was asymptomatic, and there was no evidence on physical examination of chest wall swelling. Discussion of Case and Relevant Literature The patient described in this report presented with empyema necessitans (also known as empyema necessitatis), a rare complication of pulmonary/pleural space infections in which infected fluid dissects spontaneously from the pleural cavity into the contiguous chest wall soft tissue space. Although first described in the mid-nineteenth century, 11 it has been less commonly encountered in the post-antibiotic era. Empyema necessitans usually presents as a chest wall mass, although infection may extend into the flank, groin, or thigh. 1 The infectious etiology most classically associated with this diagnosis is Mycobacterium tuberculosis, 2 although many pathogens have been described (Table 1).3-7 These etiologies include the endemic mycoses histoplasmosis, blastomycosis, and coccidioidomycosis. 4 Less common etiologies can include parasitic infections such as Paragonimus westermani. 5 Empyema necessitans can also be a complication of poor dental health, including periodontitis. 6 Cases may also be polymicrobial in nature. 7 Patients may manifest disease with fever and severe systemic toxicity, although the evolution and presentation can also be indolent in nature. 7 Surgical debridement and appropriate antimicrobial therapy represent the key elements in the management of empyema necessitans. 8 Of note, the patient described in this report tested positive for COVID-19 at the time of his hospital admission with blastomycosis. We hypothesize that the patient’s serious Blastomyces infection was potentially causally related to his antecedent SARS-CoV-2 infection. 12 It is clear that immunodeficient hosts, including individuals with primary immune deficiency, are also predisposed to fungal infections. Impaired cell-mediated immunity, neutrophil dysfunction, and C3 deficiency are typical inborn errors of immunity (IEI) that can predispose patients to such infections. The patient described in this report had a normal leukocyte count and differential. Other studies, such as quantitative serum immunoglobulin levels, lymphocyte phenotype analyses, evaluation of lymphocyte proliferative responses to phytohemagglutinin, neutrophil functional tests, and C3 levels should be considered in the evaluation of a patient with an invasive mycotic infection. Additional immunological studies were suggested in this case, but were deferred by the primary care (hospitalist) team, given the patient’s negative medical history for other serious infections. In support of the hypothesis that COVID-19 disease may have predisposed this patient to a serious Blastomyces infection, it has become increasingly clear that SARS-CoV-2 infection has a substantial impact on the human immune response. Type 1 interferons have been shown to play a critical role in innate immune responses to COVID-19 disease.13,14 SARS-CoV-2 also directly affects the host immune response, including induction of dysregulation of genes involved in B-cell and T-cell responses15,16 and results in changes in innate immunity, including perturbation of natural killer (NK) cell-mediated activity. 17 There is evidence that, for patients with COVID-19 disease, there is an increased risk of latent tuberculous infection progressing to active tuberculosis (TB). 18 COVID-19 disease is also associated with reactivation of latent cytomegalovirus infection. 19 A striking example of the impact of SARS-CoV-2 on the emergence of secondary fungal infection is the association with severe mucormycosis that was observed in India, during the height of the COVID pandemic. Over 30 000 cases have been reported to date, 20 with approximately 15% attributable mortality due specifically to Mucor infection. Although the intercurrent presence of diabetes and the concomitant use of steroids both contributed substantially to the mucormycosis disease burden, 21 COVID-19-induced T-cell exhaustion was also observed, 22 and this immunomodulatory effect of the SARS-CoV-2 virus likely increased the susceptibility of these patients to experience severe fungal disease with Mucor. A similar enhancement of risk for other fungal infections, triggered by intercurrent SARS-CoV-2 infection, has been noted for histoplasmosis, 23 cryptococcosis, 24 coccidioidomycosis, 25 and, as for the patient described in the current report, blastomycosis. 26 When considering a potential fungal superinfection in a patient with COVID-19 disease, the endemicity of fungal pathogens where the patient lives or has traveled is an important historical clue. In this instance, blastomycosis was an unsurprising finding when the etiology for our patient’s infection was determined. Blastomycosis is a relatively common diagnosis in Minnesota when compared with other states; in fact, Minnesota and Wisconsin account for 75% of all blastomycosis cases reported annually in the United States. 27 Outdoor activities and soil exposure are associated with infection throughout Minnesota, and at least one-third of Minnesotans engaged in in-state recreational activities that acquire this infection are exposed to the pathogen in a location other than their home county. 28 Our patient’s history of a summer camping trip, where he explored wooded areas and engaged in outdoor recreation, was compatible with these known features of blastomycosis epidemiology. We hypothesize that this patient’s Blastomyces infection, which was indolent and subclinical in nature, transformed into an aggressive and invasive infection, with spread to the chest wall, driven by the immunomodulatory effects of his acute COVID-19 infection. It was of interest to note that, upon molecular testing, the causative agent for the patient’s illness was B gilchristii, and not B dermatitidis. The use of whole-genome sequencing and single-nucleotide polymorphism (SNP) analyses to evaluate blastomycosis molecular epidemiology in Minnesota identified B gilchristii as the causative agent in 2 recent outbreak investigations. 29 B gilchristii has a smaller geographic range, encompassing Canada and the northern United States, than does B dermatitidis, which is found throughout the Great Lakes, the Ohio and Mississippi River valleys, and the St. Lawrence River. 29 Final Diagnoses COVID-19 infection. Pulmonary and pleural space infection due to blastomycosis, with associated empyema necessitans. Conclusion Empyema necessitans was a common complication of pneumonia in the pre-antibiotic era, but the diagnosis is rarely encountered today. This case report highlights a unique presentation of a minimally symptomatic patient with an indolent B gilchristii infection who had an acute and significant clinical deterioration after acquiring COVID-19 infection, with evolution of a fungal pneumonia that extended into his chest wall. We hypothesize that the acute SARS-CoV-2 infection, through dysregulation of host immunity (most likely due to suppression of T-cell responses), contributed to acceleration of his underlying Blastomyces infection. Invasive fungal infections are reported complications of COVID-19 disease, and clinicians should be mindful to consider this association for patients who live in endemic regions. This case illustrates how the immunomodulatory effects of COVID-19 may facilitate emergence of invasive disease for what would otherwise be a subclinical, endemic infection such as blastomycosis. In addition to the potential contributory effect of intercurrent COVID-19 in this patient, the classic IEI should be considered in such a situation. These usually present in early childhood, but in exceptional cases these do not appear until adolescence or adulthood. The patient with late-onset IEI presents with a mostly healthy childhood and often receives little medical attention. Such atypical cases may be diagnosed by using the European Society for Immunodeficiency criteria. 30 Detailed immunological examination should be considered in such settings, to avoid overlooking underlying immunodeficiency. However, immune deficiency need not be invoked in all instances of blastomycosis, since the ubiquitous and endemic nature of blastomycosis in Minnesota results in many clinically symptomatic cases that occur in immunocompetent individuals. The complication of this endemic mycosis that we describe in this case, empyema necessitans, indicates that blastomycosis should be considered in a child presenting with unexplained chest wall swelling, particularly if an underlying pulmonary/pleural infectious process is present. Diverse etiologies are described, but among the endemic mycoses, blastomycosis is a particularly important causal agent for empyema necessitans to consider in the upper Midwestern United States. Due to climate change, the geographic range of endemic mycoses is expanding in the United States, making it likely that increased numbers of cases of infection with these pathogens will be encountered in the future. 31 Author Contributions NK and MRS: Conceived of the paper and wrote the first draft of the manuscript. MI: Contributed data and performed data analyses and participated in critical review and editing of the manuscript. We thank the UMN Infectious Diseases Diagnostic Laboratory (Dr Patricia Ferrieri) and the Minnesota Department of Health Public Health Laboratory for assistance in identification of the Blastomyces isolate. Authors’ Note: This case was originally presented as a poster (unpublished) at the University of Minnesota Department of Pediatrics annual Pediatric Research, Education and Scholarship (“PRESS”) symposium, April 1, 2022 (“First Place” award, resident category). The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: MDH’s work was supported in part by a cooperative agreement with the Centers for Disease Control and Prevention as part of the Epidemiology and Laboratory Capacity for Infectious Diseases Program (U50/CK000371). Ethics Statement: The authors gratefully acknowledge the permission of this patient’s family to share the information contained in this case report. Written informed consent for publication of this patient’s information and consent for publication of corresponding images was provided by the patient’s mother. Written assent was also directly provided by the patient described in this case report. Non-Anonymized Ethical Statement With Ethics Clearance: In addition to written consent provided by the parent and written assent from the patient, who is the subject of this case report, after review with the University of Minnesota Institutional Review Board (IRB), the determination was made that the content of this manuscript did not constitute research and with the documented signed permissions, the determination was further made that this case report submission could be considered successfully endorsed for ethics clearance. ORCID iD: Mark R. Schleiss https://orcid.org/0000-0002-2108-9433 ==== Refs References 1 Haddad CJ Sim WK. Empyema necessitatis. Am Fam Physician. 1989;40 (4 ):149-152. 2 Chang SH Choe J Ghandehari S , et al . Rapidly growing Mycobacterium tuberculosis in the form of empyema necessitans: a case report. J Intensive Care Med. 2022;37 (8 ):1112-1116. doi:10.1177/08850666211044100.34515566 3 Yauba MS Ahmed H Imoudu IA Yusuf MO Makarfi HU. Empyema necessitans complicating pleural effusion associated with Proteus species infection: a diagnostic dilemma. Case Rep Pediatr. 2015;2015 :108174. doi:10.1155/2015/108174.25893125 4 Rendon A Rendon RA Bauerle O. Empyema necessitatis: unique presentation in a coccidioidomycosis case. Am J Respir Crit Care Med. 2015;191 (8 ):964-966. doi:10.1164/rccm.201411-1973LE.25876208 5 Norton LE Curtis SN Goldman JL. A 9-year-old boy with a chest mass and eosinophilia. J Pediatric Infect Dis Soc. 2016;5 (4 ):476-479. doi:10.1093/jpids/piw055.27856771 6 Tanaka A Kogami M Nagatomo Y , et al . Subcutaneous abscess due to empyema necessitans caused by Porphyromonas gingivalis in a patient with periodontitis. IDCases. 2022;27 :e01458. doi:10.1016/j.idcr.2022.e01458. 7 Vertelney H Lin-Martore M. A baffling bump: a case report of an unusual chest wall mass in a pediatric patient. Clin Pract Cases Emerg Med. 2021;5 (3 ):316-319. doi:10.5811/cpcem.2021.3.51958.34437037 8 Akgul AG Orki A Orki T Yuksel M Arman B. Approach to empyema necessitatis. World J Surg. 2011;35 (5 ):981-984. doi:10.1007/s00268-011-1035-5.21404081 9 Brown EM McTaggart LR Zhang SX , et al . Phylogenetic analysis reveals a cryptic species Blastomyces gilchristii, sp. nov. within the human pathogenic fungus Blastomyces dermatitidis. PLoS One. 2013;8 (3 ):e59237. doi:10.1371/journal.pone.0059237. 10 Chapman SW Dismukes WE Proia LA , et al . Clinical practice guidelines for the management of blastomycosis: 2008 update by the Infectious Diseases Society of America. Clin Infect Dis. 2008;46 (12 ):1801-1812. doi:10.1086/588300.18462107 11 Moore W. Clinical lecture on pleurisy with effusion in children; displacement of the heart; empyema “necessitatis”; superficial tumour; recovery. Br Med J. 1869;1 (418 ):4. doi:10.1136/bmj.1.418.4. 12 Durdevic M Durdevic D Baniulis D , et al . Severe pulmonary blastomycosis in a patient with a COVID-19-positive test: a dangerous anchoring bias. Chest. 2022;162 (4 ):A292. doi:10.1016/j.chest.2022.08.224. 13 Bastard P Rosen LB Zhang Q , et al . Autoantibodies against type I IFNs in patients with life-threatening COVID-19. Science. 2020;370 (6515 ):eabd4585. doi:10.1126/science.abd4585. 14 Snell LM McGaha TL Brooks DG. Type I interferon in chronic virus infection and cancer. Trends Immunol. 2017;38 (8 ):542-557. doi:10.1016/j.it.2017.05.005.28579323 15 Zhou Y Liao X Song X , et al . Severe adaptive immune suppression may be why patients with severe COVID-19 cannot be discharged from the ICU even after negative viral tests. Front Immunol. 2021;12 :755579. doi:10.3389/fimmu.2021.755579.34867988 16 Davitt E Davitt C Mazer MB , et al . COVID-19 disease and immune dysregulation. Best Pract Res Clin Haematol. 2022;35 (3 ):101401. doi:10.1016/j.beha.2022.101401.36494149 17 Ryan FJ Hope CM Masavuli MG , et al . Long-term perturbation of the peripheral immune system months after SARS-CoV-2 infection. BMC Med. 2022;20 (1 ):26. doi:10.1186/s12916-021-02228-6.35027067 18 Noori MAM Younes I Latif A , et al . Reactivation of tuberculosis in the setting of COVID-19 infection. Cureus. 2022;14 (3 ):e23417. doi:10.7759/cureus.23417. 19 Gatto I Biagioni E Coloretti I , et al . Cytomegalovirus blood reactivation in COVID-19 critically ill patients: risk factors and impact on mortality. Intensive Care Med. 2022;48 (6 ):706-713. doi:10.1007/s00134-022-06716-y.35583676 20 Prakash S Kumar A. Mucormycosis threats: a systemic review. J Basic Microbiol. 2023;63 (2 ):119-127. doi:10.1002/jobm.202200334.36333107 21 Al-Tawfiq JA Alhumaid S Alshukairi AN , et al . COVID-19 and mucormycosis superinfection: the perfect storm. Infection. 2021;49 (5 ):833-853. doi:10.1007/s15010-021-01670-1.34302291 22 Dandu H Kumar M Malhotra HS , et al . T-cell dysfunction as a potential contributing factor in post-COVID-19 mucormycosis. Mycoses. 2023;66 (3 ):202-210. doi:10.1111/myc.13542.36305225 23 Toscanini MA Barberis F Benedetti MF , et al . Detection of anti-Histoplasma capsulatum antibodies and seroconversion patterns in critically ill patients with COVID-19: an underdiagnosed fungal entity complicating COVID-19? Med Mycol. 2022;60 (3 ):myac012. doi:10.1093/mmy/myac012. 24 Cafardi J Haas D Lamarre T Feinberg J. Opportunistic fungal infection associated with COVID-19. Open Forum Infect Dis. 2021;8 (7 ):ofab016. doi:10.1093/ofid/ofab016. 25 Krauth DS Jamros CM Rivard SC Olson NH Maves RC. Accelerated progression of disseminated coccidioidomycosis following SARS-CoV-2 infection: a case report. Mil Med. 2021;186 (11-12 ):1254-1256. doi:10.1093/milmed/usab132.33826724 26 Nasim R Prasad A Nasim H. Postpartum COVID-19 complicated by Blastomycosis infection. Chest. 2021;160 (4 ):A274. doi:10.1016/j.chest.2021.07.282. 27 Smith DJ Williams SL , Endemic Mycoses State Partners Group, et al. Surveillance for coccidioidomycosis, histoplasmosis, and blastomycosis—United States, 2019. MMWR Surveill Summ. 2022;71 (7 ):1-14. doi:10.15585/mmwr.ss7107a1. 28 Ireland M Klumb C Smith K Scheftel J. Blastomycosis in Minnesota, USA, 1999-2018. Emerg Infect Dis. 2020;26 (5 ):866-875. doi:10.3201/eid2605.191074.32310071 29 Bagal UR Ireland M Gross A , et al . Molecular epidemiology of Blastomyces gilchristii clusters, Minnesota, USA. Emerg Infect Dis. 2022;28 (9 ):1924-1926. doi:10.3201/eid2809.220392.35997504 30 de Vries E , European Society for Immunodeficiencies (ESID) members. Patient-centred screening for primary immunodeficiency, a multi-stage diagnostic protocol designed for non-immunologists: 2011 update. Clin Exp Immunol. 2012;167 (1 ):108-119. doi:10.1111/j.1365-2249.2011.04461.x.22132890 31 Mazi PB Sahrmann JM Olsen MA , et al . The geographic distribution of dimorphic mycoses in the United States for the modern era. Clin Infect Dis. 2023;76 :1295-1301. doi:10.1093/cid/ciac882.36366776
PMC010xxxxxx/PMC10290935.txt	==== Front J Aging Health J Aging Health spjah JAH Journal of Aging and Health 0898-2643 1552-6887 SAGE Publications Sage CA: Los Angeles, CA 37329336 10.1177_08982643231179869 10.1177/08982643231179869 Aging and Resilience in the Americas: Mexico and the United States Psychological Coping and Resilience to COVID-19 in Mexico, Evidence From a National Sample https://orcid.org/0000-0001-9109-381X Montero-López Lena María PhD 1 Alonso-Reyes Pilar PhD 2 Montes-de-Oca Zavala Verónica PhD 3 https://orcid.org/0000-0002-3902-5585 Zamudio Alejandro BSc 1 Guajardo Darío PhD 1 1 Facultad de Psicología, 7180 Universidad Nacional Autónoma de México , Ciudad, México 2 Facultad de Ciencias, 7180 Universidad Nacional Autónoma de México , Ciudad, México 3 Instituto de Investigaciones Sociales, 7180 Universidad Nacional Autónoma de México , Ciudad, México María Montero-López Lena, Facultad de Psicología, Universidad Nacional Autónoma de México, División de Estudios de Posgrado e Investigación, Lab. de Ecología Social y Desarrollo Comunitario, Av. Universidad 3004, Edif. D-207, Ciudad de México, C.P. 04510. Email: monterol@unam.mx 17 6 2023 17 6 2023 08982643231179869© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Objective We report the effect of COVID-19 confinement on some psychological variables associated with mental health: Stress perception, types of coping strategies during crises, and some components related to resilience. Method A national sample of the Mexican population was considered in a total of 2775 people whose ages ranged from 15 years and older. Questionnaires that met the psychometric criteria (reliability and validity) to be used in Latino samples were used. Results The results showed that older people experienced less stress and displayed more efficient coping behaviors. Discussion Regarding the exploration of some components associated with resilience, it was found that family constitutes an important interpersonal resource for coping with the crisis related to confinement due to the COVID-19 pandemic. In the future, it is proposed to make comparisons of the psychological factors evaluated to detect and analyze possible fluctuations due to the prevalence of epidemic conditions. mental health stress Mexican population COVID-19 resilience edited-statecorrected-proof typesetterts10 ==== Body pmcIntroduction Mexico is among the Latin American countries with the highest number of deaths due to the spread of COVID-19 among the population (Statista, 2022). Due to the suggestion of voluntary seclusion issued by the federal government on March 23, 2020, the immediate consequence in terms of mental health was an increase in the perception of stress at both the individual and social level (Secretaría de Salud, 2020). According to WHO reports (2020), physical isolation linked to the pandemic increased levels of loneliness, depression, anxiety, harmful substance use, and self-harm in the general population. In addition, the economic crisis associated with employment loss for more than a million people (Feix, 2020)—with the consequent worsening of social inequalities (Montes de Oca, et al., 2021)—represented the breeding ground for a foreseeable mental health crisis. Therefore, governmental, and social initiatives to offset, reduce, or, in the best of cases, eliminate pernicious consequences for the mental health of the population should be a priority for implementation (Márquez & Ryder, 2020). In this context, it is of utmost importance to understand phenomena such as stress, coping strategies, or resilience in the population affected by the pandemic. This paper aimed to document the impact of voluntary confinement due to COVID-19 on psychological variables related to the mental health of older adults, considering a national sample of the population in Mexico. Theoretical Framework We can understand stress as a physiological and psychological response to threats, real or perceived, in contexts of uncertainty. When experiencing psychological stress, people make use of coping strategies, both adaptive and maladaptive depending on the context. These strategies involve different psychological elements, varying in complexity and magnitude (Mella-Morambuena et al., 2020). At the same time, we can understand coping as a complex and multidimensional process that involves cognitive and behavioral efforts for the management or control of both external demands coming from the environment and internal demands linked to personality characteristics or emotional states. These demands are usually evaluated as conditions that exceed the person’s normal resources or abilities to process them, which he or she has to manage (Campos & Iraurgui, 2004; Lazarus & Folkman, 1984; Simpson et al., 2018) Lazarus and Folkman (1986) propose that there are at least 6 types of coping: (a) avoidance, (b) religious support, (c) socioemotional support, (d) emotional expression, (e) positive reappraisal, and (f) acceptance. Resilience, according to the conceptual review by Rodríguez (2018), involves two attributes: on the one hand, the overall ability shown by the person to maintain effective functioning in the face of environmental adversities (Cuervo, Yanguma & Arroyave, 2011; Trujillo, 2011) and, on the other hand, an efficient adaptation to meeting the various tasks of personal development in the face of the demands of the context in which they develop (Aracena, et al., 2000). Thus, resilience is considered as a combination of cognitive, emotional, and behavioral abilities that allows the person not only to efficiently face the demands of the context but also to transcend them and transform themselves, emerging stronger and more integrated. Although the definition of resilience is controversial (Parada-Fernández & Herrero-Fernández, 2022), the specialized literature refers to some important components that allude to elements of the environment, the person, or the interaction between the two. Thus, notions such as transcending adversity, overcoming situations of risk or stress, and rebuilding or recovering from critical events, are concepts that are linked to resilience (Fraser et al., 1999; Ruiz Párraga and López Martínez, 2012). While the perception of stress depends on the individual’s perceived physical, psychological, social, and contextual resources, coping strategies are associated with behaviors that seek to restore the psycho-emotional balance disturbed by stress (Simpson et al., 2018). When stress episodes are efficiently and constructively transcended, it can be assumed that the coping strategies have been effective and that individuals showed resilience, since not only have they behaved efficiently in the face of the demands of the context, by restoring the balance lost by the stress, but they have also acquired a new perspective of the situation and their psychological resources. In the context of the COVID-19 pandemic, it is interesting to note the contrast between the mass media coverage of the fragility and high risk of older adults during the first stage of the pandemic and the data reported in the specialized literature. In the systematic review of research on stress and well-being in adults during COVID-19 quarantine by Sterina et al. (2022), older adults generally presented lower stress levels and less negative emotions under quarantine than younger adults. Vannini et al. (2021) studied the impact of the COVID pandemic on measures of perceived stress, resilience, and behavioral coping strategies in a sample of older adults from an open population in Massachusetts and reported that participants demonstrated moderate levels of stress related to COVID-19 and showed relatively high levels of resilience. Also, Burke-Garcia et al. (2021) documented the effectiveness of a community intervention program promoting mental health and addressing feelings of grief, worry, and stress experienced during this time. These results highlight the importance of understanding the perspectives and experiences of participants to identify effective support and services. The study reported by Fuller & Huseth-Zosel (2021) evidenced that older adults efficiently employ coping strategies to deal with the stress experienced by the COVID-19 pandemic. Meanwhile, several studies (Carriedo et al., 2020; Keisari et al., 2021; Lim et al., 2021; Weitzel et al., 2021; Zach, Zeev & Ophir, 2021) agree that the higher the scores on measures of resilience, the lower are those on measures of anxiety and depression presented by older adults. Because statistic data also showed that older people, despite the social over-identification of their fragility, survived COVID infections more frequently than middle-aged people, it is thus reasonable to inquire which psychological resources allowed them such efficient adaptation or resilience. There are few studies documenting similar processes in Latin American cultures (González-González, et al., 2020; Montero, 2020). In this regard, it is important to be aware of the mental health profile in developing countries because it is known that the perception of stress is a function of the demands that individuals perceive in the contexts in which they live. Particularly, Brazil and Mexico were the countries with the highest fatality rate during the first wave of the COVID-19 pandemic, while they are the countries with the highest population and social disparity in the region. Therefore, it is of strategic interest to conduct studies to document the magnitude of perceived stress, the profile of coping strategies employed by individuals according to their age, and which variables can contribute to developing and increasing their resilience in the face of events as threatening as the COVID-19 pandemic, amid a context of social and economic uncertainty. In Mexico, Rivera-Ledesma & Montero-López Lena (2007) reported that for older people, spiritual life seems to play an important role in coping with stress and maintaining health in older adulthood. On the other hand, Verdugo-Lucero (2013) documented that adolescents most frequently use acceptance of responsibility as a coping mechanism, while avoidance is the least employed coping strategy. Differences in coping may also be reflected when sex is considered as a variable. Several studies (Cabanach et al., 2009, 2015; Di-Colloredo, et al., 2007; Guadarrama et al., 2018) coincide in documenting that women prefer socially supportive coping strategies, while men employ positive reappraisal. Little is known about the factors that increase people’s resilience. Usually, indirect measures such as closed questionnaires or scales are used in the evaluation of this psychological resource. These options, although they create specific scores of variables associated with resilience, have the limitation of reducing the possibility of identifying critical components that promote and increase resilience strategies specifically in older people. In 2011, Windle, et al. reported that there were 19 instruments measuring resilience that met psychometric criteria. While numerical precision allowed for the derivation of profiles of resilient behaviors, the phenomenological subtlety of the personal, social, and contextual contributors that enable an adaptive response to adverse events is lost in contrast. In addition, in consideration of the psychosocial and health exceptionality of COVID-19 confinement, particularly during the first stage of its pandemic dissemination, it was considered pertinent to use other tools to explore openly the components associated with resilient behaviors in the Mexican population. Therefore, in the present study, the incomplete sentence format was considered more appropriate to detect patterns of resources that would promote resilience. In this way, it will be understood that, in general, people could use intrapersonal resources (related to the development of individuality, self-esteem, confidence, autonomy, etc). and interpersonal resources (related to the interaction with other people to share affections, emotions, beliefs, and values) (Suárez & Mendoza, 2008). Given the contrasting findings, this paper aimed to document the impact of voluntary confinement due to COVID-19 on psychological variables related to the mental health of older adults, considering a national sample of the population in Mexico. In congruence with the described purpose, three variables were evaluated: (a) magnitude of perceived stress, (b) type of coping behaviors employed, and (c) intra- and interpersonal resources that could favor resilience. Method Data The CoBESS-2020 survey was applied online, based on a multistage probability sampling that allowed statistical inferences with a margin of error of less than .05 and a confidence level of 95%. The sample design considered two study domains, the first of which included the Valley of Mexico, and was made up of six states: Mexico City, State of Mexico, Hidalgo, Morelos, Tlaxcala, and Puebla. The other domain corresponded to the rest of the states. Within each domain, a proportional distribution to the population size of each state was used. The sample, albeit national, was not representative. The final sample consisted of 2775 participants, with a predominance of women (70.3%). The age range fluctuated from 15 to more than 80 years, with the most prevalent decade being between 20 and 29 years of age (23.5%). In contrast, the percentage of people aged 60 years or older was 17.56%. 20.25% of the sample were unable to ascribe themselves to a race, while 75.7% identified themselves as mestizo, 2.7% as indigenous, 1.2% as white, and only .1% as Afro-Mexican. In terms of educational level, 70.5% of the participants said they had undergraduate or graduate studies, 28.75% reached secondary or high school and only .7% indicated that they had only elementary level education. Finally, the average number of people living per household was 3.4 (sd = 1.68). Procedure The CoBESS-2020 was uploaded to social networks (Facebook and WhatsApp) on June 10th, 2020, and remained uploaded for 2 months until August 30th, 2020. University professors and researchers were contacted and acted as facilitators to distribute the CoBESS-2020 questionnaire in the social networks of each federal entity, using the “snowball” technique. Less than 1% of the questionnaires were administered by telephone, specifically when the participant was unable to answer the questionnaire independently due to difficulties in the handling of digital devices. The responses were recorded automatically using the Google Forms platform. Before this application, informed consent was obtained from the participants and they were informed of the confidentiality and treatment of their data. The survey Cost Benefits of Adaptation Strategies in Health, Economy, and Society-CoBESS-2020 was applied. (Montes-de-Oca et al., 2020). This survey is part of a broader study funded by UNAM (op. cit.) whose main objective was to document the losses and possible gains in the areas of health, economy, and social convenience that the Mexican population presented as a consequence of the COVID-19 pandemic. The CoBESS-2020 survey consisted of five sections. The first section collected sociodemographic data that enabled the authors to know the particularities of the sample (Alonso, Montero, and Montes de Oca, 2021). The second section explored some variables linked to the economic dimension, such as whether the person remained employed during the pandemic, what facilities were provided to continue performing work, or how the family was provided with the financial resources to support themselves during the pandemic, among other aspects. The third section explored variables on physical and mental health, enquiring about the prevalence of chronic illnesses and substance use. The occurrence of stressful experiences and some of the coping behaviors employed were also explored. The fourth section explored the socio-spatial dimension of the housing where participants resided. Household density and facilities available for mobility were documented. We also documented whether they had any social support network and the means of communication they used, from personal to digital, and the use they made of their free time. We also explored the perception of housing security during pandemic confinement. Finally, the fifth section included four semi-open questions in the form of incomplete sentences, which sought to explore the possible manifestation of resilience behaviors Stress and Coping Measures To document the possible fluctuations in the scores corresponding to two of the three variables considered: stress perception and types of coping, validated questionnaires adapted to the Mexican population were used (Montero, 1999, 2019; Montero et al., 2020). These adaptations were based on the questionnaires by Cohen, et al. (1983) for stress perception, and by Lazarus and Folkman (1986) for coping behaviors. The coping scale assessed 6 different types of strategies: (a) avoidance, (b) religious support, (c) socioemotional support, (d) emotional expression, (e) positive reappraisal, and (f) acceptance. Both stress perception (10 items) and coping behaviors (18 items) were 4-point Likert-type scalar items, where the higher the score, the higher the quality of the variable measured. Attention was given to making the appropriate adjustments to the phrasing of the item so that it would be congruent with the impact of COVID-19 confinement. Stress perception (α = .87) as well as the six coping factors assessed: avoidance (α = .83), support in religion (α = .97), socioemotional support (α = .93), expression of emotions (α = .79), positive reappraisal (α = .86), and acceptance (α = .75) showed acceptable internal consistency indexes. To evaluate the intrapersonal and interpersonal resources associated with resilience, it was considered pertinent to report, for this study, the data derived from the item that explored more directly what the participant identified as most significant and what had helped them to cope with the confinement. A semantic analysis was made out of this item and the emotions expressed in the response to the question “What has helped me to cope with this confinement is…” were explored. Analysis We first describe the sample by age, sex, marital status, education, and socio-economic status. The age variable was taken into account as a categorical variable. Stress means were obtained according to age group and sex. The means of each type of coping by age group and sex were obtained and tests were performed to evaluate the differences in the types of coping and the level of stress. We use ordinary least squares (OLSs) regression to examine the association between coping mechanisms, age, and sex, controlling for the level of stress. These statistical analyzes were made with the R software (2021). Finally, text analysis (Silge & Robinson, 2017) was performed to detect possible patterns in the use of resources associated with resilience to the incomplete sentence described. All Stopwords were removed and a TermDocumentMatrix was created containing all the words before the incomplete sentence and sparse was removed at 0.99. The frequency of the top words was obtained for the total sample and for the younger adults (n = 20–39) and older adults (n = 60 and +) groups so as to make comparisons between the words used. To classify the resources associated with resilience, two categories were identified. The first category is intra-personal (or individual) resources which allude to the transactions between emotional, cognitive, and behavioral components that distinguish the development of each individual. The second category is interpersonal (or social), which includes the interactions that participants establish with their immediate environment and that allows them to establish communication networks among family members, friends, and members of the community. Based on this categorization, we sought to identify the resources (intra vs. interpersonal) associated with resilience in the participants, and a comparison was made between two age groups, those aged 60 years and older versus young people aged 20–39 years. Results Table 1 shows the means of perceived stress and coping strategies by age and sex. When comparing the stress perception scores by sex, significant differences were found; women, in contrast to men, were the ones who obtained higher means in this variable. Likewise, when considering the types of coping, there were significant differences between men and women, in all cases, the highest means were obtained by women, indicating significant differences in religious coping, socioemotional support, emotional expression, positive reappraisal, and acceptance.Table 1. Comparison by Age and Sex Groups Considering Stress Perception Scores and Types of Apprehension. Age- sex Perceived stress [10–50] Avoidance [3–15] Religion support [3–15] Socioemotional support [3–15] Emotional expression [3–15] Positive reappraisal [3–15] Acceptance [3–15] 15–19 27.79 (6.6) 10.67 (2.5) 7.82 (4) 6.81 (3.5) 7.72 (3.1) 10.89 (2.7) 11.03 (2.7) 20–29 26.92** (6.4) 10.34 (2.7) 6.96 (3.8) 6.71* (3) 7.09 (2.9) 11.28 (2.5) 11.04 (2.6) 30–39 25.30 (6.1) 10.05 (2.7) 7.78 (4) 6.32 (2.7) 6.54 (2.6) 11.01 (2.7) 10.87 (2.6) 40–49 23.75 (5.9) 10.01 (2.8) 8.41 (3.9) 6.52 (2.6) 6.43 (2.4) 11.22 (2.6) 10.99 (2.7) 50–59 22.57 (5.7) 10.25 (3) 9.51** (3.7) 6.34 (2.7) 6.42 (2.6) 11.48* (2.5) 11.51* (2.6) 60 y más 21.38 (5.6) 10.51 (3) 9.89 (4) 6.21 (2.7) 6.17 (2.4) 11.37 (2.8) 11.77 (2.8) Hombre 23.01 (6.2) 10.08 (2.9) 7.13 (3.9) 5.88 (2.7) 6.03 (2.6) 10.87 (2.8) 10.7 (2.8) Mujer 24.81 (6.3) 10.31 (2.8) 8.92 (4) 6.67 (2.8) 6.81 (2.6) 11.40 (2.5) 11.41 (2.6) Note. To assess whether there were significant differences in the means of stress scores and types of coping according to sex, a t test was performed. To assess whether there were significant differences between age groups, ANOVA analyses and post hoc tests were performed. Means are shown and standard deviations are in parentheses, p*<0.001, <0.05 At the same time, significant differences were found between the group aged 15–19 years and those aged 60 years and over, with the former reporting higher scores in the perception of stress. In the religion-focused coping subscale, it was the group aged 60 and over that obtained the highest mean. This same pattern was replicated when considering the socioemotional support-focused coping subscale, where differences were identified where the 60 and over group differed significantly from the 20–29 year-old group. In the emotional expression subscale, it was the group aged 15–19 years that obtained the highest mean. Table 2 shows the results of the OLS regression models for each coping strategy controlling for age, sex, education, and perceived stress. When analyzing age, it was found to be a significant predictor mainly in the religious support coping strategy, followed by avoidance and emotional expression. This suggests that older people make more frequent use of religious coping compared to younger people, except for the 50–59 years age group, where more explained variance associated with religious coping was found. Interestingly, it can be seen that the perception of stress was a significant predictor, although with low beta values in five of the six types of coping evaluated, and only in the coping of support in religion was it not a significant predictor.Table 2. Linear Regressions by Coping Type. Avoidance Religion support Socioemotional support Emotional expression Positive reappraisal Acceptance Women – 60 years and older intercept 9.49 9.63 3.39 .98 12.28 12.72 Perception of stress .02 .01 .14 .24 −0.03 −0.02* Ages 15–19 −0.45 −2.64** −0.21 −0.03 −0.23 −0.35 Ages 20–29 −0.4* −3.15 −0.3 −0.45 .07 −0.59 Ages 30–39 −0.48 −2.07 −0.47 −0.58 −0.23 −0.87 Ages 40–49 −0.52** −0.44 −0.03 −0.32* −0.07 −0.74 Ages 50–59 −0.28 2.64 −0.06 −0.04 .12 −0.26 Sex – men −0.15 −1.76 −0.53 −0.33 −0.58 −0.75 Educational level .31 .36** −0.08 .004 −0.02 −0.16* Note. p** <0.001, *<0.05. As for sex, men reported significantly lower scores for all coping strategies except avoidance. Finally, when analyzing the effects of education, it was found that the higher the level of education, the greater the use of avoidance and religious support coping strategies and the lower the use of acceptance. To analyze the possible groupings of resources (intra vs. interpersonal) linked to resilience in the face of COVID-19 confinement, Figure 1 shows the most frequent words used in the incomplete sentence “What has helped me to cope with this confinement is…”. As can be seen, 16 words were the most used by the participants, of which 6 represented an interpersonal resource, 7 an intrapersonal resource, and the remaining 3 words described a combination of both resources. The word family was the most frequently used word by participants, followed by valuing and health. This suggests that family constitutes the main interpersonal resource for the sample in general.Figure 1. Most frequent words used before the incomplete sentence “What has helped me cope with this confinement is…” as a function of the type of resource linked to resilience. As a second step in the analysis of resilience resources, the responses given by young adults to the incomplete sentence referred to were separated by comparing them with those of older adults. For young adults, the most frequent words were: value (n = 85; intrapersonal resource), family (n = 72; interpersonal resource), patience (n = 49; intrapersonal resource), and time (n = 40; intrapersonal resource). On the other hand, the words most frequently used by the older adults were: alive (n = 55; intra-interpersonal resource), value (n = 51; intrapersonal resource), family (n = 49; interpersonal resource), health (n = 43; interpersonal-intrapersonal resource), patience (n = 36; intrapersonal resource), and being (n = 28; interpersonal-intrapersonal resource). There were coincidences in both groups in the reference to three concepts: value, family, and patience. Together, this suggests the need to value the family and to have patience in the situation of COVID-19 confinement. It is interesting to note that young adults (aged 20–39 years) referred to interpersonal concepts, while older adults (aged 60 years and over) used words with a combination of interpersonal and intrapersonal resources. Discussion The recognition of mental health as a human right (De la Fuente et al., 1997) emphasizes the promotion of mental health, especially among the most vulnerable groups, in which older adults stand out due to their population representation (INEGI, 2021). Accordingly, this study aimed to document the impact of confinement due to COVID-19 on some psychological variables related to the mental health of older adults. Three variables were considered: (a) magnitude of perceived stress, (b) coping strategies employed in the face of COVID-19 confinement, and (c) intrapersonal and interpersonal resources that favor resilience in the face of the stress produced by the pandemic. From data obtained from a nationwide sample, we provided evidence of the level of stress experienced by participants, documented the use of different coping strategies employed in the face of perceived stress, and identified some intra- and interpersonal resources linked to resilience that participants were likely to develop in the face of extreme physical and social stress due to the COVID-19 pandemic. To make a comprehensive analysis of the findings obtained, this discussion was organized according to the level of perceived stress, the coping strategies employed, and the intra- and inter-personal resources that facilitated resilience in the participants in the studied sample. On the Level of Perceived Stress The results showed that women are more likely to perceive greater stress in crises in contrast to men. This agrees with the findings of Meda-Lara, et al. (2022); Zamarripa, et al., (2020), and González-González, et al. (2020). It ratifies the perceptual characteristics of women to detect changes in the environment both physical and social and to develop perceptual sensitivity. It is interesting to note that the older the participants of the sample studied, the less perceived stress they showed (see Table 1). This is consistent with the theory of socioemotional selectivity proposed by Carstensen, et al. (1999) in the sense that older people filter more the information they receive from their environment and only respond to stimuli that are meaningful to them. On Types of Coping Strategies Avoidance coping did not show differences by age, group, or sex. Support in religion showed that older women employed this type of coping strategy significantly more often in contrast to men and younger people (see Table 1). This could be a cultural manifestation (Berry, 1990; Díaz-Guerrero, 1982) that allows older women to adapt more efficiently to the demands of the context. Regarding socioemotional support coping, participants in the 20–29 age group were those who sought the most socioemotional support. Similarly, people between the ages of 15 and 29 reported significantly higher levels of emotional expression coping strategies. This evidence accounts for the type of coping used by younger people focused on emotional processes. Regarding the positive reappraisal type of coping, the group aged 50–59 was the one with the highest score when compared to the other age ranges. A possible hypothesis that explains this result consists in what Erickson (1982) calls a generativity crisis. In this crisis, the person focuses on building a legacy for the generations that follow, and positive reappraisal, as well as acceptance, are efficient coping strategies to rescue significant lessons from lived experiences and relabel them as a legacy. This conjecture is supported by the results found with the acceptance type of coping, where the groups of 50 and 60 and over, and women, were the ones who most often presented this type of coping. We found that age was significantly associated with five of the six types of coping evaluated. Only in the emotional support coping strategy did age not show a significant association. Thus, the results of the regressions showed that, when considering avoidance, the higher the age, the lower the avoidance, and the higher the education level, the higher the avoidance. As for religious support coping strategy, the older the age and the higher the education level, the greater the support for religion. However, age was not significant for the use of socio-emotional support as a coping strategy. Being female and older did predict the expression of emotions and acceptance as forms of coping regarding the COVID-19 confinement. On the Categories of Intra- and Inter-personal Resources Linked to Resilience For the total sample, interpersonal resources (related to interaction with other people to share affections, emotions, beliefs, and values) followed by intrapersonal resources were the most used. The group aged 20–39 years showed that they used intrapersonal resources more, in contrast to the resources indicated by the participants aged 60 years and older, who used both resources. This evidence is significant because it shows, on the one hand, the preeminence of the family as a resource “sine qua non” both to face crises and to transcend and overcome them, with greater psychological, social, and emotional strength. The approach to detecting resilience resources in this sample was incipient because priority was given to the free expression of the resources identified by the participants, rather than classifying them in advance and waiting for their response on pre-established categories. It is pertinent to emphasize that in this study the indirect measurement of psychosocial processes was transcended and participants were given a voice through the use of the incomplete sentence technique (Sacks & Levy, 1967). This technique allowed participants to describe directly and in their own words the emotions they experienced during confinement, as well as the resources they used to transcend the negative experiences associated with the pandemic. Scope and Limitations Although this study was based on a national sample, and this represents one of its advantages, it is important to point out that the exhaustiveness in the inclusion of a total of 66 items could facilitate a halo effect in the responses and thus bias the results. The distribution of the CoBESS-2020 questionnaire by digital means also represented a bias in obtaining the sample, since literate digital-savvy people could’ve had an easier time answering this questionnaire. A lot of Mexican Older Adults do not have access to a computer or other digital means, and those who might have it, might not be as digital-literate as younger adults, so a very specific profile of an older adult is the one who could’ve had access to the questionnaire. It would be desirable for future studies to obtain a national representation with multistage sampling that guarantees the inclusion of all the groups that make up the population profile of Mexico. Among the achievements of this study, the opportunity to obtain a vivid and close record at a crucial time in the development of the COVID-19 pandemic stands out. This study covered the first wave of dispersion of this disease and represents an invaluable base for future population contrasts, considering similar variables. Conclusion The confinement due to the COVID-19 pandemic has ended. Now is the time to tend to the mental health needs of the population. As stated at the beginning of this paper, governmental and social initiatives to offset, reduce, or eliminate pernicious consequences for the mental health of the population should be a priority for implementation. The results of this study might help those initiatives to have a common ground of knowledge. Knowing how older adults perceive stress; cope with the stress regarding the confinement, and the important elements regarding those coping mechanisms, like the importance of family for older adults, might be of use for the stakeholders in charge of public policies. Another study should be conducted in the future to document possible fluctuations in the different variables considered after almost 2 years of the pandemic. It will be interesting to evaluate the degree of adaptation achieved and the challenges perceived by citizens after the health, economic, and socio-political measures were adopted. The possible costs and benefits achieved in terms of mental health will be reported shortly. ORCID iDs Maria Montero-Lopez Lena https://orcid.org/0000-0001-9109-381X Alejandro Zamudio https://orcid.org/0000-0002-3902-5585 The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research project was completed with the support of the 2021-2023 cycle Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica (DGAPA-PAPIIT-IG300221) granted by the National Autonomous University of Mexico. ==== Refs References Alonso-Reyes P. Montero-López Lena M. Montes de Oca V. (2021). Costos-beneficios de las estrategias de adaptación en salud, economía y sociedad ante COVID-19 en México 2020. Universidad Nacional Autónoma de México. Aracena M. Catillo R. Haz A. M. Francisco C. Muñoz S. Luis B. Román F. (2000). Resiliencia al maltrato físico infantil: Variables que diferencian a los sujetos que maltratan y no maltratan físicamente a sus hijos en el presente, y que tienen una historia de maltrato físico en su infancia. Revista de Psicología de La Universidad de Chile, 9 (1 ), 11–28. 10.5354/0719-0581.2000.18543 Berry J. W (1990). Psychology of acculturation: Understanding individuals moving between cultures. In Brislin R. W. (Ed.), Applied cross-cultural psychology (pp. 232–253). Sage. Burke-Garcia A. Johnson-Turbes A. Mitchell E. W. Vallery Verlenden J. M. Puddy R. Mercado M. C. Nelson P. Rabinowitz L. Xia K. Wagstaff L. Feng M. Caicedo L. Tolbert E. Tolbert E. (2021). How right now? Supporting mental health and resilience amid COVID-19. Traumatology, 27 (4 ), 399–412. 10.1037/trm0000322 35360002 Cabanach R. G. Fariña F. Freire C. González P. Ferradás M. del M. (2015). Diferencias en el afrontamiento del estrés en estudiantes universitarios hombres y mujeres. European Journal of Education and Psychology, 6 (1 ), 19. 10.30552/ejep.v6i1.89 Cabanach R. G. Millán P. G. Freire C. (2009). El afrontamiento del estrés en estudiantes de ciencias de la salud. Diferencias entre hombres y mujeres. Aula Abierta, 37 (2 ), 3–10. https://dialnet.unirioja.es/servlet/articulo?codigo=3088531 Campos M. Iraurgui J. (2004). Afrontamiento y regulación emocional de hechos estresantes un meta-análisis de 13 estudios. Boletín de Psicología, 82 , 25–44. https://dialnet.unirioja.es/servlet/articulo?codigo=1302317 Carriedo A. Cecchini J. A. Fernandez-Rio J. Méndez-Giménez A. (2020). COVID-19, Psychological well-being and physical activity levels in older adults during the nationwide lockdown in Spain. The American Journal of Geriatric Psychiatry, 28 (11 ), 1146–1155. 10.1016/j.jagp.2020.08.007 32919872 Carstensen L. L. Isaacowitz D. M. Charles S. (1999). Taking time seriously: A theory of socioemotional selectivity. The American Psychologist, 54 (3 ), 165–181. 10.1037//0003-066x.54.3.165 10199217 Cohen S. Kamarck T. Mermelstein R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24 (4 ), 385–396. 10.2307/2136404 6668417 Cuervo Rodríguez J. J. Yanguma C. Arroyave M. (2011). Comprensiones de la resiliencia en los libros editados en español y localizados en seis bibliotecas de Bogotá Colombia. Diversitas, 7 (1 ), 57–71. 10.15332/s1794-9998.2011.0001.04 De la Fuente R. Medina-Mora M. E. Caraveo J. (1997). Salud mental en México. Instituto Mexicano de Psiquiatría/Fondo de Cultura Económica. de Salud S. (2020). Inicia fase 2 por coronavirus COVID-19. https://www.gob.mx/salud/prensa/095-inicia-fase-2-por-coronavirus-covid-19 Díaz-Guerrero R. (1982). Psicología del mexicano. Descubrimiento de la etnopsicología. Trillas. Di-Colloredo C. Cruz D. P. A. Moreno Y. J. (2007). Descripción de los estilos de afrontamiento en hombres y mujeres ante la situación de desplazamiento. Psychologia, Avances de la disciplina, 1 (2 ), 125–156. https://www.redalyc.org/pdf/2972/297224996002.pdf Erikson E. H. (1982). The life cycle completed. Norton. Feix N. (2020). México y la crisis de la COVID-19 en el mundo de trabajo: Respuestas y desafíos (pp. 1–23). Organización Internacional del Trabajo. https://www.ilo.org/wcmsp5/groups/public/–-americas/–-ro-lima/–-ilo-mexico/documents/publication/wcms_757364.pdf Fraser M. W. Galinsky M. J. Richman J. M. (1999). Risk, protection, and resilience: Toward a conceptual frame-work for social work practice. Social Work Research, 23 (3 ), 131–143. 10.1093/swr/23.3.131 Fuller H. R. Huseth-Zosel A. (2021). Lessons in resilience: Initial coping among older adults during the COVID-19 pandemic. The Gerontologist, 61 (1 ), 114–125. 10.1093/geront/gnaa170 33136144 González-González A. Toledo-Fernandez A. Romo-Parra H. Reyes-Zamorano E. Betancourt-Ocampo D. (2020). Psychological impact of sociodemographic factors and medical conditions in older adults during the COVID-19 pandemic in Mexico. Salud Mental, 43 (6 ), 293–301. 10.17711/SM.0185-3325.2020.040 Guadarrama R. Carmona V. Márquez O. Veytia M. Padilla I. M. (2018). Diferencias en la aplicación de estrategias de afrontamiento adolescente entre alumnos hombres y mujeres del nivel medio superior. Interdisciplinaria: Revista de Psicología y Ciencias Afines, 35 (1 ), 205–216. 10.16888/interd.2018.35.1.11 Instituto Nacional de Estadística y Geografía (2021). Estadísticas a propósito del día internacional de las personas adultas mayores. [INEGI]. https://www.inegi.org.mx/contenidos/saladeprensa/aproposito/2021/EAP_ADULMAYOR_21.pdf (2021, 29 de septiembre). Keisari S. Hoffman Y. Ring L. Palgi Y. (2021). The Moderating effect of older adults’ receptive arts engagement on the association between resilience and anxiety symptoms during coronavirus breakout. The Journal of Nervous and Mental Disease, 209 (6 ), 443–448. 10.1097/NMD.0000000000001326 34037551 Lazarus R. S. Folkman S. (1984). El concepto de afrontamiento en estrés y procesos cognitivos. Martínez Roca. Lazarus R. S. Folkman S. (1986). Estrés y procesos cognitivos. Martínez Roca. Lim X. Y. Yap A. C. Mahendran R. Yu J. (2021). The interplay between anxiety, fear, protective behaviors, compassion, and resilience among older adults during a COVID-19 lockdown: A structural equation modeling study. Translational Behavioral Medicine, 11 (5 ), 1172–1178. 10.1093/tbm/ibaa143 33793946 Márquez C. Y. Ryder R. (2020). Covid-19 en México: Las múltiples aristas de la crisis. Política Exterior. https://www.politicaexterior.com/covid-19-en-mexico-las-multiples-aristas-de-la-crisis/ Meda-Lara R. M. Munoz-Valle J. F. Juarez-Rodriguez P. Figueroa-Lopez C. Herrero M. de Santos Ávila F. Palomera-Chavez A. Yeo Ayala C. Moreno-Jimenez B. (2022). Psychological responses to COVID-19 in a Mexican population: An exploratory study during second and third phases. Psychology, Health and Medicine, 27 (1 ), 257–264. 10.1080/13548506.2021.1883689 Mella-Morambuena J. López-Angulo Y. Sáez-Delgado Y. Del Valle M. (2020). Estrategias de afrontamiento en apoderados durante el confinamiento por la COVID-19. CienciAmérica, 9 (2 ), 322–333. 10.33210/ca.v9i2.324 Montero-López Lena M. (1999). Soledad: Desarrollo y validación de un inventario multifacético para su medición. Tesis doctoral. Universidad Nacional Autónoma de México. http://132.248.9.195/pdbis/269524/Index.html Montero-López Lena M. (2019). Pobreza como dolor social, su impacto en la salud física y mental en la vejez. Documento interno UNAM-DGAPA. Montero-López Lena M. (2020). Psicogerontología y COVID19: Experiencia en Latinoamérica [Webinar]. Facultad de Psicología, UNAM. https://www.youtube.com/watch?v=PLNC1nxwxLk Montero-López Lena M. Zarebski G. Corte B. Guidotti C. Heredia D. Kabanchik A. Rivas M. (2020). Cuestionario REDIP-COVID19. Documento Interno REDIP. Montes-de-Oca V. Alonso M. P. Montero-López Lena M. Vivaldo-Martínez M. (2021). Sociodemografía de la desigualdad por COVID-19 en México. Revista Mexicana de Sociología, 83 , 67–91. 10.22201/iis.01882503p.2021.0.60169 Montes-de-Oca V. Montero-López Lena M. Alonso-Reyes M. D. P. (2020). Encuesta Costos-beneficios de las estrategias de adaptación en salud, economía y sociedad-CoBESS-2020, en México ante la pandemia por COVID-19. DGAPA-UNAM, Base de datos. Organización Mundial de la Salud [OMS] (2020). Salud mental y COVID-19. Organización Panamericana de la Salud. https://www.paho.org/es/salud-mental-covid-19 Parada-Fernández P. Herrero-Fernández D. (2022). Propiedades psicométricas y análisis del cambio fiable de la Resilience Scale en una muestra normativa española. Revista Mexicana de Psicología, 39 (1 ), 45–57. (ISSN 0185-6073. https://www.researchgate.net/publication/361951618_Propiedades_psicometricas_y_analisis_del_cambio_fiable_de_la_Resilience_Scale_en_una_muestra_normativa_espanola Rivera-Ledesma A. Montero-López Lena M. (2007). Medidas de afrontamiento religioso y espiritualidad en adultos mayores mexicanos. Salud Mental, 30 (1 ), 39–47. Rodríguez D. (2018). Una revisión al constructo resiliencia. Historia y panorama actual. Tesis de grado, Universidad de San Buenaventura Colombia. http://bibliotecadigital.usb.edu.co/bitstream/10819/5694/1/Revisi/n_Constructo_Resiliencia_Rodriguez_2017.pdf R software (2021). Download R-4.1.3 for windows. The R-project for statistical computing. Ruiz Párraga G. T. López Martínez A. E. López Martínez A. E. (2012). Resiliencia psicológica y dolor crónico. Escritos de Psicología/Psychological Writings, 5 (2 ), 1–11. 10.5231/psy.writ.2012.1001 Sacks J. Levy S. (1967). El test de frases incompletas. In Abt L. E. Bellak L. (Eds.), Psicología proyectiva. Paidós. Silge J. Robinson D. (2017). Text mining with R: A tidy approach. O’Reilly Media, Inc. Simpson D. Suárez L. Cox L. Connolly S. (2018). The role of coping strategies in understanding the relationship between parental support and psychological outcomes in anxious youth. Child and Adolescent Social Work Journal, 35 (4 ), 407–421. 10.1007/s10560-018-0531-y Statista (2022, 3 April). Número de casos confirmados de coronavirus (COVID-19) en América Latina y el Caribe al 3 de abril de 2022, por país. https://es.statista.com/estadisticas/1105121/numero-casos-covid-19-america-latina-caribe-pais/ Sterina E. Hermida A. P. Gerberi D. J. Lapid M. I. (2022). Emotional resilience of older adults during COVID-19: A systematic review of studies of stress and well-being. Clinical Gerontologist, 45 (1 ), 4–19. 10.1080/07317115.2021.1928355 34080527 Suárez I. Mendoza B. (2008). Desarrollo de la inteligencia emocional y de la capacidad para establecer relaciones interpersonales e intrapersonales. Laurus, 14 (27 ), 76–95. Trujillo S. (2011). Resiliencia: ¿proceso o capacidad? Una lectura crítica del concepto de resiliencia en 14 universidades colombianas. Revista Iberoamericana de Psicología, 4 (11 ), 13–22. 10.33881/2027-1786.rip.4102 Vannini P. Gagliardi G. P. Kuppe M. Dossett M. L. Donovan N. J. Gatchel J. R. Quiroz Y. T. Premnath P. Y. Amariglio R. Sperling R. A. Marshall G. A. (2021). Stress, resilience, and coping strategies in a sample of community-dwelling older adults during COVID-19. Journal of Psychiatric Research, 138 , 176–185. 10.1016/j.jpsychires.2021.03.050 33862301 Verdugo-Lucero J. C. Ponce de León-Pagaza B. G. Guardado-Llamas R. E. Meda-Lara R. M. Uribe-Alvarado J. I. Guzmán-Muñiz J. (2013). Estilos de afrontamiento al estrés y bienestar subjetivo en adolescentes y jóvenes. Revista Latinoamericana de Ciencias Sociales, Niñez y Juventud, 11 (1 ), 79–91. http://www.scielo.org.co/pdf/rlcs/v11n1/v11n1a05.pdf Weitzel E. C. Löbner M. Röhr S. Pabst A. Reininghaus U. Riedel-Heller S. G. (2021). Prevalence of high resilience in old age and association with the perceived threat of COVID-19: Results from a Representative Survey. International Journal of Environmental Research and Public Health, 18 (13 ), 7173. 10.3390/ijerph18137173 34281110 Windle G. Bennett K. M. Noyes J. (2011). A methodological review of resilience measurement scales. Health and Quality of Life Outcomes, 9 , 8. art. 8 10.1186/1477-7525-9-8 21294858 Zach S. Zeev A. Ophir M. Eilat-Adar S. (2021). Physical activity, resilience, emotions, moods, and weight control of older adults during the COVID-19 global crisis. European Review of Aging and Physical Activity: Official Journal of the European Group for Research Into Elderly and Physical Activity, 18 (1 ), 5. 10.1186/s11556-021-00258-w 33648448 Zamarripa J. Delgado-Herrada M. Morquecho-Sanchez R. Banos R. de la Cruz-Ortega M. Duarte-Felix H. (2020). Adaptability to social distancing due to COVID-19 and its moderating effect on stress by gender. Salud Mental, 43 (6 ), 273–278. 10.17711/SM.0185-3325.2020.037
PMC010xxxxxx/PMC10290936.txt	==== Front J Intellect Disabil J Intellect Disabil spjld JLD Journal of Intellectual Disabilities 1744-6295 1744-6309 SAGE Publications Sage UK: London, England 37328257 10.1177_17446295231184117 10.1177/17446295231184117 Original Research Following 4 months of social distancing during COVID-19 Pandemic in Brazil did not change aspects of functioning in children and adolescents with developmental disabilities: A longitudinal study https://orcid.org/0000-0001-7883-3123 Brugnaro Beatriz Helena Department of Physical Therapy, Child Development Analysis Laboratory (LADI), Federal University of São Carlos (UFSCar) , São Carlos, SP, Brazil https://orcid.org/0000-0003-1390-2284 Fernandes Gesica Department of Physical Therapy, Child Development Analysis Laboratory (LADI), Federal University of São Carlos (UFSCar), São Carlos, SP, Brazil https://orcid.org/0000-0001-7699-3094 Vieira Fabiana Nascimento Department of Physical Therapy, Child Development Analysis Laboratory (LADI), Federal University of São Carlos (UFSCar), São Carlos, SP, Brazil Pavão Silvia Letícia Departament of Prevention and Rehabilitation in Physical Therapy, Federal University of Paraná , Curitiba, PR, Brazil Rocha Nelci Adriana Cicuto Ferreira Department of Physical Therapy, Child Development Analysis Laboratory (LADI), Federal University of São Carlos (UFSCar), São Carlos, SP, Brazil Beatriz Helena Brugnaro, Department of Physical Therapy, Child Development Analysis Laboratory (LADI), Federal University of São Carlos (UFSCar), Rod. Washington Luis, km 235, 13565-905, São Carlos, SP, Brazil. Email: bia10.helena@gmail.com 16 6 2023 16 6 2023 17446295231184117© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. The COVID-19 pandemic and its demands of social distancing have created challenges in the lives of children/adolescents with developmental disabilities and their families, which would change aspects of children’s functioning. The objetive of this study was to evaluate changes in some components of functioning of children/adolescents with disabilities following 4 months of social distancing during a period of high contamination rate in the year 2020 in Brazil. Participated 81 mothers of children/adolescents with disabilities, 3-17 years, most of them (80%) diagnosed with Down syndrome, cerebral palsy and autism spectrum disorder. Remote assessments of functioning’ aspects including IPAQ, YC-PEM/ PEM-C, Social Support Scale and PedsQL V.4.0. Wilcoxon tests compared the measures, with significance level <0.05. No significant changes in participant’s functioning were identified. Social adjustments required to facing the pandemic during two points in time in the midst of the pandemic did not change the evaluated aspects of functioning in our sample of Brazilian. functioning children disabilities COVID-19 Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) 2019/13570-6 2019/13716-0 2020/05685-5 2021/15016-6 Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Financial code 001 edited-statecorrected-proof typesetterts10 ==== Body pmcWhat this paper adds? • The level of physical activity and participation at home in children with developmental disabilities did not change between the pandemic period analyzed. • The social support received by the caregiver of children with developmental disabilities did not change between the pandemic period analyzed. • The child's health-related quality of life did not change during the period evaluated. Introduction The Coronavirus (COVID-19) pandemic, recognized in March 2020 by the World Health Organization (WHO), has brought numerous and unprecedented challenges to health systems and daily life of the population. In Brazil, the period of high contamination rate in the year 2020 occurred between June and September 2020, when the highest rates of positive cases of COVID-19 were registered (Silva et al, 2020) until that moment. Without available vaccines, social distancing became the most protective/drastic measure to contain the viral spread (Nussbaumer-Streit et al., 2020; Lippi et al., 2020). Considering the relevance of the environment for children/adolescents’ development (Kramer et al., 2012), social distancing seems to have a negative impact on their lives, potentially determining worse cognitive development and higher rates of anxiety and depression (Almeida et al., 2021). The interruption of outdoor activities, deprivation of social contact, associated with the maintenance of school demands by means of video lessons, resulted in many changes, like an increase in the time of exposure to screens (TV, smartphones, computers, and tablets) (French et al 2020). The long periods of remote teaching and the required adjustments to keep the didactic schedule also increased the volume of school tasks carried out at home, demanding time and dedication not only from the child but also from their parents. Therefore, parental stress, due to overlapping demands of household chores, remote work, educational support for their children and lack of social support, might create anxiety, irritability, and lack of patience for raising children. Although the globally advocated social distancing measures might have had potential to negatively impact everyone, children and adolescents with developmental disabilities may have been more intensely affected by these measures due to their greater health vulnerability (Okuyama et al., 2021). For this population, the abrupt interruptions in elective medical and therapeutic procedures and the complete withdrawal of face-to-face school and leisure activities were among the biggest challenges arising from social distancing (Aquino et al., 2020; Batista et al., 2021). Studies prior to the COVID-19 pandemic reported that children with developmental disabilities presented lower quality of life (Isa et al., 2016; Kuru and Piyal, 2018) and physical activity levels (Einarsson et al., 2015; Aviram et al., 2019) when compared to their typical peers. Moreover, their parents and caregiver showed lower perceptions of social support (Holanda et al., 2015, Kuru and Piyal, 2018). Therefore, in the pandemic scenario, the peculiarity of the social changes might have affected the functioning of both these children and their parents/caregivers. Most of the studies addressing the effects of COVID-19 pandemic have focused on specific components of health of the individuals, such as, caregivers’ quality of life (Al Awaji et al., 2021), behavioral aspects (Summers et al., 2020; Marinho et al., 2019) or physical activity level of children with disabilities (Marinho et al., 2019; Theis et al., 2021; Suarez-Balcazar et al., 2021). We point out the need to address the potential changes observed during pandemic on the functioning of children and adolescents with developmental disabilities, emphasizing functioning in the light of the International Classification of Functioning, Disability and Health (ICF), by assesing multiple biopsychosocial components of health. Comparing functioning in children with Down syndrome one month before and three months after pandemic beginning, a study found a significant reduction in the social support reported by the children’s caregivers, and an increase in children’s participation at home (Brugnaro et al., 2021). Possibly, the longer time spent by these children and their parents at home, with increased contact with domestic activities, generated greater opportunities for participation in this environment. Taking into account the multidirectionality between the health components of individuals for determination of their functioning (Castro et al., 2020; Halberstad et al., 2019), we believe that the extensive duration of social distancing measures imposed to control the pandemic (Lippi et al., 2020; Nussbaumer-Streit et al., 2020) would have impacted functioning of these children. In addition, considering the vulnerability of the population of children and adolescents with developmental disabilities, this pandemic situation might affect other components of their health. The knowledge of these changes may help rehabilitation professionals involved with these children to better comprehend the relevance of social aspects for their health, as well as might guide them to develop and adopt measures to promote functioning for this population. Therefore, the present study proposes to carry out a remote assessment (Brugnaro et al., 2022) longitudinally of some aspects of functioning (physical activity level, home participation, home environment support, social support and quality of life related to health) of children and adolescents with developmental disabilities during a period of high contamination rate in the year 2020 in Brazil. We expect that aspects of functioning evaluated will be undermined after 4 months of pandemic, considering the challenges imposed by social distancing, and the difficulties that children with developmental disabilities and their caregivers already face prior to the pandemic. This study will possible new insights about the functioning assessment of children with developmental disabilities in Brazil during pandemic, in order to provide important data about how they were facing the pandemic during this period. Methods Research design This was a longitudinal study with a convenience sample. It was approved by the local ethics committee (protocol number: 31786920.8.1001.5504) following the recommendations of the Check-list for Reporting Results of Internet E-Surveys (CHERRIES) statement guideline (Eysenbach, 2004). Participants Parents of children and adolescents aged 3 to 17 years diagnosed with developmental disorders were assessed. We consider as developmental disabilities: (1) diagnosed neuromotor impairments, such as, cerebral palsy, developmental coordination disorders, myelomeningocele, chronic syndromes with motor impairments and neurodegenerative diseases (Field & Roxborough, 2012; Field & Roxborough, 2011), and (2) other diagnosed conditions affecting cognitive development, such as, Down syndrome (Agarwal Gupta and Kabra, 2014) and autism spectrum disorder (Baird et al., 2003). Participants’ age was chosen according to the range within the cut line of the used instruments and based on the Brugnaro et al. (2022) study, since we used part of the same assessment. The study was widespread through communication media and social networks. Participants were selected from contact on demand, direct invitation contact from researchers or in partnership with institutions that take care of children/adolescents with developmental disabilities. Contacts were established by WhatsApp or e-mail. We did not include parents who did not sign the Informed Consent Form (ICF) and child’s the Minor’s Consent Form (MCF). Those who wanted to leave the research, for any reason, were discontinued from the study. Procedures All the included participants were assessed by electronic forms (EF), and also by a telephone interview. The EF contained the ICF, MCF and a characterization form addressing child’s profile (age, sex, diagnosis, mothers’ age, caregiver education level, type of residence, number of rooms in the residence and Socioeconomic Classification) and validated questionnaires (IPAQ-short version, Social Support Scale and PedsQL V.4.0). Phone interviews were used by the researchers allowing parents’ responses to YC-PEM/PEM-CY (according to the child’s age). These calls were conducted by only one evaluator ensuring consistency and reliability for the assessments. Following the recommendations of the CHERRIES, parents/caregivers received a link containing ICF and MCF. Study design involved two assessments (AI and AII) separated by an interval of four months. The first assessment was carried out between June/September 2020, during a period of high contamination rate in the year 2020 in the country where the study was carried out. Parents should answer all the measures items within 10 calendar days of EF submission. Outcome measures We assessed the sample characterization and some aspects of functioning based on the biopsychosocial model in the ICF, in two modalities: EF and telephone interview, which will be described below. EF: Sample characterization The caregiver's education level was obtained through a closed question for the participant to choose the option that best describes (Incomplete Elementary, Complete Elementary, Incomplete high school, Complete high school, Incomplete Higher Education, Complete Higher Education) which were categorized by frequency of occurrence. Participants’ socio-economic classification was obtained using the Brazil Economic Classification Criterion (CCBE), according to ABEP - Brazilian Association of Research Companies. ABEP is an economic segmentation instrument that uses the survey of household characteristics (presence and quantity of some household items of comfort and educational level of the family head) to differentiate the population. The criterion assigns points according to each household characteristic and performs the sum of these points. Subsequently, correspondence is made between the value obtained in the sum and score ranges established by the economic classification, from highest to lowest, defined by A, B1, B2, C1, C2, D-E. Sample characterization data is shown on Table 1.Table 1. Characterization of the participants of the first and second evaluation. Assessment I - n (%) Assessment II - n (%) Child's Gender (n = 81) (n = 60) Female 30 (37.0%) 18 (30.0%) Male 51 (63.0%) 42 (70.0%) Child’s Age (completed years) (n = 81) (n = 60) Mean 7,58 7,65 Standard deviation 3,55 3,61 Child’s Age description (completed years) n n 3 8 4 4 8 6 5 10 9 6 11 10 7 9 7 8 9 6 9 6 2 10 4 4 11 4 2 12 3 3 13 4 3 14 0 0 15 2 1 16 1 1 17 2 2 Developmental Disability (n = 81) (n = 60) Down syndrome 34 (42.0%) 23 (38.3%) Cerebral Palsy 16 (19.8%) 12 (20.0%) Autism Spectrum Disorder 14 (17.3%) 11 (18.3%) Others 15 (18.5%) 12 (20.0%) Was not informed 2 (2.5%) 2 (3.3%) Maternal Age (completed years) (n = 81) (n = 60) Mean 38,42 38,53 Standard deviation 8,93 8,63 Caregiver education level (n = 81) (n = 60) Incomplete Elementary 9 (11.1%) 6 (10.0%) Complete Elementary 5 (6.2%) 4 (6.7%) Incomplete High School 8 (9.9%) 5 (8.3%) Complete High School 26 (32.1%) 19 (31.7%) Incomplete Higher Education 5 (6.2%) 3 (5.0%) Complete Higher Education 28 (34.6%) 23 (38.3%) Type of residence (n = 81) (n = 60) House 64 (79.0%) 46 (76.7%) Apartment 17 (21.0%) 14 (23.3%) Number of rooms in the residence (n = 81) (n = 60) 0-3 6 (7.4%) 5 (8.3%) 4-6 56 (69.1%) 39 (65.0%) 7-9 16 (19.8%) 14 (23.3%) 10 3 (3.7%) 2 (3.3%) Socioeconomic Classification (n = 81) (n = 60) D-E 2 (2.5%) 2 (3.3%) C2 21 (25.9%) 11 (18.3) C1 27 (33.3%) 25 (41.7%) B2 23 (28.4%) 17 (28.3%) B1 5 (6.2%) 3 (5.0%) A 3 (3.7%) 2 (3.3%) Legend: n = number of participants. When fill in the EF participants should indicate the type of social distancing child and caregiver were adopting: no social distancing (not avoiding external physical contact); partially social distancing (only leaving home to access food, medication, therapies, or medical consultation); or total social distancing (do not leave the house under any circumstances). Also, they completed the time of social distancing caregiver was performing (Was not distancing; 0 to 1 month; to 2 months; 2 to 3 months; 3 to 4 months; More than 4 months) and if the child was undergoing in-person therapy during the pandemic (yes or no). These results are presented on Table 2.Table 2. Characterization of Pandemic period and level of physical activity. Assessment I - n (%) Assessment II - n (%) Type of Social Distancing (Child) (n = 81) (n = 60) Total 20 (24.7%) 0 (0.0%) Partial 59 (72.8%) 55 (91.7%) Was not in distancing 2 (2.5%) 5 (8.3%) Type of Social Distancing (Caregiver) (n = 81) (n = 60) Total 7 (8.6%) 0 (0.0%) Partial 70 (86.4%) 53 (88.3%) Was not in distancing 4 (4.9%) 7 (11.7%) Social Distancing Time (Caregiver) (n = 81) (n = 60) Was not distancing 4 (4.9%) 0 (0.0%) 0 to 1 month 1 (1.2%) 16 (26.7%) 1 to 2 months 1 (1.2%) 4 (6.7%) 2 to 3 months 39 (48.1%) 6 (10.0%) 3 to 4 months 1 (1.2%) 34 (56.7%) More than 4 months 35 (43.2%) 0 (0.0%) Was the child undergoing in-person therapy during the pandemic? (n = 81) (n = 60) Yes 41 (50.6%) 36 (60.0%) No 40 (49.4%) 24 (40.0%) IPAQ - Short version (n = 71) (n = 58) Very Active 10 (14.1%) 13 (22.4%) Active 26 (36.6%) 18 (31.0%) Insufficiently Active A 15 (21.1%) 10 (17.2%) Insufficiently Active B 14 (19.7%) 11 (19.0%) Sedentary 6 (8.5%) 6 (10.3%) Legend: n = number of participants. EF: IPAQ- Short Version The level of physical activity was remotely accessed via EF using the International Physical Activity Questionnaires - short form (IPAQ-short form). IPAQ is a global instrument, widely used to assess physical activity level (Lima et al., 2019), and validated for Brazilians (Guedes et al., 2005). By means of eight standardized questions, IPAC assesses the frequency, duration, and intensity of activities performed by the individual, in the week prior to the assessment date. The activities questioned involve, 'light physical activity; 'moderate physical activity, and 'vigorous physical activity. The results obtained were classified as 'inactive' (sedentary), 'insufficiently active B', 'insufficiently active A', 'active' and 'very active' (Lima et al., 2019; Melo et al., 2016). IPAC scores were increasingly categorized for statistical analyses. Table 2 presents these results. Telephone interview: YC-PEM/PEM-CY The Young Children's Participation and Environment Measure (YC-PEM) and the Children's Participation and Environment Measure (PEM-CY) are corresponding instruments that assess the frequency and involvement in participation (Coster et al, 2011, 2013, 2014; Bedell et al, 2011, Khetani et al, 2014). They can be used, respectively, in infants/children and children/adolescents aged between 0-5 years and between 6-17 years, with typical development or with any type of developmental disabilities, including physical, cognitive, or emotional, and must be answered by parents or caregivers (Coster et al, 2011). The present study used the versions with translation and cultural adaptation for Brazil (Galvão et al., 2018; Silva Filho et al., 2019). Only part of participation at the home environment was used, considering the situation of social distancing imposed by the COVID-19 pandemic. Both versions have the same structure, varying only in the number of questions, and are composed of two parts: (a) participation, which involves the constructs frequency and involvement, and (b) environment, which involves characteristics of supports and barriers and the availability of services and resources. For part (a) ‘participation', the YC-PEM instrument has 13 items in the home session and the PEM-CY, 10 items. Thus, for each type of activity, it is asked (1) How often the child/adolescent participated in certain situations over the past 4 months; (2) How involved the child/adolescent is when they participate in 1 or 2 of the activities they perform most frequently. In order to normalize the YC-PEM and PEM-CY data considering the different number of questions, and allow to use as a same measure variable, the mean frequency score was used, which is obtained by dividing the total frequency score (sum of the scores from all items) by the total number of items of each version (YC-PEM: 13; PEM-CY: 10). For involvement in each activity (2), the mean involvment score was also calculated dividing the total involvment score (sum of the scores from all items) by the total number of items of each version (YC-PEM: 13; PEM-CY: 10). All scores (mean frequency, mean involvement) for the home environment were used in the statistical analyses. Higher scores indicate, respectively, greater frequency and greater involvement. See Table 3.Table 3. Inferential data, mean and standard deviation of the variables analyzed in the first and second assessments during the COVID-19 pandemic. IPAQ AI IPAQ AII Frequency A1 Frequency AII Involvement AI Involvement AII Social Support A1 Social Support AII PedsQl - V4.0 A1 PedsQl - V 4.0 AII N Statistic 71 58 69 53 69 53 78 58 69 57 Minimum Statistic 1 1 6 6 2 3 31 24 17 15 Maximum Statistic 5 5 8 7 5 6 95 95 84 85 Mean Statistic 3.28 3.36 6.51 6.45 4.31 4.36 62.01 63.12 53.67 50.56 Std. Deviation Statistic 1.185 1.307 .342 .289 .544 .610 19.619 21.386 16.120 17.530 Variance Statistic 1.405 1.709 .117 .084 .296 .372 384.896 457.371 259.859 307.284 Kurtosis Statistic −.811 -1.019 1.635 −.470 .897 .206 −1.211 −1.468 −.690 −.696 Std. Error .563 .618 .570 .644 .570 .644 .538 .618 .570 .623 Z −.141b −1.184c −.994b −.363c −2.386c Asymp. Sig. (2-tailed) .888 .236 .320 .716 .017 Legend: AI = Assessment I; AII = Assessment II; IPAQ = International Physical Activity Questionnaire, short version; PedsQL = Pediatric Quality of Life Inventory; a = Wilcoxon Signed Ranks Test, b = Based on negative ranks, c = Based on positive ranks. EF: Social Support Scale The Social Support Scale measures the social support provided to the main caregiver of the child/adolescent. This scale is composed of 19 items that assess, according to validation for the Brazilian population, three dimensions of social support: positive social interaction/affective support; emotional/information support; and material support and has high internal consistency in all its domains (Griep et al., 2005). For each item, the respondent answers, on a scale from 1 to 5 points, how much he/she considers that he/she has that particular social support, in the frequency of “never” (1); “rarely” (2); “sometimes” (3); “almost always” (4); and “always” (5) (Griep et al., 2005). The final score is obtained by summing the points of all items, and this raw value was used in the statistical analysis. Thus, higher scores indicate that the respondent has greater social support. See Table 3. EF: PedsQLTM V. 4.0 PedsQLTM V. 4.0 Generic Core Scales module, with the parent proxy versions, was used to assess the health-related quality of life of the children/adolescents with developmental disabilities based on the responsible person's report (Klatchoian et al., 2008). The instrument has adequate internal consistency for the Brazilian population (Klatchoian et al., 2008). There are specific versions for age groups: 13-18 years (23 items), 8-12 years (23 items), 5-7 years (23 items), and 2-4 years (21 items). Different versions only change the examples in survey questions, consistent with the age gap evaluated. The answer options are the same as for the other module. Versions corresponding to the age of the participants were used. Items are reversed scored and linearly transformed to a 0-100 scale as follows: 0 = 100, 1 = 75, 2 = 50, 3 = 25, 4 = 0, and the Total Score was obtained from the sum of all the items over the number of items answered on all the Scales. Higher scores indicate the higher health-related quality of life of children. See Table 3. Statistical Analysis The Kolmogorov-Smirnov test attested the non-normal pattern of data distribution. Descriptive values across assessments were obtained by the mean and standard deviation in Excel® spreadsheets. Significant changes in outcome variables between AI and AII were investigated using the Wilcoxon test with Bonferroni correction (Armstrong, 2014). For all analyses, a significance level of 0.01 was adopted (0.05 / number of comparisons made: 0.05/5= 0.01). Statistical Package for Social Science (SPSS®, version 24.0, Chicago, IL, USA) was used to perform the analyses. We used only cases with valid data and missing values are treated as missing using the pairwise deletion method. Results One hundred twenty-four mothers of children and adolescents with developmental disabilities were invited to participate in the research. In assessment I (AI), 81 mothers participated, and 60 mothers agreed to continue in the research (assessment II - AII). Figure 1 illustrates the flowchart of the study about participants. The participating children and adolescents were mostly male (AI= 63%; AII= 70%) and diagnosed with Down syndrome (AI= 42%; AII= 38.3%). The characterization data of the total sample can be found in Table 1.Figure 1. Study flowchart. Legend: n = number of participants. In the AI, most children and adolescents (72.8%) and caregivers (86.4%) were in partial distancing. This distancing regime was also observed in the second assessment, with 91.7% of children/adolescents and 88.3% of caregivers considering being in partial distancing. Descriptive data and measures referring to the pandemic experience can be seen in Table 2. We did not find significant changes on outcome variables addressing aspects of functioning between AI and AII. Table 3 indicates the results of descriptive and statistics tests. Discussion We aimed to investigate if 4 months of social distancing during a period of high contamination rate in the year 2020 in Brazil changed aspects of the functioning of children and adolescents with developmental disabilities. There was no significant change in the physical activity levels of the children evaluated. Our results do not agree with those reported by other authors, who point to a significant reduction in the level of physical activity in this population (Theis et al., 2021; Suarez-Balcazar et al., 2021; Marinho et al., 2019). Still, in the first assessment, we could see that, in our sample, about 48% of the participants already had an insufficient level of physical activity, a percentage that changed to 46% in the second assessment. So, we can notice that there were no significant changes. In this way, the characteristics of the pandemic peak scenario did not contribute to reducing these levels, since they were already too low in the first assessment. In any case, this high occurrence of insufficient physical activity in the population of children and adolescents with developmental disabilities draws attention to the importance of health professionals in guiding this population, seeking to ensure satisfactory levels of regular physical activity, even in the home environment, for example, proposing active family games, with balls, circuits of activities, or dance. The study by Masi et al. (2021) identified that, during the pandemic, children were viewing more television and digital media and exercising less. Considering that the level of physical activity may be related to mental health in children and adolescents with typical development during the pandemic (Okuyama et al., 2021), it is important that both aspects are evaluated and monitored. We also did not observe changes in the frequency of participation or in the involvement of individuals when participating in activities in home environments in the period evaluated. However, a previous study reported a significant increase in participation in the home of this population, soon after the immediate onset of the pandemic (Brugnaro et al., 2021). Therefore, when analyzing the results of the present study, we can infer that although the pandemic scenario has immediately changed numerous components of children's functioning (Theis et al., 2021; Suarez-Balcazar et al., 2021; Brugnaro et al., 2021; Marinho et al., 2019), including the participation at home, its continuity does not seem to have contributed to an even greater decrease/increase in activities and participation components during the 4 months of high contamination rate in the year 2020. So, the changes caused immediately after the social distancing started, these domains tend to keep constant. We can also think that, for these families, the routine has an important role, being more difficult to achieve some changes throughout the pandemic, having probably changed more soon after its sudden onset. Considering the social support received by the caregiver, no significant changes were identified. It is observed that the participants scored, on average of 65.27% in the first assessment, and 66.44% in the second assessment. It is known that family members of children with disabilities generally have less social support (Hassanein et al., 2021), in line with our data. In addition, the pandemic brought several challenges to the entire population, and this did not change the social support received by participating caregivers - neither more nor less. One might think that the support network, despite being small, can be considered loyal, even during the peak of the pandemic. Considering aspects of quality of life related to health, no changes were found between the assessments. We can observe that the quality of life was slightly above 50 points in both assessments, with a variable score between 0 and 100, which represents approximately half of the maximum possible value. The literature indicates that children with disabilities have lower quality of life when compared to typical peers (Voll, 2001; Dahan-Oliel, et al., 2012). Thus, during the period of the pandemic, the constant dedication of the family to maintain the routine and the conditions of care, were able to maintain the quality of life between the assessments. An important aspect is that the characterization data of the evaluated individuals shows that at the end of the second assessment, more than 90% of the children assessed were in partial isolation. We can highlight some important notes about our sample, because the participants have specific characteristics that can influence the results found. For example, our sample was composed for mothers with advanced age (mean age about 38 years), and according to previous studies, mothers with advanced age (Silva et al., 2008) can positively impact the global development of the child. Also, the maternal schooling of our sample was predominantly complete higher education and complete high school, which can be considered a high level of education, and study has indicated that higher mother's level of education can indirectly influence the child-mother interaction (Chai & Choi, 2021), facilitating the period to face the pandemic. Another important aspect is that almost 80% of participants live in a house, which can provide space for child play and do things in an open place. In addition, the majority of children and caregivers were in partial social distancing, which can reduce the impact of distancing time, and half or more were doing physical therapy during this time. So, the set of these factors may have acted as environmental facilitators, and have favored the results of no change in the aspects of functioning of the evaluated children. Limitation e strengths This study had several limitations. First, we did not have data about theses aspects before pandemic start, therefore, our results are limited only at the two points in time in the midst of the pandemic. Also, different types of developmental disabilities were included, which may have hampered the findings. In addition, it was not investigated whether the family and the child were receiving telehealth guidelines or interventions, which could help better cope with the pandemic situation. We did not verify which environmental factors could have influenced the absence of changes in the components evaluated. Thus, environmental facilitators may have driven this outcome. Also, the study included participants from one community in one country and can not be generalized to others contexts. Finally, the different experiences of social distancing presented by the participants may be a bias in the study, and during the second assessment, no family was in complete social distancing. Moreover, the partial social distancing is an aspect that is difficult to control and may vary between families. We highlight that, being a longitudinal study, we had 25% of drop out rate. Therefore, although 75% of the assessed families did not show significant changes in the functioning of their children, the results would have been different if all the families had participated in the second assessment. We consider as a strength the fact that the assessment took place in the months that comprised the height of the pandemic in the country where the research was carried out. Thus, it was possible to evaluate these aspects at a time when the pandemic was the most critical, with the greater need for social distancing, allowing the identification of biopsychosocial aspects at that specific moment. In addition, because the sample is comprehensive, it provides an overview of children with developmental disabilities, which can guide further investigations on the subject, specific to each health condition. Another strength is that our study evaluated different aspects of functioning for the same population, unlike other studies that evaluate isolated components, for example, quality of life and level of physical activity. When evaluating everything for the same population, the biopsychosocial approach is favored. Conclusions Considering the characteristics of our sample, and the aspects of functioning we proposed to analyze, there were no changes between the 4 months during a period of high contamination rate in the year 2020 in Brazil. These results, however, do not exclude the importance of constant assessments of this population, in order to monitor their functioning aspects, and better cope with the pandemic. Acknowledgements We thank all participants of the research, FAPESP and CAPES for financial support and the Brazilian Institutions Amor pra Down, Mano Down, Avança Down, Amigos Especiais de Limeira, Associação de Reabilitação Infantil Limeirense (ARIL), Associação de Pais e Amigos dos Excepcionais (APAE) do Estado de São Paulo, Casa da Criança Paralítica de Campinas, Centro Municipal de Atendimento Educacional Especializado (CMAEE) de Piraquara (Paraná) and Radio Magnificat de Limeira (São Paulo) for the support and dissemination of the research. ORCID iDs Beatriz Helena Brugnaro https://orcid.org/0000-0001-7883-3123 Gesica Fernandes https://orcid.org/0000-0003-1390-2284 Fabiana Nascimento Vieira https://orcid.org/0000-0001-7699-3094 The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the São Paulo Research Foundation (FAPESP) under Grant [number 2019/13570-6, 2019/13716-0, 2020/05685-5 and 2021/15016-6] and by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) [Finance Code 001]. Ethics statement: This study was approved by the local ethics committee (protocol number: 31786920.8.1001.5504). ==== Refs References Agarwal Gupta N. Kabra M. (2014). Diagnosis and management of down syndrome. Indian journal of pediatrics, 81 (6 ), pp. 560–567. doi:10.1007/s12098-013-1249-7.24127006 Al Awaji N.N. Almudaiheem A.A. Mortada E.M. (2021). Assessment of caregivers’ perspectives regarding speech-language services in Saudi Arabia during COVID-19. PLOS ONE, 16 (6 ), p. e0253441. doi:10.1371/journal.pone.0253441.34157033 Almeida I.L.de L. Rego J.F. Teixeira A.C.G. Moreira M.R. (2021). Social Isolation and Its Impact on Child and Adolescent Development: A Systematic Review. Revista Paulista de Pediatria, [online], 40 , p. e2020385. doi:10.1590/1984-0462/2022/40/2020385.34614137 Aquino E.M.L. Silveira I.H. Pescarini J.M. Aquino R. Souza-Filho J.A. de Rocha A.dos S. Ferreira A. Victor A. Teixeira C. Machado D.B. Paixão E. Alves F.J.O. Pilecco F. Menezes G. Gabrielli L. Leite L. Almeida M.da C.C.de Ortelan N. Fernandes Q.H.R.F. Ortiz R.J.F. (2020). Medidas de distanciamento social no controle da pandemia de COVID-19: potenciais impactos e desafios no Brasil. Ciência & Saúde Coletiva, [online], 25 (1 ), , pp. 2423–2446. doi:10.1590/1413-81232020256.1.10502020.32520287 Armstrong R.A. (2014). When to use the Bonferroni correction. Ophthalmic and Physiological Optics, [online] 34 (5 ), pp.502–508. doi:10.1111/opo.12131.24697967 Aviram R. Harries N. Rabani A.S. Amro A. Nammourah I. Al-Jarrah M. Raanan Y. Hutzler Y. Bar-Haim S. (2019). Comparison of Habitual Physical Activity and Sedentary Behavior in Adolescents and Young Adults With and Without Cerebral Palsy. Pediatric Exercise Science, [online], 31 (1 ), pp.60–66. doi:10.1123/pes.2017-0285.30272530 Batista T.S.C. Rebouças D.A. de A. Almeida L.A.da H. Santana R.S. Fontes Junior J. Santos G.B. Silva É.R. Tofani P.S. (2021). Telemonitoring in SUS users by physiotherapy students and physiotherapist during the COVID-19 pandemic: experience report. Brazilian Journal of Health Review, 4 (3 ), pp.11071–11082. doi:10.34119/bjhrv4n3-114. Baird G. Cass H. Slonims V. (2003). Diagnosis of autism. BMJ (Clinical research ed.), 327 (7413 ), 488–493. doi:10.1136/bmj.327.7413.488. Bedell G.M. Khetani M.A. Cousins M.A. Coster W.J. Law M.C. (2011). Parent Perspectives to Inform Development of Measures of Children’s Participation and Environment. Archives of Physical Medicine and Rehabilitation, 92 (5 ), pp.765–773. doi:10.1016/j.apmr.2010.12.029.21530724 Brugnaro B. H. Vieira F. N. Fernandes G. de Camargo O. K. Fumincelli L. de Campos A. C. Pavão S. L. Rocha N. A. C. F. (2022). Exploration of the Feasibility of Remote Assessment of Functioning in Children and Adolescents with Developmental Disabilities: Parents' Perspectives and Related Contextual Factors. International journal of environmental research and public health, 19 (22 ), pp. 15101. doi:10.3390/ijerph192215101.36429822 Brugnaro B.H. de Camargo O.K. Corsi C. de Campos A.C. Fernandes G. Pavão S.L. Rocha N.A.C.F. (2021). Functioning of children and adolescents with Down syndrome and the association with environmental barriers and facilitators during the COVID-19 pandemic. Journal of Intellectual Disabilities, p. 174462952110327. doi:10.1177/17446295211032763. Castro G.G. de Nascimento L.C.G. do Figueiredo G.L.A. (2020). Applicability of the ICF-CY in evaluating children with disabilities and family support: an integrative literature review. Revista CEFAC, [online] 22 (1 ). doi:10.1590/1982-0216/202022111518. Chai HY Choi MY . (2021). [Structural Equation Modeling of the Interaction between Mothers and Preschool-Age Children]. J Korean Acad Nurs, 51 (5 ), pp. 630–642. Korean. doi: 10.4040/jkan.21096. PMID: 34737254.34737254 Coster W. Bedell G. Law M. Khetani M.A. Teplicky R. Liljenquist K. Gleason K. Kao Y.-C. (2011). Psychometric evaluation of the Participation and Environment Measure for Children and Youth. Developmental Medicine & Child Neurology, 53 (11 ), pp. 1030–1037. doi:10.1111/j.1469-8749.2011.04094.x.22014322 Coster W. Law M. Bedell G. Anaby D. Khetani M. Teplicky R. (2014). PEM-CY User's Guide (Version 1.1). Coster W. Law M. Bedell G. Liljenquist K. Kao Y.-C. Khetani M. Teplicky R. (2013). School participation, supports and barriers of students with and without disabilities. Child: Care, Health and Development, 39 (4 ), pp. 535–543. doi:10.1111/cch.12046.23763254 Dahan-Oliel N. Shikako-Thomas K. Majnemer A. (2012). Quality of life and leisure participation in children with neurodevelopmental disabilities: a thematic analysis of the literature. Quality of life research: an international journal of quality of life aspects of treatment, care and rehabilitation, 21 (3 ), pp. 427–439. doi:10.1007/s11136-011-0063-9.22101860 Einarsson I.Ó. Ólafsson Á. Hinriksdóttir G. Jóhannsson E. Daly D. Arngrímsson S.Á. (2015). Differences in Physical Activity among Youth with and without Intellectual Disability. Medicine & Science in Sports & Exercise, 47 (2 ), pp. 411–418. doi:10.1249/mss.0000000000000412.24983335 Eysenbach G. (2004) Improving the quality of Web surveys: the Checklist for Reporting Results of Internet E-Surveys (CHERRIES). Journal of medical Internet research 6 (3 ): e34. Available at: 10.2196/jmir.6.3.e34.15471760 Field D. A. Roxborough L. A. (2012). Validation of the relation between the type and amount of seating support provided and Level of Sitting Scale (LSS) scores for children with neuromotor disorders. Developmental neurorehabilitation, 15 (3 ), pp. 202–208. doi:10.3109/17518423.2012.673177.22582851 Field D. A. Roxborough L. A. (2011). Responsiveness of the Seated Postural Control Measure and the Level of Sitting Scale in children with neuromotor disorders. Disability and rehabilitation. Assistive technology, 6 (6 ), pp. 473–482. doi:10.3109/17483107.2010.532285.21110727 French J.A. Brodie M.J. Caraballo R. Devinsky O. Ding D. Jehi L. Jette N. Kanner A. Modi A.C. Newton C.R. Patel A.A. Pennell P.B. Perucca E. Sander J.W. Scheffer I.E. Singh G. Williams E. Wilmshurst J. Cross J.H. (2020). Keeping people with epilepsy safe during the COVID-19 pandemic. Neurology, 94 (23 ), pp. 1032–1037. doi:10.1212/wnl.0000000000009632.32327490 Galvão É.R.V.P. Cazeiro A.P.M. De Campos A.C. Longo E. (2018). Participation and Environment Measure for Children and Youth (PEM-CY): cross-cultural adaptation for use in Brazil. Revista de Terapia Ocupacional da Universidade de São Paulo, 29 (3 ), pp. 237–245. doi:10.11606/issn.2238-6149.v29i3p237-245. Griep R.H. Chor D. Faerstein E. Werneck G.L. Lopes C.S. (2005). Construct validity of the Medical Outcomes Study's social support scale adapted to Portuguese in the Pró-Saúde Study. Cadernos de Saúde Pública, [online], 21 , pp.703–714. doi:10.1590/S0102-311X2005000300004.15868028 Guedes D.P. Lopes C.C. Guedes J.E.R.P. (2005). Reproducibility and validity of the International Physical Activity Questionnaire in adolescents. Revista Brasileira de Medicina do Esporte, 11 (2 ), pp.151–158. doi:10.1590/s1517-86922005000200011. Hassanein EEA Adawi TR Johnson ES . (2021). Social support, resilience, and quality of life for families with children with intellectual disabilities. Res Dev Disabil, 112 , pp. 103910. doi: 10.1016/j.ridd.2021.103910. Epub 2021 Feb 20. PMID: 33621821.33621821 Halberstadt B.F. Moraes A.B. de Souza A.P.R. de (2019). Evaluation of children with Down’s Syndrome through ICF-CJ: comparison of parents ‘and therapists’ vision. Distúrbios da Comunicação, 31 (3 ), pp. 454–464. doi:10.23925/2176-2724.2019v31i3p454-464. Holanda C.M. de A. Andrade F.L.J.P. de Bezerra M.A. Nascimento J.P. da S. Neves R. da F. Alves S.B. Ribeiro K.S.Q.S. (2015). Support networks and people with physical disabilities: social inclusion and access to health services. Ciência & Saúde Coletiva, [online], 20 (1 ), pp. 175–184. doi:10.1590/1413-81232014201.19012013.25650611 Isa S.N.I. Ishak I. Ab Rahman A. Mohd Saat N.Z. Che Din N. Lubis S.H. Mohd Ismail M.F. (2016). Health and quality of life among the caregivers of children with disabilities: A review of literature. Asian Journal of Psychiatry, [online], 23 , pp. 71–77. doi:10.1016/j.ajp.2016.07.007.27969083 Khetani M. Marley J. Baker M. Albrecht E. Bedell G. Coster W. Anaby D. Law M. (2014). Validity of the Participation and Environment Measure for Children and Youth (PEM-CY) for Health Impact Assessment (HIA) in sustainable development projects. Disability and Health Journal, 7 (2 ), pp. 226–235. doi:10.1016/j.dhjo.2013.11.003.24680052 Klatchoian D.A. Len C.A. Terreri M.T. Silva M. Itamoto C. Ciconelli R.M. Varni J.W. Hilário M.O. (2008). Quality of life of children and adolescents from São Paulo: reliability and validity of the Brazilian version of the Pediatric Quality of Life Inventory version 4.0 Generic Core Scales.Jornal de pediatria, 84 (4 ), pp. 308–315. doi:10.2223/JPED.1788.18679557 Kramer J.M. Olsen S. Mermelstein M. Balcells A. Liljenquist K. (2012). Youth with disabilities' perspectives of the environment and participation: a qualitative meta-synthesis. Child: care, health and development, 38 (6 ), pp. 763–777. doi:10.1111/j.1365-2214.2012.01365.x.22372695 Kuru N. Piyal B. (2018). Perceived social support and quality of life of parents of children with Autism. Nigerian Journal of Clinical Practice, [online], 21 (9 ), pp. 1182–1189. doi:10.4103/njcp.njcp_13_18.30156205 Lima M.F.C. de Lopes P.R.N.R. Silva R.G. Faria R.C. de Amorim P.R. dos S. Marins J.C.B. (2019). Questionnaires to assess the habitual physical activity level among Brazilian adolescents: a systematic review. Revista Brasileira de Ciências do Esporte, [online], 41 , pp. 233–240. doi:10.1016/j.rbce.2018.03.019. Lippi G. Henry B.M. Bovo C. Sanchis-Gomar F. (2020). Health risks and potential remedies during prolonged lockdowns for coronavirus disease 2019 (COVID-19). Diagnosis, 7 (2 ), pp. 85–90. doi:10.1515/dx-2020-0041.32267243 Marinho A.R. Severo M. Correia D. Lobato L. Vilela S. Oliveira A. Ramos E. Torres D. Lopes C. (2019). Total, added and free sugar intakes, dietary sources and determinants of consumption in Portugal: The National Food, Nutrition and Physical Activity Survey (IAN-AF 2015–2016). Public Health Nutrition, 23 (5 ), pp. 869–881. doi:10.1017/s1368980019002519.31486357 Masi A. Mendoza Diaz A. Tully L. Azim S.I. Woolfenden S. Efron D. Eapen V. (2021). Impact of the COVID ‐19 pandemic on the well‐being of children with neurodevelopmental disabilities and their parents. Journal of Paediatrics and Child Health, 57 (5 ). doi:10.1111/jpc.15285. Nussbaumer-Streit B. Mayr V. Dobrescu A.I. Chapman A. Persad E. Klerings I. Wagner G. Siebert U. Christof C. Zachariah C. Gartlehner G. (2020). Quarantine alone or in combination with other public health measures to control COVID-19: a rapid review. Cochrane Database of Systematic Reviews, (4 ). doi:10.1002/14651858.cd013574. Okuyama J. Seto S. Fukuda Y. Funakoshi S. Amae S. Onobe J. Izumi S. Ito K. Imamura F. (2021). Mental Health and Physical Activity among Children and Adolescents during the COVID-19 Pandemic. The Tohoku Journal of Experimental Medicine, 253 (3 ), pp. 203–215. doi:10.1620/tjem.253.203.33775993 Silva N.C.B. Nunes C.C. Betti M.C.M. Rios K.S.A . (2008). Variáveis da família e seu impacto sobre o desenvolvimento infantil. Temas em Psicologia, 16 (2 ), pp. 215–229. Retrieved October 07, 2022, from http://pepsic.bvsalud.org/scielo.php?script=sci_arttext&pid=S1413-389X2008000200006&lng=en&tlng= Silva Filho J.A.da Cazeiro A.P.M. Campos A.C.de Longo E. (2019). Young Children’s Participation and Environment Measure (YC-PEM): translate and cross-cultural adaptation for use in Brazil. Revista De Terapia Ocupacional Da Universidade De São Paulo, 30 (3 ), pp. 140-149. doi:10.11606/issn.2238-6149.v30i3p140-149. Silva L.L.S. da Lima A.F.R. Polli D.A. Razia P.F.S. Pavão L.F.A. Cavalcanti M.A.F. de H. Toscano C.M. (2020). Social distancing measures in the fight against COVID-19 in Brazil: description and epidemiological analysis by state. Cadernos de Saúde Pública, [online], 36 (9 ). doi:10.1590/0102-311x00185020. Suarez-Balcazar Y. Mirza M. Errisuriz V.L. Zeng W. Brown J.P. Vanegas S. Heydarian N. Parra-Medina D. Morales P. Torres H. Magaña S. (2021). Impact of COVID-19 on the Mental Health and Well-Being of Latinx Caregivers of Children with Intellectual and Developmental Disabilities. International Journal of Environmental Research and Public Health, 18 (15 ), p. 7971. doi:10.3390/ijerph18157971.34360263 Summers D.J. Cheng D.H.-Y. Lin P.H.-H. Barnard D.L.T. Kvalsvig D.A. Wilson P.N. Baker P.M.G. (2020). Potential lessons from the Taiwan and New Zealand health responses to the COVID-19 pandemic. The Lancet Regional Health – Western Pacific, [online], 0 (0 ). doi:10.1016/j.lanwpc.2020.100044. Theis N. Campbell N. De Leeuw J. Owen M. Schenke K.C. (2021). The Effects of COVID-19 Restrictions on Physical Activity and Mental Health of Children and Young Adults with Physical and/or Intellectual Disabilities. Disability and Health Journal, [online], 14 (3 ), p.101064. doi:10.1016/j.dhjo.2021.101064.33549499 Voll R. (2001). Aspects of the quality of life of chronically ill and handicapped children and adolescents in outpatient and inpatient rehabilitation. International journal of rehabilitation research. Internationale Zeitschrift fur Rehabilitationsforschung. Revue internationale de recherches de readaptation, 24 (1 ), pp. 43–49. doi:10.1097/00004356-200103000-00006.11302463
PMC010xxxxxx/PMC10290937.txt	==== Front Tour Hosp Res Tour Hosp Res spthr THR Tourism and Hospitality Research 1467-3584 1742-9692 SAGE Publications Sage UK: London, England 10.1177_14673584231184161 10.1177/14673584231184161 Article Benefit-triggered or trust-guided? Investigation of customers’ perceptions towards AI-adopting hotels amid and post COVID-19 pandemic https://orcid.org/0000-0002-8836-8939 Ghazi Karam 648641 The Higher Institute for Tourism and Hotels in Alexandria (EGOTH) , Egypt Kattara Hanan Alexandria University, 496089 Faculty of Tourism and Hotels , Egypt https://orcid.org/0000-0002-9483-3550 Salem Islam Elbayoumi College of Economics and Business Administration, University of Technology and Applied Sciences, Salalah, Oman and Department of Hotel Management, Faculty of Tourism and Hotels, Alexandria University , Alexandria, Egypt Shaaban Mohammad Nabil University of Prince Mugrin, College of Business and Tourism, International Hospitality Management Department, Saudi Arabia, and Alexandria University , Faculty of Tourism and Hotels, Alexandria, Egypt Karam Ghazi, High Institute of Tourism and Hotels in Alexandria (EGOTH), Alexandria 55555, Egypt. Email: dr.karam.ghazi@egoth-alex.edu.eg 20 6 2023 20 6 2023 14673584231184161© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. This research exhibits and empirically validates an expansion of the Unified Theory of Acceptance and Use of Technology (UTAUT2) and integrates artificial intelligence (AI) and pandemic threats to explain customers’ utilitarian-versus-emotional behavioral intentions towards AI-adopting hotels amid and post-COVID-19. Utilizing data gathered from 416 customers, the findings confirmed that customers’ perceived importance of AI amid and post-COVID-19 has a direct positive effect on their behavioral intentions towards hotels adopting those technologies, with perceived benefits of technology playing a more significant mediating role than customers’ trust intervening in that correlation. This provides evidence for the utilitarian perception of customers during crises and offers updated insights into the dynamics that constitute and trigger hotel customers’ behavioral intentions toward AI. The results provide hoteliers with a valid understanding and rationalization of how to utilize AI to address customers’ crucial concerns and interests amid and post-COVID-19 and in similar crises. Artificial intelligence behavioral intentions customer trust perceived benefits UTAUT2 edited-statecorrected-proof typesetterts10 ==== Body pmcIntroduction Artificial intelligence (AI) has been widely used in the contemporary hospitality industry and is deeply changing and reshaping the behaviors and experiences of both tourists and businesses (Zhong et al., 2020; Ivanov et al., 2020; Gursoy and Chi, 2020; Kim et al., 2021; Knani et al., 2022; Saydam, et al., 2022; Li, et al., 2022). From the perspective of hospitality businesses, as the supply side, the adoption of AI solutions improves service quality while increasing operational capability and efficiency, lowering costs, and consequently creating a competitive advantage (Lukanova and Ilieva, 2019). While providing interactive, personalized, and customized services to tourists can improve the customer experience (Lu, 2019), it cannot improve the supply side. AI frequently refers to machine learning, deep learning, robotics, the Internet of Things, and the use of big data (Lukanova and Ilieva, 2019). In hospitality companies, AI offers versatile applications, like biometric authentication (hence, biometrics), robotic process automation, virtual and augmented reality, self-service technologies, and so forth (Ivanov et al., 2020; Tussyadiah, 2020). AI applications are used to authenticate online bookings and check-in/out, access guestrooms and outlets, operate in-room functions, process inquiries and payments, facilitate luggage transfer, provide concierge services, clean and disinfect guestrooms and public areas, deliver foodservice, and pose as in-room assistants (Murphy and Rottet, 2009; Lukanova and Ilieva, 2019; Ivanov et al., 2020). However, there are some risks associated with tourists’ distrust, fear of interacting, security and privacy concerns, and human contact preferences (Tussyadiah, 2020; Li et al., 2022; Saydam et al., 2022; Knani et al., 2022). Moreover, the hospitality industry is highly vulnerable to outbreaks of epidemics and diseases (like COVID-19, SARS, H1N1, MERS, Ebola, etc.), which severely damage the industry and deeply affect the behaviors and experiences of both tourists and businesses in terms of safety, economic spending, conviction, and attitude (Ivanov et al., 2020; Zeng et al., 2020; Lin, Chi, & Gursoy, 2020; Kim et al., 2021). Particularly, the COVID-19 pandemic has urged the hotel industry to accelerate the exploitation of AI-contactless solutions to recover customer trust and encourage demand through keeping social and physical distance and lowering costs (Seyitoğlu and Ivanov, 2020; Romero and Lado, 2021; Perić, and Vitezić, 2021; Kim et al., 2021). Hence, increasing interest is directed toward investing in and adopting AI and its applications in hospitality enterprises to enrich the customer service experience and minimize pandemic-related health and economic risks (Ivanov et al., 2020; Gaur et al., 2021). Since AI solutions, in conjunction with the COVID-19 pandemic, have reformed customers' attitudes, expectations, and experiences, hotel marketers and operators must have a thorough understanding of their acceptance of such trends, particularly during and after the COVID-19 pandemic (Lin et al., 2020; Ivanov et al., 2020; Gaur et al., 2021; Lin et al., 2020; Perić, and Vitezić, 2021). Yet, there is a lack of empirical research on the role of AI to manage safety and health risks in hospitality settings amid and post-pandemic era (Ivanov, et al., 2020; Zhong et al., 2020; Zeng et al., 2020; Seyitoğlu and Ivanov, 2020). Accordingly, hospitality scholars have called for further research to understand the role of AI contactless solutions to assure and recover customer trust and safety amid and post-pandemic era (Hao et al., 2020; Romero and Lado, 2021; Pillai et al., 2021; Perić, and Vitezić, 2021; Lin et al., 2020; Gaur et al., 2021). There is no empirical study that investigates whether the COVID-19 pandemic obliges intelligent automation of hotel services, and there is also no dedicated investigation of whether customers embrace utilitarian or emotional perspectives towards staying in hotels during and post-COVID-19 and towards newly introduced initiatives. This research thus aims at determining how the pandemic compels hotel customers’ utilitarian versus emotional tendencies and their behavioral intentions toward AI-adopting hotels. Working towards that aim, this study employed a noteworthily extended utilization of Venkatesh et al. (2012)’s unified theory of acceptance and use of technology 2 (UTAUT2) to examine whether customers embrace utilitarian or emotional perspectives towards staying in hotels amid and post-COVID-19. That is, the literature still needs more studies that integrate customer trust, in addition to novel variables, into UTAUT2, in order to gain more structured, meaningful outcomes. There are limited studies that involved customers’ trust to reveal their behavioral intentions in the hospitality context in general (Jalilvand et al., 2017) and amid and post-pandemic in particular (Hao, 2021; Park and Tussyadiah, 2020; Ruan et al., 2020; Tussyadiah et al., 2020; Perić and Vitezić, 2021). The authors therefore adopted the perspective of Hao (2021), who perceived that some contactless technologies might provoke some uncertainties and trust issues among customers. Consequently, trust occupies an essential position and represents a focal addition to UTAUT2 in the current study’s aim to inquire whether hotel customers are more inclined to be utilitarian or emotional. The concepts of utilitarian versus emotional nature are comparable and equivalent to UTAUT2’s variables, and are most suitable to the current study’s objective. Hence, twofold purposes are attained. In addition to the intervening function of AI benefits, the model extends to examine the intervening and mediating role of “COVID-triggered customers’ trust in AI-adopting hotels” (hence, trust) towards their behavioral intentions. Trust represents a customer’s emotional perspective. To provide a more thorough comprehension of the role of technologies, probable moderating variables need to be investigated (Beatson, 2010; Salem & Čavlek, 2016). To recapitulate, extending the UTAUT2’s variables has the evident benefits of enriching the UTAUT2’s current stand by a novel set of variables that have not been included in the model before, suiting the current study’s needs and aims that would not otherwise be attained if UTAUT2 is utilized as it is, and, eventually, providing a solid base for future research endeavors to build upon for unreservedly integrating more contemporary variables, while simultaneously enjoying the well-structured skeleton of UTAUT2. Thus, unlike the original UTAUT2 model, the current study’s UTAUT2 equivalents of habit, social influence, and facilitating conditions will not act as independent variables, but rather will be scrutinized as moderators (as shown in Figure 1). This study and its proposed extended model will thus direct hoteliers in predicting customer trends concerning AI in hotels and in formulating and implementing effective strategies for the adoption of AI technologies during pandemic periods and similar crises.Figure 1. Conceptual model. Literature review and hypotheses development AI importance and Behavioral Intentions (BIs) Behavioral Intentions (BIs) refer to an individual’s likelihood of engaging in a specific behavior, which helps to predict the actual behavior in one’s decision-making process (Ajzen, 1991). Based on the Theory of Planned Behavior (Ajzen, 1991) and the Theory of Reasoned Action (Fishbein and Ajzen, 1977), prior studies have extensively confirmed that customer awareness and attitudes towards novel technology usage and importance have an influence on BIs (e.g., Shin and Kang, 2020; Zhong et al., 2020; De Kervenoael et al., 2020; Kim, et al., 2021). Shin and Kang (2020) revealed that lower expected interaction and better cleanliness associated with technology innovation positively affected hotel booking intention during the COVID-19 pandemic. The importance of personalization and entertainment, as well as the safety and security of smart hotels, had a leading role to shape customer behavior (Elkhwesky et al., 2022; Kim et al., 2021). Moreover, UTAUT2-based previous findings performance expectancy has significant effect on attitudes and intentions to use various innovative technologies (Chen et al., 2022; Hao, 2021; Jung and Cha, 2022; Morosan and DeFranco, 2016; Wu, et al., 2021). Thus, the AI importance’ (UTAUT2’s performance expectancy) role in shaping customers’ BI is evident. As such, we propose: H1: AI-importance has a positive and significant influence on BIs. AI-importance and customer trust COVID-19-provoked anxiety was a positive predictor of attitude towards virtual reality tourism (Talwar, et al., 2022). Empirical research emphasized that intelligent technologies, including AI, serve as objects of customer trust (Tussyadiah et al., 2020). Ruan et al. (2020) similarly showed that technological proficiency and service innovation accomplishment positively affect trust by customers’ perceived value. Based upon the interdisciplinary trust model proposed by McKnight and Chervany (2001), a series of studies identified that trust propensity has positive influences on developing trust beliefs toward intelligent technologies (e.g., Tussyadiah et al., 2020). Thus, the following hypothesis is proposed: H2: AI-importance has a positive and significant influence on customer trust. AI-importance and AI-benefits Several studies confirmed the positive effect of AI perceived importance (UTAUT2’s performance expectancy) on AI-benefits (UTAUT2’s effort expectancy) in the hospitality context (De Kervenoael et al., 2020; Ruan et al., 2020; Shin and Kang, 2020; Tussyadiah, 2020). In addition, Shin and Kang (2020) found that minimized interaction and high cleanliness associated with new technologies positively affected perceived health risks during the COVID-19 pandemic. Likewise, de Kervenoael et al. (2020) confirmed that robot characteristics positively affect AI perceived benefits. As such, and as a variation of UTAUT2, it is proposed that: H3: AI-importance has a positive and significant influence on AI-benefits. AI-benefits and customer trust Prior studies collectively established that the AI-benefits (UTAUT2’ effort expectancy) play an important positive role in forming customers’ perceived trust (Ameen, et al., 2021; Lee and Lee, 2019; Ruan et al., 2020). Specifically, Ameen et al. (2021) demonstrated the positive effect of perceived convenience and AI-enabled service quality on trust. Pai et al. (2018) confirmed that biometric technology benefits positively affected visitors’ perceptions of trust. Ruan et al. (2020) showed that customers’ perceived value of technological competence and service innovation implementation positively affect trust. Therefore, it is assumed that: H4: AI-benefits has a positive and significant influence on customer trust. Customer trust and Behavioral Intentions (BIs) Trust has been shown to be a strong determinant of BIs (e.g., Tussyadiah et al., 2020; Park and Tussyadiah, 2020; Ruan et al., 2020; Lee and Lee, 2019). Customer trust, as perceived by Martínez and Del Bosque (2013), represented a necessity for the healthy relationship with the hotel, represented in creating a positive attitude, loyalty, and satisfaction. Also, Cha (2020) confirmed that perceived trust has a significant positive effect on the intention to use restaurant robots, and Pai et al. (2018) found that perceived trust has a positive impact on the intention to use the biometric systems in the hospitality industry. Based upon the trust model proposed by McKnight and Chervany (2001), other studies supported the positive relationship between trust and positive intentions towards AI and service robots (Tussyadiah et al., 2020; Park and Tussyadiah, 2020). Consistently, the above studies supported the notion of the theory of planned action (Ajzen, 1991), implying that beliefs, particularly trust, are directly associated with corresponding intentions. Consequently, it is proposed that: H5: Customer Trust has a positive and significant influence on BIs. AI-benefits and Behavioral Intentions (BIs) In UTAUT2, AI-benefits, current study’s equivalent of effort expectancy, directly affect customers’ intentions towards new technological adoption (Jung and Cha, 2022). Consistently, AI-benefits are extensively researched to ascertain their positive effect on customers’ behaviors and attitudes, for instance, towards using social robots (De Kervenoael et al., 2020), for service robots in restaurants (Jung and Cha, 2022), and for touchless, mobile-phone-based payment in restaurants (Chen et al., 2022). Lin et al. (2020) highlighted that robots’ benefits support customers’ positive emotions toward AI robotics. As such, it is hypothesized that: H6: AI benefits have a positive and significant influence on BIs. The mediating effects of perceived AI-benefits Previous studies established that customers first need to perceive a novel technology application as beneficial to develop positive behavioral intentions towards it. AI-benefits worked as a mediating variable in the relationship between both AI technology and trust (Lee and Lee, 2019). Ruan et al. (2020) revealed that perceived value mediated the association between both technological provision and trust, and between service innovation accomplishment and trust. Pinxteren et al. (2019) exhibited that the influence of perceived trust on customers’ intentions to use humanoid service robots was fully mediated by customers’ perceived enjoyment. AI benefits’ mediating is thus recognized to influence the relationship among technological novelties and customers’ reactions, suggesting the subsequent hypotheses: H7a: AI-benefits positively and significantly mediate the relationship between AI-importance and customer trust. H7b: AI-benefits positively and significantly mediate the relationship between AI-importance and BIs. The mediating effects of customer trust in AI-adopting hotels Relevant studies supported trust as a mediator between AI importance and AI-benefits (Ameen et al., 2021; Ruan et al., 2020), as well as mediating the AI benefits correlation to behavioral intentions (Wang, et al., 2015; Lee and Lee, 2019). Ruan et al. (2020) revealed that trust mediates both the relationship between perceived value and brand image, and between perceived value and perceived quality. Ameen et al. (2021) found that trust also mediates the effects of convenience, personalization, and AI-enabled service quality on AI-enabled customer experience. Lee and Lee (2019) proved that brand trust mediates the relationship between customers’ engagement with branded hotel applications and brand loyalty. Thus, this efficaciously mediating role of trust is consistent with the trust model (McKnight and Chervany, 2001), suggesting the subsequent hypotheses: H8a: Customer trust positively and significantly mediates the association between AI-importance and BIs. H8b: Customer trust positively and significantly mediates the association between AI-benefits and BIs. The moderating effect of preferences for personal service (personal-service) Preferences for Personal Service (hence, personal-service) refers to the desire to interact with service employees during the service encounter (Dabholkar and Bagozzi, 2002). The influences of customers’ preferences for personal service, as the UTAUT2’s “Habit,” have been highlighted in previous studies, on customers satisfaction and, hence, commitment, (Beatson, 2010), and intentions (Chen et al., 2022; Gupta et al., 2018; Morosan and DeFranco, 2016). Prior studies provided empirical evidence of the negative link between personal-service and the benefits of technology (Ameen et al., 2021) and customer AI-related trust (Tussyadiah et al., 2020). For illustration, Shin and Kang (2020) exhibited a positive effect of expected interaction with employees on the perceived health risk of hotel customers during the COVID-19 pandemic. Ameen et al. (2021) demonstrated that perceived sacrifice due to a lack of human interaction has a negative direct effect on AI-enabled customer experience. It has been even reported that self-service represents a significant moderating pose in technology-related models (Beatson, 2010; Dabholkar and Bagozzi, 2002). The moderating role of personal-service is predicted, proposing the following hypotheses: H9a: Personal-service positively and significantly moderates the impact of AI-importance on customer trust. H9b: Personal-service positively and significantly moderates the impact of AI-importance on AI-benefits. The moderating effect of perceived corporate reputation Perceived Corporate Reputation (hence, reputation, resembling UTAUT2’s “Social Influence”) is defined as the customer’s overall evaluation of the organization’s past actions and expectations regarding its future actions, in view of its efficiency in relation to the main rivals (Walsh et al., 2009). Prior studies provided empirical evidence on the positive influence of reputation on trust in tourism and hospitality contexts (e.g., Jalilvand et al., 2017; Chang, 2013). Moreover, reputation has a significant impact on AI-benefits; represented as perceived value, service quality, satisfaction, and perceived trust (Perez-Aranda, et al., 2019; Chang, 2013; Dardeer et al., 2017). Thus, customers who perceive a hotel as being highly reliable are most likely to develop trust and a more positive perception of acquired AI-benefits. With this situation, and the pre-hypothesized impact of AI importance on trust and AI-benefits, the moderating role of reputation is predicted, proposing the following hypotheses: H10a: Reputation positively and significantly moderates the impact of AI-importance on customer trust. H10b: Reputation positively and significantly moderates the impact of AI-importance on AI-benefits. The moderating effect of perceived customer-company identification Perceived Customer-Company Identification (hence, identification, representing UTAUT2’s “Social Influence”) is defined as the extent to which customers perceive the company identity as trustworthy (Bhattacharya and Sen, 2003). Identification positively pushes towards customer loyalty, as a form of behaviroal intention (Gupta et al., 2018), commitment, more constant and long-term preference, setting customers more tolerant towards trivial alterations in the product/service (Bhattacharya and Sen, 2003), and accepting and utilizing modern technologies (Hao, 2021). Identification rationalizes the motives that inspire individuals to associate to the organization, and fostering similarities with its members and differences with non-members (Martínez and Del Bosque, 2013). Prior research has supported the association between identification and trust (So, et al., 2013; Rather, 2018). Other studies also indicated the positive impact of identification on AI benefits through improving their perceptions of service quality, perceived value, satisfaction and brand trust (So et al., 2013). Besides, identification’s intervention positive role on customer trust has been reported by Martínez and Del Bosque (2013). Therefore, customers who positively identify themselves with a particular hotel are most likely to build trust and promote a better perception of gained AI benefits. Based on this, the moderating role of identification is predicted, proposing the following hypotheses: H11a: Identification positively and significantly moderates the impact of AI-importance on customer trust. H11b: Identification positively and significantly moderates the impact of AI-importance on AI benefits. The moderating effect of perceived ease of use Perceived Ease of Use (hence, easiness, representing UTAUT2’s “Facilitating Conditions”) is defined as “the degree to which a person believes that using a particular system would be free from effort” (Davis, 1989). Frequently, models and theories integrated easiness as an antecedent of acceptance of technology, perceived usefulness, and intention to be adopted, since the easier it is to utilize technology, it is perceived as being more useful (Venkatesh, 2000). Thus, in order to properly adopt digitalization, it is imperative to comprehend the leading provision for easiness, as an essential moderating step towards developing and maintaining customers’ acceptance and due adoption (Wu et al., 2021). The literature frequently supported the positive effect of easiness on AI-benefits in hospitality (De Kervenoael et al., 2020). Hao (2021) and Chen et al. (2022) stated that facilitating conditions are a major antecedent of accepting and adopting novel technologies. The literature also provided support for the relationship between easiness and trust (Agag and El-Masry, 2016). For example, Pai et al. (2018) confirmed that the easiness of biometric technology positively affected visitors’ perceptions of trust. Consequently, customers who perceive a certain technological application as easy to use have the most potential to perceive it as more beneficial, and trust the organization providing it. Thus, the moderating role of easiness is predicted, proposing the following hypotheses: H12a: Easiness positively and significantly moderates the impact of AI-importance on customer trust. H12b: Easiness positively and significantly moderates the impact of AI-importance on AI-benefits.1 Methods Questionnaire development and pre-testing This study is an exploratory and empirical one that adopted a quantitative approach by using a survey questionnaire. Before commencing data gathering, a pre-test was carried out with seven expert academics and 16 frequent hotel customers. Based on their responses, a few minor changes were addressed to enhance the questionnaire content validity. Respondents were asked to participate in the survey only if they had patronized 3-, 4-, or 5-star hotels in the previous 6 months. Since the travel and vacation industry’s activities almost ceased due to the pandemic, it was impossible to approach actual hotel customers to survey respondents. Rather, the preliminary question aimed at guaranteeing that respondents had experienced recent hotel accommodation and relevant hospitality services, to better assess the AI-importance in hotels, including how much the respondents would trust AI-adopting hotels and what their behavioral intentions would be. The questionnaire includes three main sections. In the first one, AI importance as a multi-dimensional construct was measured through three main dimensions; namely, the biometrics (eight statements adopted from Murphy and Rottet, 2009), robotics (eight statements adopted from Ivanov et al., 2017), and self-service technologies (nine statements adopted from Ivanov et al., 2017). AI importance was operationalized as higher-order factors. These dimensions were gauged by using a 5-point Likert scale, where 1 = not important at all and 5 = very important. Another section of the questionnaire was assessing; AI benefits in hotels (four statements adopted from Lu, 2019), trust (four statements adopted from Martínez and Del Bosque, 2013) and their behaviroal intentions towards hotels adopting AI (four statements adopted from Kim, et al., 2009). A third area of questions focused on evaluating the personal-service (four statements adopted from Beatson, 2010), reputation (four statements adopted from Jalilvand et al., 2017), easiness (three statements adopted from Venkatesh, 2000), and identification (four statements adopted from Martínez and Del Bosque, 2013). By using a 5-point Likert scale, the survey questions were formulated, where 1 = strongly disagree and 5 = strongly agree. The questionnaire ended with some demographic questions. Population and sample Respondents were preliminarily asked to participate in the survey only if they had patronized 3-, 4-, or 5-star hotels in the previous 6 months. This opening question aimed at guaranteeing that respondents had experienced recent hotel accommodation and relevant hospitality services, to better assess the variables of the study. A sample frame for the population was unknown, therefore a non-probability sampling approach was selected, whereby a two-stage sampling process has been employed: firstly, the convenience sampling and then the snowball techniques. Convenience sampling was applied to reach ‘convenient’ sources of respondents (San Martín and Herrero, 2012) from several regions around the world. When the population is large, convenient sampling is successful (Etikan, et al., 2016), as in this study. In addition, it provides strong data when participation is high (Coviello and Jones, 2004). It is also the frequently selected sampling technique in hospitality studies due to the impossibility of accessing all available electronic sources of hotel customers around the world. As mentioned before, and due to the surge of COVID-19, it was not possible to approach actual hotel guests. Due to this dilemma, the authors had to choose between a statistically valid sample of the most accessible part of the target population, which will be absolutely limited, and a statistically less-valid and less-representable sample of broader coverage. That is why we resorted to snowball sampling, which is used when characteristics to be possessed by samples are rare and/or difficult to find, and where a population is hard to locate and tough to choose subjects to assemble them as samples for research. Snowball sampling may also help discover characteristics about usually inaccessible populations. And to avoid the common bias of snowball sampling, respondent-driven sampling was, in part, implemented. The authors could not be able to determine the tally of respondents recruited by each study subject initially contacted. Still, initial study subjects were asked to select from among their acquaintances, their own peers. This approach would mostly assure minimum level of homophily and homogeneity on attributes in the population. The questionnaire was designed using Google Forms. Respondents were contacted through a number of hotel chains, that distributed the e-form questionnaire through their websites and customer databases. Second, a snowball sampling technique was utilized to collect data from respondents through an online survey. Because of its relevant restrictions, the online survey was shared with conceivable participants within professional and social networks (e.g., WhatsApp, and Linked-In), which they then shared with their networks (Salem et al., 2021, 2022, 2023; Ghazi, 2018; Ghazi and Ammar, 2018; Ghazi, et al., 2023). The typical sample size for Structural Equation Modeling (SEM) is approximately 200 cases. However, more than 400 cases are adequate for examining a theoretical inclusive model (Kline, 2011). The responses obtained were from 451 previous hotel clients, of which 416 were valid. From the total usable sample (416), 60.8% of the respondents were men and 39.2% were women. Respondents represented diverse age groups; 36.1% were from 28 to 37, 29.3% were from 38 to 47, 18.8% were from 18 to 27, 12.3% were from 48 to 57, and 3.6% were from 58 years old and above. Most of the respondents were master’s degree holders (54.1%), followed by those with university education (22.8%), then Ph.D. holders (19.7%), and finally those with secondary education (3.4%). Respondents were from different geographic areas; about 31% were from the MENA region, 27.6% were from Europe and North America, 24% were from Asia, and 17.3% were from Africa. Data analysis and results Data analysis Partial Least Squares-Structural Equation Modeling (PLS-SEM) was used by exploiting WarpPLS7 (Kock, 2020). Prior to testing the model, normality, multicollinearity and common method bias (CMB) tests were carried out by using the average full collinearity variance inflation factor (AFVIF) that affirmed all variables had values of less than 3.00, which is ideal (Kock, 2020). The PLS-SEM assessment comprised of a two-step process; the measurement model through confirmatory factor analysis (CFA), followed by testing the hypothesized structural relationships among the key constructs included in the conceptual model (Hair et al., 2017). Measurement model In this research, the guidelines of Hair et al. (2017) have been followed, recommending the selection of reflective constructs. We first assessed the convergent validity which includes the composite reliability (CR), Cronbach’s Alpha, average variance extracted (AVE), and the variance inflation factor (VIF). Table 1 displays the value of the CR and Cronbach Alpha values, which exceeded the appropriate level of 0.7. AVE values were higher than the value of 0.5 (range, 0.584–0.811), which intimates an adequate convergent validity (Fornell and Larcker, 1981). Moreover, all variables had VIF values of below 3.3 (range 1.142–3.126), which is perfect, in addition to the lack of both multicollinearity and common method bias (Kock, 2020). CFA is shown in Appendix A.Table 1. Convergent validity. Variable Composite reliability Cronbach’s alpha AVE VIF Biometric authentication 0.926 0.908 0.613 2.016 Robotic process automation 0.918 0.898 0.584 2.404 Self-serve technologies 0.960 0.955 0.668 1.979 Customer trust 0.917 0.877 0.734 1.586 Perceived benefits 0.927 0.895 0.761 1.931 Preference for personal service 0.945 0.922 0.811 1.142 Perceived corporate reputation 0.924 0.890 0.751 2.270 Perceived ease of use 0.921 0.884 0.744 1.518 Customer-company-identification 0.908 0.864 0.713 3.126 Behavioral intention 0.931 0.901 0.771 2.281 Note. AVE = average variance extracted; VIF = variance inflation factor. Secondly, as shown in Table 2, the square root of AVE for each construct with correlations among the latent variables was examined, revealing an acceptable discriminant validity (Fornell and Larcker, 1981). Additionally, Henseler et al. (2015) introduced an alternative and novel method to verify discriminant validity that concentrates on the multitrait-multimethod matrix to assess discriminant validity; the heterotrait-monotrait (HTMT) ratio of correlations (see Table 2). To meet the HTMT criterion, each value must be equal to or below 0.85. All study variables have values less than 0.85, showing that discriminant validity was satisfactory.Table 2. Discriminant validity. Constructs 1 2 3 4 5 6 7 8 9 10 Squared roots of average variance extracted (AVE) 1. Biometric authentication 0.783 — — — — — — — — — 2. Robotic process automation 0.637 0.764 — — — — — — — — 3. Self-serve technologies 0.566 0.652 0.817 — — — — — — — 4. Customer trust 0.300 0.202 0.174 0.857 — — — — — — 5. Perceived benefits 0.376 0.439 0.412 0.429 0.872 — — — — — 6. Preference for personal service 0.209 0.107 0.031 0.224 0.071 0.900 — — — — 7. Perceived corporate reputation 0.390 0.359 0.346 0.507 0.586 0.033 0.867 — — — 8. Perceived ease of use 0.300 0.357 0.309 0.366 0.513 0.119 0.444 0.862 — — 9. Customer-company-identification 0.413 0.403 0.284 0.402 0.471 0.122 0.569 0.421 0.844 — 10. Behavioral intention 0.412 0.510 0.440 0.452 0.537 −0.007 0.637 0.448 0.557 0.878 Heterotrait-monotrait HTMT 1. Biometric authentication — — — — — — — — — — 2. Robotic process automation 0.708 — — — — — — — — — 3. Self-serve technologies 0.609 0.705 — — — — — — — — 4. Customer trust 0.335 0.228 0.192 — — — — — — — 5. Perceived benefits 0.419 0.490 0.446 0.484 — — — — — — 6. Preference for personal service 0.228 0.157 0.084 0.249 0.079 — — — — — 7. Perceived corporate reputation 0.433 0.403 0.376 0.575 0.656 0.059 — — — — 8. Perceived ease of use 0.330 0.401 0.337 0.415 0.574 0.133 0.500 — — — 9. Customer-company-identification 0.471 0.403 0.314 0.458 0.537 0.139 0.649 0.482 — — 10. Behavioral intention 0.457 0.567 0.476 0.511 0.598 0.048 0.712 0.501 0.631 — HTMT ratios (good if <0.90, best if <0.85). Structural model and hypotheses testing The structural model was estimated using goodness-of-fit indices, standardized path coefficients (β-value), significance level (t statistic), effect sizes (f 2 ), and R2 estimates (Hair et al., 2017). To estimate the model fit, standardized root mean square residual (SRMR) was employed (Henseler et al., 2015). An SRMR value of 0 would indicate an ideal fit, and generally, an SRMR value of ≤0.1 is rated as satisfactory for PLS models (Kock, 2020). In this study, an SRMR of 0.065 resulted in a satisfactory model fit. As shown in Table 3, all hypothesized relationships were supported, except for H9a (AI importance - > Personal-service - > trust: β-value = -0.068, and t value = -1.395), H11a (AI importance - > CCI- > trust: β-value = 0.052 and t value = 0.822), H11b (AI importance - > CCI - > AI benefits: β-value = 0.040 and t value = 0.809), and H12a (AI importance - > Easiness > trust: β-value = 0.039 and t value = 0.806). Furthermore, R2 values below 0.25 show a weak accuracy, those lower than 0.50 indicate a moderate accuracy, and values below 0.75 imply a solid predictive accuracy. AI importance during the COVID-19 pandemic explained 41% of the variance in behavioral intentions (R2 = 0.41). The R2 value of 41% is greater than the 0.25, implying a noteworthy model. Moreover, the effect size (f2) exhibits whether the effects stipulated by path coefficients are small, medium, or large. Kock (2020) clarified that the values commonly implied are 0.02, 0.15, and 0.35, sequentially. As shown in Table 3, most relationships had a medium effect.Table 3. Hypotheses-testing summary. No. Hypotheses Beta p-Value t Supported? f 2 H1 AI-importance has a positive influence on customers’ BIs 0.421 <0.001 9.1 Yes 0.215 H2 AI-importance has a positive influence on customer trust 0.276 <0.001 5.8 Yes 0.075 H3 AI-importance has a positive influence on AI-benefits 0.475 <0.001 10.3 Yes 0.226 H4 AI-benefits have a positive on customer trust 0.352 <0.001 8.5 Yes 0.172 H5 Trust has a positive influence on BIs 0.369 <0.001 7.9 Yes 0.161 H6 AI-benefits have a positive influence on BIs 0.409 <0.001 8.1 Yes 0.292 H7a AI-benefits positively mediate AI-importance and trust 0.188 <0.001 1.7 * Yes 0.052 H7b AI-benefits positively mediate AI-importance and BIs 0.179 <0.001 7.5 Yes 0.095 H8a Customer trust positively mediates AI-importance and BI 0.102 0.002 9.2 Yes 0.054 H8b Customer trust positively mediates AI-benefits and BIs 0.141 <0.001 8.8 Yes 0.077 H9a Personal-service positively moderates AI-importance and customer trust −0.068 0.082 −1.395 No 0.023 H9b Personal-service positively moderates AI-importance and AI-benefits −0.217 <0.001 −4.6 Yes 0.023 H10a Reputation positively moderates AI-importance and customer trust 0.110* 0.021 2.285 Yes 0.025 H10b Reputation positively moderates AI-importance and AI-benefits 0.252 <0.001 5.317 Yes 0.111 H11a Identification positively moderates AI-importance and customer trust 0.052 0.223 0.822 No 0.019 H11b Identification positively moderates AI importance and AI-benefits 0.040 0.206 0.809 No 0.009 H12a Easiness positively moderates AI-importance and trust 0.039 0.210 0.806 No 0.019 H12b Easiness positively moderates AI-importance and AI-benefits 0.190 <0.001 3.975 Yes 0.078 t value for two-tailed tests: 1.960 (p < 0.001), t value for one-tailed tests: 1.645 (p < 0.01). Table 4 exhibits the mediation analysis results. The variance accounted for (VAF) confirms the indirect effect size and the total effect are connected. Values above 80% exemplify full mediation, those within 20% and 80% imply partial mediation, and values below 20% reveal no mediation force (Zhao et al., 2010). Table 4 exhibits that all mediating effects are partially supported.Table 4. Summary of mediation results. Paths Significance Confidence intervals VAF Mediation Direct effect Indirect effect 95% LL 95% UL % Outcome AI-importance on customer trust via AI benefits 0.083 0.188 0.094 0.282 0.693 69.3 Partially mediated AI-importance on BIs via AI-benefits 0.352 0.179 0.112 0.246 0.337 33.7 Partially mediated AI-importance on BIs via trust 0.421 0.102 0.035 0.168 0.195 20 Partially mediated AI-benefits and BIs via trust 0.409 0.141 0.075 0.208 0.256 25.6 Partially mediated p < 0.01; *p < 0.05; LL, Lower level; UL, Upper level; VAF, Variance Accounted For. The moderation effects by using the two-stage approach were also measured. The formula proposed by Kock (2020) was used to evaluate the variations in path coefficients between AI importance and both AI benefits and trust models. The t-statistics were determined and are exhibited in Table 3. The projected standardized path coefficients for the effect of the moderator on AI benefits (β = −0.217; p < 0.001) were significant (see Table 3). Therefore, personal-service dampens the positive association between AI-importance and AI-benefits (see Figure 2). Likewise, as shown in Table 3 and Figure 2, reputation strengthens the positive relationship between AI importance, trust and AI benefits, while easiness increases the positive relationship between AI importance and AI benefits. As explained earlier, H9a, H11a, H11b, and H12a were not supported, as no moderation effects were experienced.Figure 2. Moderation analysis. Conclusions and implications Conclusions As a result of the pandemic, potential hotel customers are more concerned with utilitarian, concrete benefits and outcomes than with emotional perspectives such as trust. Likewise, triggered by the pandemic, customers are also more compelled by tangible influencing variables, such as preference for personal service, and ease of use, than social influences, such as corporate reputation and identification with the company. It was confirmed that the higher the potential guests’ perceived AI-importance, the higher their behavioral intentions towards AI-applying hotels will be during and post-COVID-19. This conforms to previous studies (e.g., Shin and Kang, 2020; Zhong et al., 2020), confirming that customers’ behavioral intentions are positively influenced by their awareness of the value of novel technologies. This also corresponded to UTAUT2’s assumption that performance expectancy has a significant effect on attitudes and intentions to use various innovative technologies (Hao, 2021; Jung and Cha, 2022; Morosan and DeFranco, 2016; Wu et al., 2021). This urging need for tangible value was apparent in that the AI-benefits are more influential across various investigated correlations than trust, either as a direct predictor, a dependent variable, or as an efficacious mediator. This implies that the pursuit of solid gains by potential guests surpasses and precedes their need for trust. The results indicate that the more potential customers perceive AI as being important, the more this would build their trust in AI-adopting hotels, which conforms with Tussyadiah et al. (2020) and Ruan et al. (2020). Furthermore, Wu et al. (2021) confirmed that AI-importance has a more significant effect on AI-benefits. This result proves that the current study’s variation of UTAUT2 is significant, whereby performance expectancy significantly and positively impacts effort expectancy. Besides, it was proved that AI-benefits have a substantial effect on trust. This conforms to prior studies, which revealed that AI-benefits positively influenced trust, in addition to mediating the impact of AI-importance on trust (Ameen et al., 2021; Lee and Lee, 2019; Ruan et al., 2020). In addition, it was supported that trust has a significant effect on BI, a result that was also supported through previous studies (Tussyadiah et al., 2020; Ruan et al., 2020; Wu et al., 2021). Nevertheless, results once again assert the worth of AI-benefits for customers, where it is found to have a significant effect on behavioral intentions. An added argument that verifies the fundamental role of AI benefits is that results show that AI-benefits play a partial mediating role through the impact of AI importance on trust. This implies that even trust, as a common key variable, is also preceded and foreshadowed by AI-benefits. This agrees with studies proving that various forms of AI-benefits mediate the impact of AI-importance on trust (Ameen et al., 2021; Pinxteren et al., 2019; Ruan et al., 2020). The findings also revealed that the correlation between AI-importance and behavioral intentions is partially mediated by AI-benefits, to a greater extent than trust. These results validate the investigation and the inclusion of the customers’ utilitarian versus emotional perceptions into UTAUT2. Furthermore, results confirmed the necessity of gaining trust, which significantly mediates the relationship between AI-benefits and behavioral intentions. A rather insubstantial mediating magnitude assures the previously induced inference that customers now seek more tangible outcomes. Customer trust was found to play a mediating role between AI-importance and AI-benefits (Ameen et al., 2021; Ruan et al., 2020) and between AI-benefits and behavioral intentions (Wang et al., 2015; Lee and Lee, 2019). This strongly suggests that AI-benefits play a critical role in the decision-making process of supporting and dealing with AI-adopting hotels. These findings are consistent with previous research (Davis, 1989; De Kervenoael et al., 2020; Lin et al., 2020) indicating that perceived AI-benefits, value, and usefulness have a direct impact on customers’ intentions and positive emotions to support and use new technologies. AI-benefits have also been shown to play a mediating role in the relationship between AI-importance and behavioral intentions (Pinxteren et al., 2019; Ruan et al., 2020). Results additionally indicated that personal service is not moderating the relationship between AI-importance and trust. A logical inference is that personal-service, as a moderator, dampened the relationship between AI-importance and AI-benefits. This sheds light on the fact that, in spite of the pandemic and its compelling impacts on the hospitality industry, and despite the tendency towards recent technological applications and solutions, the human touch is still much appreciated in the hospitality industry, and it will deter the full automation of service processes. This coincides with previous empirical studies that proved the negative association between personal-service and AI-benefits of technology (Shin and Kang, 2020; Ameen et al., 2021). The fact that tangible values are preferred to the indirectly rewarding, emotional aspects is also perceptible in other moderating relationships. The rather concrete variable “reputation,” which typically necessitates the formulation of a solid foundation (Walsh et al., 2009), moderates the relationship between AI-importance and both AI-benefits and trust. Furthermore, the effect of reputation, similar to UTAUT2’s “Social Influence,” is more significant when it modifies the relationship between AI-importance and AI-benefits rather than trust. This implies that the more reputable the hotel is, the more potential customers would perceive AI practices as being more beneficial and would subsequently trust the AI-adopting hotel. These findings are consistent with previous research indicating the importance of reputation in forming trust (Jalilvand et al., 2017; Chang, 2013) and influencing AI-benefits (Perez-Aranda et al., 2019; Chang, 2013). The results also indicated that identification, which represents UTAUT2’s “social influence”, which is a rather sentimental and intangible construct, does not moderate the relationships between AI-importance and trust nor between AI-importance and AI-benefits. This finding contradicts previous research that suggested the AI-importance of identification in achieving AI benefits (So et al., 2013; Rather, 2018). This disagreement can be attributed to COVID-19 circumstances, which engender different perceptions and varied responses to variables than those commonly established, and which push customers towards the pursuit of more tangible outcomes. Moreover, the results confirmed that ease, representing UTAUT2’s “facilitating conditions,” does not moderate its relationship with trust. This contradicts existing literature that supports the impact of ease on trust building (Agag and El-Masry, 2016; Chen et al., 2022; Hao, 2021; Pai et al., 2018). This might also be due to the same rationale: COVID-19 has directed potential customers’ interests towards more functionality and pursuing material outcomes. Meanwhile, the results confirmed previous research that ease, as a utilitarian perception (Davis, 1989), moderates the impact of AI-importance on its benefits (De Kervenoael et al., 2020). Theoretical implications This study presents a novel perception of how to integrate trendy operational solutions to deal with and recover from a crisis. In addition, it incorporates customer-decision-related utilitarian and emotional variables in the proposed model, which provides a workable extension to UTAUT2. AI solutions have been integrated with key moderating and mediating variables, with the ultimate goal of configuring how to build positive behavioral intentions towards AI-adopting hotels during COVID-19, and in similar crises. This study complements the empirical research gap concerning utilizing AI applications for amid and post-pandemic management (Ivanov et al., 2020; Seyitoğlu and Ivanov 2020; Shin and Kang, 2020), utilizing UTAUT2 to frame and streamline the study findings. Additionally, the current study examined AI-importance, AI-benefits, trust, identification, reputation, personal-service, and easiness variables, to provide a holistic, generalizable model that fits in most operational settings to positively steer customers’ behavioral intentions towards the hotel’s best interests, to make up for relevant shortages spotted in previous studies (Ruan et al., 2020; Shin and Kang, 2020; Tussyadiah, 2020; Zhong et al., 2020). Another apparent, statistically validated implication is the study’s applicable extension to UTAUT2’s to a broader spectrum through exchanging the traditional UTAUT2’s variables. This aimed at specifically incorporating a vast array of hypothesized correlations that would most fit into the operational status quo of hotels during and post COVID-19 and similar crises. Furthermore, other research-related outcomes are augmenting UTAUT2’s framework by additional, parallel variables to broaden the applicability of UTAUT2, and ultimately inspiring researchers to utilize the UTAUT2 as a launching point towards more exploratory studies according to concurrent needs of the hospitality industry. Managerial implications The current study posits several practical implications for involved stakeholders, particularly hotel managers. The major implication is expressed by Gaur et al. (2021), who urged hospitality practitioners to foster available knowledge to recover the COVID-19 crisis via resorting to digitalization solutions, with AI on top of technological innovations, not only for the sake of recovery, but also for thriving business and re-developing guests’ interest. Particularly, the extended UTAUT2 model examined is provided for effective managerial implications to institute the proper and purposeful adoption of AI technologies, and to survive and recover from the pandemic, by developing and sustaining customers’ interest and trust during and post COVID-19, and in similar crises. First and foremost, managers should exploit and make best use their customers’ pursuit of material, tangible gains; a need that is driven by fear from contagion and by the desire to obtain required services and products seamlessly. Thus, rather than just jacking a trend, hoteliers should steer their implementation of AI towards specific operational objectives that fulfil customers’ concurrent, COVID-19-driven expectations and needs, comprising cognitive and social benefits, value, convenience, practicality, quality, reliability, user-friendliness, security, and privacy. All those notions need to be supported by hotel managers to obtain the full potential of AI solutions. Second, customer trust is an important determinant among AI-related variables, and the eventual behavioral intentions. Hoteliers should develop and maintain their customers’ health-related and operational trust through providing AI applications that are easy to use, dependable, effective, time and cost-efficient, and, above all, guarantee the proper social distancing that does not deter smooth operations. Thus, trust should be provoked by linking it to material benefits to best appeal for customers, rather than just being an emotional prospect that might not matter to customers in times of crises. Hence, hotels would associate and gain both cognitive and emotional support from customers, eventually leading to positive behavioral intentions. Third, it is also imperative that managers, besides maximizing benefits, and supporting trustworthiness, should take due measures to stabilize and enhance their corporate reputation. This is attainable throughout monitoring and measuring customers’ reaction to and evaluation of the hotel’s goods, services, communication activities, interactions with the company and/or its representatives (Walsh, et al., 2009), and enhancing the corporate transparency, assuming social responsibility, pursuing business ethics, and preventing unfair competition (Almeida and Coelho, 2019). Furthermore, the corporate reputation should be portrayed as providing AI-assisted operations for the best interests of customers, their crisis-related concerns, and their pursuit for more facilitated, yet personalized, services. Fourth, It is not only compelling to adopt and implement AI; but it is also crucial that managers devise workable, feasible measures to periodically track and assess customers’ behavioral intentions as a rational, eventual outcome. Fifth, it was advised by several respondents that AI can be applied in functions that already involve a tangible product, like food and beverage outlets and housekeeping tasks. However, where functions are purely service centered, like reservations, check-in and check-out, it is preferred for them to be administered through human contact, where AI should only be integrated to enhance the service experience, not to substitute the human-based hospitable service encounters. That is, AI should be advertised as a means of facilitating and enriching customers’ service experience, thus enabling the “human and personal” hospitality element to be enhanced towards more personalized service, rather than advertising AI as a tool for decreasing personal contact, that would otherwise resent customers. Since perceived “Customer-Company Identification” is associated with the company’s trustworthiness, thus, customers’ inclination towards such an identification, as an emotional construct, will not be substantial during crises. Consequently, identification should be elicited and incited through including AI-assisted service capacities. AI usage should not then be solely a means of enhancing services and product delivery, rather, AI should be directed as a utility towards soliciting customers’ needs and wants, and incorporating them into designing and creating personalized packages, services, and communications. Accordingly, customers will not perceive identification as an ineffectual element in forming their behaviroal intentions towards AL-adopting hotels. Rather, identification will be appreciated as being associated with quantifiable, pertinent outcome, particularly during and after crises. Moreover, hotel managers have to carefully monitor data privacy and security issues, since some respondents welcomed using AI applications, but were hesitant to use biometric identification to pay their bills, while others were totally against biometric identification. Thus, hotels should provide their guests with multiple options for these concerns. In addition to the indispensable security concerns, to influence behavioral intentions further positively, it is essential for hospitality managers and practitioners to boost their AI applications’ perceived usefulness, ease of use, interactivity, responsiveness, and innovativeness. Finally, those AI-inherent features must not come on the account of anthropomorphism, so as not to miss the hospitality-inbuilt human touch and the personalized flair. Ultimately then, customers will build up and effectuate their proclivity to use AI applications. Since COVID-19 is almost done, it is rather essential to comprehend that those implications are equally essential and applicable to stimulate customers’ behavioral intentions during and amid other types of crises, and to minimize or compromise the effect of other types of turmoil. AI applications’ versatility directly supports the aspects that are usually most adversely affected, improving service quality, increasing operational capability and efficiency, lowering costs, and consequently creating a competitive advantage for AI-adopting hotels during crises. AI-contactless solutions help recover customer trust and encourage demand through keeping social and physical distance and lowering costs, which are the major customers’ concerns during health and economic risks. Limitations and future directions The current study was directed towards customers of hotels. Further studies are needed to examine the proposed model in other tourism and hospitality sectors and in specific geographical areas, discerning the needs of various types of customers. Furthermore, this study focused only on the positive aspects and AI-benefits during the pandemic. Future research should investigate the negative aspects and drawbacks of AI, and whether they would hinder customers from accepting AI in response to the threats posed by COVID-19. Further research should investigate how each type of AI fits individually into the proposed model, rather than biometric authentication, robotics, and self-serve technologies. Moreover, more concern have to be directed to the utilization of Metaverse and ChatGPT, and their role towards minimizing crisis-related operational drawbacks. Rather than just surveying customers, a natural extension of the current study is to scrutinize AI and its applications during crises from an organizational and strategic perspectives by surveying hotel chains’ trend-setters and regional directors, tourism and hospitality establishment owners, managers, supervisors, and employees. To gain more insights into how customers process their decisions, a qualitative, rather than quantitative, methodology should be utilized to acquire deeper acumen on how customers proceed through each phase of the guest cycle, starting from planning their trip, and extending through their post-service preferences. Finally, researchers and theorists should be encouraged to launch novel, augmented versions from UTAUT2 to specifically suit and investigate recent, urging technology-related operational needs and research gaps, and extend UTAUT2’s viability to more spectrums. Author Biographies Karam Ghazi holds a PhD in hotel studies from Alexandria University, Egypt. He worked as Associate Professor at the Higher Institute for Tourism and Hotels in Alexandria (EGOTH), Egypt. His research and publications encompass both national and international domains. His research interests are crisis management, technology, safety and security, marketing, human resources and other issues related to the tourism and hospitality business. He has a relevant number of research publications in distinguished journals such as; Tourism Management, Tourism Management Perspectives, and Tourism and Hospitality Research. Hanan Kattara is a professor of hotel management at Alexandria University. Her research and publications encompass both national and international domains. Her prime research area is in human resource management applied to the hospitality and tourism sector. She also has contributions in general hotel management, marketing, hotel operations and new trends in the industry. She developed excellent professional skills with a comprehensive blend of hands-on industrial and academic hospitality experience. Professor Kattara has a relevant number of research publications in distinguished journals. She is actively involved in collaborative research networks and projects with outstanding international education and research institutions, as well as consultancy work with governmental authorities and hospitality enterprises. Islam Elbayoumi Salem is an associate professor at the Faculty of Tourism and Hotels, Alexandria University, Egypt and at the Business Administration Department, Salalah College of Applied Sciences, University of Technology and Applied Sciences, Oman. His research interests are hospitality leadership, hospitality marketing, hospitality technology, hotels’ outsourcing, and structural equation modelling (SEM) CB-SEM and PLS-SEM analysis in tourism and fuzzy-set configuration approach. Mohammad Shaaban is an Associate Professor of Hospitality Management at University of Prince Mugrin, Saudi Arabia, on sabbatical leave from Alexandria University, Egypt. His research activities and publications revolve on human resources management, organizational behavior, and marketing, with the application on hospitality industry. His overseas academic and research experiences extended to Greece, Dubai, Spain, Kazakhstan, and Saudi Arabia. He is a Certified Hospitality Department Trainer from the American Hotel and Lodging Association (AHLA), in addition to the Entrepreneurship and Small Business Certification, accredited from Certiport-A Pearson Vue Business. His responsibilities and activities include mentoring, training, teaching, and designing various hospitality courses and modules for educational and governmental institutions. ORCID iDs Karam Ghazi https://orcid.org/0000-0002-8836-8939 Islam Salem https://orcid.org/0000-0002-9483-3550 Appendix Appendix A: Confirmatory factor analysis PLS approach. Construct/items Mean SE Loadings p-value Confidence 2.5% Intervals 97.5% AI-biometric authentication (BA) 3.75 0.970 — — — — BA.1 Online booking of a hotel room 3.31 1.56 0.641 <0.001 0.553 0.729 BA.2 Check-in and check-out processes 3.84 1.19 0.829 <0.001 0.743 0.915 BA.3 Entering the guestroom 3.85 1.17 0.779 <0.001 0.693 0.866 BA.4 Operating in-room services and functions 3.80 1.25 0.833 <0.001 0.747 0.919 BA.5 Locking the guestroom upon the guest’s exit 3.96 1.23 0.821 <0.001 0.735 0.907 BA.6 Processing inquiries, ordering and booking hotel’s products and services 3.63 1.20 0.804 <0.001 0.717 0.890 BA.7 Identifying guests when accessing various outlets 3.66 1.20 0.825 <0.001 0.739 0.911 BA.8 Processing and confirming various payments 3.89 1.15 0.708 <0.001 0.620 0.795 AI-robotic process automation (RPA) 3.41 0.914 — — — — RPA.1 Front Desk robots performing check-in and check-out functions 3.17 1.03 0.686 <0.001 0.599 0.774 RPA.2 Porter robots for luggage transferring 3.36 1.23 0.770 <0.001 0.683 0.857 RPA.3 concierge robots 3.30 1.29 0.768 <0.001 0.681 0.855 RPA.4 Vacuum cleaning and disinfectant robots for housekeeping purposes 3.71 1.19 0.771 <0.001 0.684 0.857 RPA.5 In-room assistant robots 3.65 1.16 0.741 <0.001 0.654 0.828 RPA.6 Delivery robots 3.45 1.17 0.805 <0.001 0.719 0.891 RPA.7 Robot restaurant servers, bartenders, and baristas 3.02 1.28 0.770 <0.001 0.683 0.857 RPA.8 Robot-assisted bill-payment 3.56 1.19 0.794 <0.001 0.707 0.880 AI-self-serve technologies (SST) 3.57 0.971 — — — — SST.1 Self-service check in and check-out lobby kiosks 3.58 1.1956 0.799 <0.001 0.713 0.885 SST.2 Self-service mobile check in/out 3.71 1.1924 0.827 <0.001 0.741 0.913 SST.3 Self-service kiosks for information (concierges services) 3.51 1.1760 0.827 <0.001 0.741 0.913 SST.4 Ordering hotel’s products and services using the hotel-specific mobile application 3.762 1.1443 0.823 <0.001 0.736 0.909 SST.5 Restaurant table-side ordering 3.630 1.1481 0.863 <0.001 0.778 0.949 SST.6 Restaurant table-side entertainment 3.233 1.2108 0.794 <0.001 0.708 0.881 SST.7 Restaurant table-side payment 3.740 1.1822 0.854 <0.001 0.768 0.940 SST.8 Conveyor/Roller-coaster restaurants 3.317 1.2303 0.821 <0.001 0.735 0.907 SST.9 Using glass cubbies 3.385 1.2128 0.833 <0.001 0.747 0.919 SST.10 Using chatbots 3.466 1.2530 0.802 <0.001 0.716 0.889 SST.11 Offering virtual Voice assistants in room standalone devices 3.577 1.1858 0.770 <0.001 0.683 0.857 SST.12 Using a mobile Native Languages translations 3.909 1.1349 0.792 <0.001 0.706 0.879 Preference for personal service (PPS) 4.00 0.916 — — — — PPS.1 Face-to-face contact in providing services makes the process enjoyable 4.147 0.9769 0.908 <0.001 0.823 0.993 PPS.2 I like interacting with the person who provides the service 4.024 1.0548 0.943 <0.001 0.858 1.027 PPS.3 I like making conversation with the person who is providing the service 3.978 0.9679 0.853 <0.001 0.767 0.939 PPS.4 I have a preference for dealing with contact staff in service settings 3.858 1.0717 0.896 <0.001 0.811 0.981 Perceived benefits (PB) 3.84 0.808 — — — — PB.1 The hotel artificial intelligence reduces my searching time to access the hotel products that I need 3.868 0.9180 0.880 <0.001 0.795 0.966 PB.2 The hotel artificial intelligence can provide me with the convenience of instantly accessing the hotel products that I need 3.875 0.8719 0.884 <0.001 0.798 0.969 PB.3 I think that using artificial intelligence in a hotel can offer me a wider range of hotel products 3.767 0.9699 0.830 <0.001 0.744 0.916 PB.4 Overall, I feel that the hotel artificial intelligence is beneficial to access the hotel products 3.837 0.9530 0.893 <0.001 0.808 0.979 Perceived ease of use (PEU) 3.72 0.817 — — — — PEU.1 Learning to deal with artificial intelligence in hotels would be easy for me 3.861 0.9390 0.838 <0.001 0.752 0.924 PEU.2 My interactions with artificial intelligence in hotels would be clear and understandable 3.764 0.9196 0.909 <0.001 0.824 0.994 PEU.3 My interactions with artificial intelligence in hotels would not require a lot of mental effort 3.546 1.0026 0.816 <0.001 0.729 0.902 PEU.4 Overall, I believe artificial intelligence is easy to use 3.714 0.9352 0.883 <0.001 0.798 0.969 Perceived customer-company-identification (CCI) 3.15 0.889 — — — — CCI.1 When someone criticizes hotels that provide AI, it feels like a personal insult 2.887 1.0818 0.827 <0.001 0.741 0.913 CCI.2 I am very interested in what others think about hotels that provide AI. 3.450 0.9755 0.768 <0.001 0.681 0.854 CCI.3 When someone compliments hotels that provide AI, then it feels like a personal compliment 3.207 1.0439 0.890 <0.001 0.805 0.976 CCI.4 When I talk about hotels that provide AI, I usually say “we” rather than “they” 3.059 1.1090 0.887 <0.001 0.801 0.972 Perceived corporate reputation (PCR) 3.72 0.822 — — — — PCR.1 Highly regarded 3.724 0.9304 0.853 <0.001 0.767 0.938 PCR.2 Successful 3.788 0.9511 0.884 <0.001 0.799 0.969 PCR3 Well-established 3.805 0.9483 0.873 <0.001 0.787 0.958 PCR.4 Stable 3.560 0.9675 0.857 <0.001 0.771 0.943 Customer trust (CT) 3.67 0.839 — — — — CT.1 Hotels’ services will make me feel a sense of security 3.522 1.0503 0.824 <0.001 0.738 0.910 CT.2 Hotels will provide quality services 3.603 0.9713 0.921 <0.001 0.836 1.006 CT.3 Services of hotels will be a quality-assurance process 3.644 1.0028 0.899 <0.001 0.814 0.984 CT.4 Hotels will be interested in their customers 3.901 0.8984 0.776 <0.001 0.689 0.862 Behavioral intention (BI) 3.63 0.832 — — — — BI.1 I would recommend hotels operating with AI to other people 3.709 0.9075 0.870 <0.001 0.785 0.956 BI.2 I would tell other people positive things about hotels operating with AI. 3.793 0.8618 0.860 <0.001 0.774 0.945 BI.3 I consider hotels operating with AI as my first choice compared to other hotels 3.416 1.0494 0.886 <0.001 0.801 0.972 BI.4 I have a strong intention to repeat visits to hotels operating with AI. 3.582 0.9684 0.896 <0.001 0.811 0.981 The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ==== Refs References Agag G El-Masry A (2016) Why do customers trust online travel websites? Drivers and outcomes of customer trust toward online travel websites. Journal of Travel Research 56 (3 ): 347–369. Ajzen I (1991) The theory of planned behavior. Organizational Behavior and Human Decision Processes 50 (2 ): 179–211. Almeida M Coelho A (2019) The antecedents of corporate reputation and image and their impacts on employee commitment and performance: The moderating role of CSR. Corporate Reputation Review 22 (1 ): 10–25. Ameen N Tarhini A Reppel A , et al. (2021) Customer experiences in the age of artificial intelligence. Computers in Human Behavior 114 : 106548.32905175 Beatson A (2010) Self-service technology and the moderating effects of consumer characteristics. International Journal of Services Technology and Management 14 (2/3 ): 278–302. Bhattacharya CB Sen S (2003) Consumer–company identification: A framework for understanding consumers’ relationships with companies. Journal of Marketing 67 (2 ): 76–88. Cha SS (2020) Customers’ intention to use robot-serviced restaurants in Korea: relationship of coolness and MCI factors. International Journal of Contemporary Hospitality Management 32 (9 ): 2947–2968. Chang KC (2013) How reputation creates loyalty in the restaurant sector. International Journal of Contemporary Hospitality Management 25 (4 ): 536–557. Chen C Chen H Wang Y (2022) Cash, credit card, or mobile? Examining customer payment preferences at chain restaurants in Taiwan. Journal of Foodservice Business Research 25 (2 ): 148–167. Coviello NE Jones MV (2004) Methodological issues in international entrepreneurship research. Journal of Business Venturing 19 (4 ): 485–508. Dabholkar PA Bagozzi RP (2002) An attitudinal model of technology-based self-service: moderating effects of consumer traits and situational factors. Journal of the Academy of Marketing Science 30 (3 ): 184–201. Dardeer M Tag-Eldeen A Salem I (2017) The influence of physical work environment on hotel back-of-the-house employees’ satisfaction and productivity: A Case Study on Hilton Hotels. Journal of Tourism Research 5 (18 ): 45–64. Davis FD (1989) Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13 (3 ): 319–340. de Kervenoael R Hasan R Schwob A , et al. (2020) Leveraging human-robot interaction in hospitality services: Incorporating the role of perceived value, empathy, and information sharing into visitors’ intentions to use social robots. Tourism Management 78 : 104042. Elkhwesky Z El Manzani Y Elbayoumi Salem I (2022) Driving hospitality and tourism to foster sustainable innovation: A systematic review of COVID-19-related studies and practical implications in the digital era. Tourism and Hospitality Research. Etikan I Musa S Alkassim R (2016) Comparison of convenience sampling and purposive sampling. American Journal of Theoretical and Applied Statistics 5 (1 ): 1–4. Fishbein M Ajzen I (1977) Belief, attitude, intention, and behavior: An introduction to theory and research. Philosophy and Rhetoric 10 (2 ): 177–188. Fornell C Larcker DF (1981) Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research 18 : 39–50. Gaur L Afaq A Singh G , et al. (2021) Role of artificial intelligence and robotics to foster the touchless travel during a pandemic: a review and research agenda. International Journal of Contemporary Hospitality Management 33 (11 ): 4079–4098. Gursoy D Chi CG (2020) Effects of COVID-19 pandemic on hospitality industry: review of the current situations and a research agenda. Journal of Hospitality Marketing and Management 29 (5 ): 527–529. Gupta A Dogra N George B (2018) What determines tourist adoption of smartphone apps? An analysis based on the UTAUT-2 framework. Journal of Hospitality and Tourism Technology 9 (1 ): 50–64. Ghazi KM (2018) Hotel marketing deception practices and its effect on Guests’ image. Journal of Tourism Research 20, 88–108. Ghazi K Ammar S (2018) International tourists’ attitudes and satisfaction toward airport F&B, and duty-free concessions. Journal of Tourism Research 20, 41–109. Ghazi KM El-Said O Salem IE Smith M (2023) Does performance appraisal legitimacy predict employee sabotage and innovative behaviors? The mediating role of performance appraisal satisfaction. Tourism Management Perspectives 47, 101117. Hair Jr JF Matthews LM Matthews RL , et al. (2017) PLS-SEM or CB-SEM: updated guidelines on which method to use. International Journal of Multivariate Data Analysis 1 (2 ): 107–123. Hao F (2021) Acceptance of contactless technology in the hospitality industry: extending the unified theory of acceptance and use of technology 2. Asia Pacific Journal of Tourism Research 26 (12 ): 1386–1401. Hao F Xiao Q Chon K (2020) COVID-19 and China’s hotel industry: Impacts, a disaster management framework, and post-pandemic agenda. International Journal of Hospitality Management 90, 102636. Henseler J Ringle CM Sarstedt M (2015) A new criterion for assessing discriminant validity in variance-based structural equation modeling. Journal of the Academy of Marketing Science 43 (1 ): 115–135. Ivanov S Webster C Berezina K (2017) Adoption of robots and service automation by tourism and hospitality companies. Revista Turismo & Desenvolvimento 27 (28 ): 1501–1517. Ivanov S Webster C Stoilova E , et al. (2020). Biosecurity, automation technologies and economic resilience of travel, tourism and hospitality companies. 1–34.‏ Jalilvand M Nasrolahi Vosta L Kazemi Mahyari H , et al. (2017) Social responsibility influence on customer trust in hotels: mediating effects of reputation and word-of-mouth. Tourism Review 72 (1 ): 1–14. Jung S Shin YC Oh KS , et al. (2022) Perceived control and work-related stress mediate the effects of grit on depression among employees. Brain Sciences 13 (1 ): 9–16.36671991 Kim J Montes AA Han H (2021) The role of expected benefits towards smart hotels in shaping customer behavior: comparison by age and gender. Sustainability 13 (4 ): 1698. Kim T Kim WG Kim HB (2009) The effects of perceived justice on recovery satisfaction, trust, word-of-mouth, and revisit intention in upscale hotels. Tourism Management 30 (1 ): 51–62. Kline RB (2011) Convergence of structural equation modeling and multilevel modeling. Knani M Echchakoui S Ladhari R (2022) Artificial intelligence in tourism and hospitality: Bibliometric analysis and research agenda. International Journal of Hospitality Management 107 : 103317. Kock N (2020) WarpPLS User Manual: Version 7.0. Laredo, TX, USA: ScriptWarp Systems. Lee SA Lee J (2019) Enhancing customers’ brand loyalty via branded hotel apps. Journal of Quality Assurance in Hospitality & Tourism 20 (3 ): 339–361. Li M Yin D Qiu H , et al. (2022) Examining the effects of AI contactless services on customer psychological safety, perceived value, and hospitality service quality during the COVID-19 pandemic. Journal of Hospitality Marketing & Management 31 (1 ): 24–48. Lin H Chi OH Gursoy D (2020) Antecedents of customers’ acceptance of artificially intelligent robotic device use in hospitality services. Journal of Hospitality Marketing & Management 29 (5 ): 530–549. Lu T (2019) Opportunities of Artificial Intelligence in Hospitality Industry for Innovative Customer Services. Doctoral Dissertation, Ho Chi Minh City, Vietnam. Lukanova G Ilieva G (2019) Robots, artificial intelligence, and service automation in hotels. Robots, artificial intelligence, and service automation in travel, tourism and hospitality. Emerald Publishing Limited. Martínez P Rodríguez del Bosque I (2013) CSR and customer loyalty: The roles of trust, customer identification with the company and satisfaction. International Journal of Hospitality Management 35 : 89–99. McKnight DH Chervany NL (2001) What trust means in e-commerce customer relationships: An interdisciplinary conceptual typology. International Journal of Electronic Commerce 6 (2 ): 35–59. Morosan C DeFranco A (2016) It's about time: Revisiting UTAUT2 to examine consumers’ intentions to use NFC mobile payments in hotels. International Journal of Hospitality Management 53 : 17–29. Murphy H Rottet D (2009) An exploration of the key hotel processes implicated in biometric adoption. International Journal of Contemporary Hospitality Management 21 (2 ): 201–212. Pai CK Wang TW Chen SH , et al. (2018) Empirical study on Chinese tourists’ perceived trust and intention to use biometric technology. Asia Pacific Journal of Tourism Research 23 (9 ): 880–895. Park S Tussyadiah IP (2020) How guests develop trust in hosts: an investigation of trust formation in P2P accommodation. Journal of Travel Research 59 (8 ): 1402–1412. Perez-Aranda J Vallespín M Molinillo S (2019) Hotels’ online reputation management: benefits perceived by managers. International Journal of Contemporary Hospitality Management 31 (2 ): 615–632. Perić M Vitezić V (2021) Tourism getting back to life after COVID-19: Can artificial intelligence help? Societies 11 (4 ): 115–213. Pillai SG Haldorai K Seo WS , et al. (2021) COVID-19 and hospitality 5.0: Redefining hospitality operations. International Journal of Hospitality Management 94 : 102869.34785847 van Pinxteren MM Wetzels R Rüger J , et al. (2019) Trust in humanoid robots: Implications for services marketing. Journal of Services Marketing 33 (4 ): 507–518. Rather RA (2018) Investigating the impact of customer brand identification on hospitality brand loyalty: A social identity perspective. Journal of Hospitality Marketing & Management 27 (5 ): 487–513. Romero J Lado N (2021) Service robots and COVID-19: exploring perceptions of prevention efficacy at hotels in generation Z. International Journal of Contemporary Hospitality Management 33 (11 ): 4057–4078. Ruan WQ Zhang SN Liu CH , et al. (2020) A new path for building hotel brand equity: The impacts of technological competence and service innovation implementation through perceived value and trust. Journal of Hospitality Marketing & Management 29 (8 ): 911–933. Salem IEB Čavlek N (2016) Evaluation of hotel website contents: existence-importance analysis. Journal of Hospitality and Tourism Technology 7 (4 ): 366–389. Salem IE Elbaz AM Elkhwesky Z , et al. (2021) The COVID-19 pandemic: The mitigating role of government and hotel support of hotel employees in Egypt. Tourism Management 85 : 104305.36567706 Salem IE Aideed H Alkathiri NA Ghazi KM (2022) Retaining talented employees during COVID-19 pandemic: The leverage of hotel pandemic response strategies. Tourism and Hospitality Research, 14673584221141294. Salem IE Abbas H Mousa M , et al. (2023) Does dysfunctional customer behavior really impact the turnover intention of hotel guest-contact employees? The role of wisdom leadership and job embeddedness. Journal of Hospitality Marketing & Management 32 (2 ): 150–173. San Martín H Herrero Á (2012) Influence of the user’s psychological factors on the online purchase intention in rural tourism: Integrating innovativeness to the UTAUT framework. Tourism Management 33 (2 ): 341–350. Saydam MB Arici HE Koseoglu MA (2022) How does the tourism and hospitality industry use artificial intelligence? A review of empirical studies and future research agenda. Journal of Hospitality Marketing & Management 31 (8 ): 908–936. Seyitoğlu F Ivanov S (2020) A conceptual framework of the service delivery system design for hospitality firms in the (post-) viral world: The role of service robots. International Journal of Hospitality Management 91 : 102661.32952262 Shin H Kang J (2020) Reducing perceived health risk to attract hotel customers in the COVID-19 pandemic era: Focused on technology innovation for social distancing and cleanliness. International Journal of Hospitality Management 91 : 102664.32921871 So KF King C Sparks BA , et al. (2013) The influence of customer brand identification on hotel brand evaluation and loyalty development. International Journal of Hospitality Management 34 : 31–41. Talwar S Kaur P Nunkoo R , et al. (2022) Digitalization and sustainability: Virtual reality tourism in a post pandemic world. Journal of Sustainable Tourism: 1–28. Tussyadiah I (2020) A review of research into automation in tourism: Launching the Annals of Tourism Research Curated Collection on Artificial Intelligence and Robotics in Tourism. Annals of Tourism Research 81 : 102883. Tussyadiah I Zach F Wang J (2020) Do travelers trust intelligent service robots? Annals of Tourism Research 81 : 102886. Venkatesh V (2000) Determinants of perceived ease of use: Integrating control, intrinsic motivation, and emotion into the technology acceptance model. Information Systems Research 11 (4 ): 342–365. Venkatesh V Thong J Xu X (2012) Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Quarterly 36 (1 ): 157–178. Walsh G Mitchell V Jackson P , et al. (2009) Examining the antecedents and consequences of corporate reputation: A customer perspective. British Journal of Management 20 (2 ): 187–203. Wang L Law R Guillet B , et al. (2015) Impact of hotel website quality on online booking intentions: eTrust as a mediator. International Journal of Hospitality Management 47 : 108–115. ‏. Wu R Lee J Tian X (2021) Determinants of the intention to use cross-border mobile payments in Korea among Chinese tourists: An integrated perspective of UTAUT2 with TTF and ITM. Journal of Theoretical and Applied Electronic Commerce Research 16 (5 ): 1537–1556. Zeng Z Chen P Lew A (2020) From high-touch to high-tech: COVID-19 drives robotics adoption. Tourism Geographies 22 (3 ): 724–734. Zhao X Lynch JG Jr Chen Q (2010) Reconsidering Baron and Kenny: Myths and truths about mediation analysis. Journal of Consumer Research 37 (2 ): 197–206. Zhong L Sun S Law R , et al. (2020) Impact of robot hotel service on consumers’ purchase intention: a control experiment. Asia Pacific Journal of Tourism Research 25 (7 ): 780–798.
PMC010xxxxxx/PMC10291207.txt	==== Front Exp Biol Med (Maywood) Exp Biol Med (Maywood) EBM spebm Experimental Biology and Medicine 1535-3702 1535-3699 SAGE Publications Sage UK: London, England 37350444 10.1177/15353702231171902 10.1177_15353702231171902 Original Research Predictive value of serum HIF-1α and VEGF for arrhythmia in acute coronary syndrome patients Li Bin 1 Feng Qiuting 2 Yu Cheng 2 Yang Jun 2 Qin Xian 2 Li Xing 2 Cao Jianing 2 Xu Xin 2 Yang Chenjian 2 https://orcid.org/0000-0003-2234-195X Jin Yan 12 1 Wuxi No.2 People’s Hospital, The Affiliated Wuxi Clinical College of Nantong University, Wuxi 214002, China 2 Department of Cardiology, Jiangnan University Medical Center (JUMC), Wuxi 214002, China Yan Jin. Email: epjinyan@njmu.edu.cn 23 6 2023 23 6 2023 1535370223117190228 12 2022 22 2 2023 © 2023 by the Society for Experimental Biology and Medicine 2023 The Society for Experimental Biology and Medicine This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Percutaneous coronary intervention (PCI) has been widely used in the alleviation of myocardial ischemia in patients with acute coronary syndrome (ACS). However, the incidence of reperfusion arrhythmia (RA) after PCI is high, which seriously affects the prognosis of ACS patients. Therefore, this study aimed to study the predictive value of serum HIF-1α and VEGF levels before PCI for RA in ACS patients post PCI. A total of 200 ACS patients who underwent PCI were selected and divided into those with RA after PCI (RA, n = 93) and those without RA after PCI (non-RA, n = 107) according to Lown grade. Spearman correlation analysis was applied for the relationship between serum hypoxia inducible factor-1α (HIF-1α) and vascular endothelial growth factor (VEGF) levels and Lown grade. Patients with RA after PCI tended to have higher levels of creatine kinase muscle and brain isoenzyme (CK-MB), serum HIF-1α and VEGF before surgery. Low left ventricular ejection fraction (LVEF), high CK-MB, high serum VEGF and HIF-1α were risk factors for RA in ACS patients within 24 h after PCI. Receiver operating characteristic (ROC) analysis revealed that serum HIF-1α and VEGF levels could predict RA in ACS patients after PCI, and the combined detection could increase the sensitivity of single HIF-1α detection and the specificity of single VEGF detection. Lown grade was positively correlated with the serum HIF-1α and VEGF concentrations. In conclusion, serum HIF-1α and VEGF levels before PCI are risk factors for the occurrence of RA in ACS patients after PCI, and have certain predictive values for the occurrence of RA in ACS patients after PCI. PCI ACS RA VEGF HIF-1α edited-statecorrected-proof typesetterts1 ==== Body pmcImpact Statement The study demonstrated that Lown grade was positively correlated with the serum HIF-1α and VEGF concentrations. Introduction Acute coronary syndrome (ACS) is related to the rupture of unstable plaques in coronary artery sclerosis.1,2 The clinical manifestations of ACS are episodic chest tightness and chest pain. 3 In severe cases, heart failure and even sudden death may occur. 3 At present, percutaneous coronary intervention (PCI) is a common method to improve myocardial ischemia in patients with ACS. 4 It could effectively dredge occluded blood vessels and improve myocardial perfusion of patients. 5 However, after coronary recanalization, ACS patients have a high probability of reperfusion arrhythmia (RA), which seriously affects the prognosis of PCI. 6 Relevant studies have shown that RA could increase the risk of re-infarction after PCI in patients with myocardial infarction, and even lead to death. 7 Therefore, exploring the factors related to RA in patients with ACS after PCI is conducive to early intervention to reduce the occurrence of complications after PCI. Hypoxia-inducible factor-1 (HIF-1) is a heterodimeric transcription factor composed of a constitutively expressed β subunit (HIF-1β) and an oxygen-dependent α subunit (HIF-1α). 8 Among them, HIF-1α is the main regulator of hypoxia signaling pathway. 9 HIF-1α degradation is blocked under hypoxia, resulting in rapid accumulation. 10 Numerous studies have confirmed that HIF-1α is widely expressed in cardiovascular dysfunctions, such as acute myocardial infarction and cardiomyopathy, and plays an important role in regulating cellular oxygen supply and energy metabolism, cell proliferation and apoptosis, and new angiogenesis.11,12 HIF-1α regulates a variety of downstream genes involved in multiple physiological and pathological processes,13,14 including vascular endothelial growth factor (VEGF). VEGF plays a critical role in promoting the blood vessels generation and is important in the pathological process of ACS. 15 Currently, it has been revealed that there was a correlation between serum HIF-1α and VEGFa levels and myocardial injury, 16 but few studies have been done on the correlation between ACS patients with RA after PCI. In view of this, this study explored the predictive value of serum HIF-1α and VEGF levels before PCI for RA in ACS patients post PCI. Materials and methods Patients A total of 200 ACS patients were selected. This study was approved by the ethics committee of Wuxi No.2 People’s Hospital. All diagnosis and treatment measures were collected from patients and their families with informed consent and signed informed consent. The diagnostic criteria of ACS refer to the “2015 China Emergency Acute Coronary Syndrome Clinical Practice Guidelines Diagnostic Criteria.” ST-segment elevation myocardial infarction (STEMI): severe chest pain lasting >30 min; ST-segment arched dorsal elevation on ECG; cardiac troponin T or I positive; hybrid creatine kinase isoenzyme level >2 times of the reference value. Diagnosis of non-ST-segment elevation myocardial infarction (NSTEMI): persistent chest pain; transient or new ST-segment depression, or T wave inversion and flattening on ECG; cardiac troponin T or I positive; hybrid creatine kinase isoenzyme levels >2 times of the reference value. Diagnosis of unstable angina (UA): chest pain; ST-segment depression or T wave inversion and flattening on ECG; cardiac troponin T or I negative; hybrid creatine kinase isoenzyme levels may be elevated but ⩽2 times of the reference value. Inclusion criteria: patients diagnosed with ACS by the above diagnostic criteria for the first time; the time from onset to admission was ⩽48 h; PCI was performed; informed consent was given. Exclusion criteria: old myocardial infarction; associated with heart failure, valvular heart disease and other heart diseases; contraindications to PCI; history of preoperative arrhythmia; arrhythmia or taking antiarrhythmic drugs at admission; infection, autoimmune disease, malignant tumor, and abnormal liver and kidney function. Lown grade All patients underwent 24-h Holter monitoring during PCI to monitor the occurrence of RA at any time. RA was graded using the Lown grading methods. In this study, patients with RA were classified as Lown grade ⩾ 1. Grade 0: normal; Grade 1: Occasional premature ventricular contractions with a frequency of less than 30 beats/h or 6 beats/min; Grade 2: Premature ventricular contractions frequently occur, with a frequency of more than 30 times/h or 6 times/min; Grade 3: Occurrence of polygenic or polymorphic premature ventricular contractions; Grade 4: Frequent paired premature ventricular contractions or recurrent ventricular tachycardia; Grade 5: Ron T or ventricular fibrillation occurs. Determination of serum HIF-1α and VEGF levels Before PCI, 3 mL of fasting venous blood was collected from ACS patients, and was placed in a vacuum anticoagulation tube. Serum HIF-1α and VEGF concentrations were measured using the enzyme-linked immunosorbent assay (ELISA) following standard instructions. Human/Mouse HIF-1α ELISA Kit was purchased from Beyotime (PH368, Shanghai, China) and Human VEGF ELISA Kit (ab222510) was purchased from Abcam (Cambridge, MA). Statistical methods The data presented are mean ± standard deviation (SD) or n (percentage). The comparisons of data were done by Mann–Whitney test or Fisher’s exact test or chi-square test. Anderson–Darling test, D’Agostino & Pearson test, Shapiro–Wilk test and Kolmogorov–Smirnov test were used to test the normality of the data before analysis. Multivariate logistic regression analysis was used to analyze the risk factors of RA in patients with ACS after PCI. P < 0.05 was regarded as statistically significant. Results Baseline characteristics of ACS patients before PCI with or without RA onset A total of 200 patients with ACS who underwent PCI were included in this study. All patients underwent 24 h dynamic electrocardiography during PCI to monitor the occurrence of RA. The patients with RA were classified by the Lown grading method. In this study, the patients with RA were considered to have a Lown grading ⩾ 1. According to this grading standard, we divided patients into those with RA after PCI (RA, n = 93) and those without RA after PCI (non-RA, n = 107). We then compared the demographic data of ACS patients before PCI (Table 1). No significant differences were found between the two groups in age (56.8 ± 9.2 versus 57.6 ± 9.9, P = 0.217), gender (47.7% versus 55.9%, P = 0.259), ACS type (P = 0.555), BMI (23.45 ± 3.92 versus 24.01 ± 4.13, P = 0.183), smoking history (41.1% versus 52.7%, P = 0.119), diabetes history (26.2% versus 37.6%, P = 0.094), hypertension (29.9% versus 41.9%, P = 0.103), and hyperlipidemia (40.2% versus 50.5%, P = 0.156). In addition, there were no significant differences in heart rate (76.81 ± 13.29 versus 79.32 ± 14.46, P = 0.139), blood pressure (128.47 ± 24.96 versus 135.28 ± 26.72, P = 0.269), high density lipoprotein cholesterol (1.52 ± 0.63 versus 1.35 ± 0.74, P = 0.219), low density lipoprotein cholesterol (2.81 ± 1.02 versus 3.07 ± 0.94, P = 0.132), and total cholesterol (3.81 ± 1.21 versus 4.19 ± 1.37, P = 0.224) before PCI. However, there were significant differences between the two groups in left ventricular ejection fraction (LVEF) (58.97 ± 8.68 versus 51.25 ± 9.16, P < 0.001), creatine kinase muscle, and brain isoenzyme (CK-MB) (118.95 ± 45.73 versus 205.44 ± 51.26, P < 0.001), HIF-1α (328.43 ± 133.41 versus 422.54 ± 163.51, P < 0.001), and VEGF (61.62 ± 24.37 versus 75.33 ± 29.90, P < 0.001) levels. Table 1. Baseline characteristics of ACS patients with and without arrhythmia (RA) onset after percutaneous coronary intervention (PCI). Non-RA (n = 107) RA (n = 93) P value Age (years) 56.8 ± 9.2 57.6 ± 9.9 0.217 Gender Male 51 (47.7%) 52 (55.9%) 0.259 Female 56 (52.3%) 41 (44.1%) Body mass index (kg/m2) 23.45 ± 3.92 24.01 ± 4.13 0.183 Prior or current smoke 44 (41.1%) 49 (52.7%) 0.119 Prior or current diabetes mellitus 28 (26.2%) 35 (37.6%) 0.094 Prior or current hypertension 32 (29.9%) 39 (41.9%) 0.103 Prior or current hyperlipidemia 43 (40.2%) 47 (50.5%) 0.156 Heart rate (b.p.m.) 76.81 ± 13.29 79.32 ± 14.46 0.139 LVEF (%) 58.97 ± 8.68 51.25 ± 9.16 <0.001 SBP (mmHg) 128.47 ± 24.96 135.28 ± 26.72 0.296 DBP (mmHg) 87.16 ± 15.95 91.98 ± 17.26 0.382 HDL-C (mmol/L) 1.52 ± 0.63 1.35 ± 0.74 0.219 LDL-C (mmol/L) 2.81 ± 1.02 3.07 ± 0.94 0.132 TC (mmol/L) 3.81 ± 1.21 4.19 ± 1.37 0.224 TG (mmol/L) 1.47 ± 0.63 1.73 ± 0.71 0.118 CK- MB (ng/mL) 118.95 ± 45.73 205.44 ± 51.26 <0.001 Serum HIF-1α (pg/mL) 328.43 ± 133.41 422.54 ± 163.51 <0.001 Serum VEGF (pg/mL) 61.62 ± 24.37 75.33 ± 29.90 <0.001 ACS classification STEMI 20 (18.7%) 23 (24.7%) 0.555 NSTEMI 45 (42.1%) 38 (40.9%) UA 42 (39.2%) 32 (34.4%) RA: reperfusion arrhythmia; LVEF: left ventricular ejection fraction; SBP: systolic blood pressure; DBP: diastolic blood pressure; HDL-C: high density lipoprotein cholesterol; LDL-C: low density lipoprotein cholesterol; TC: total cholesterol; TG: triglyceride; CK-MB: creatine kinase-muscle/brain; HIF-1α: hypoxia inducible factor 1α; VEGF: vascular endothelial growth factor; ACS: acute coronary syndrome; STEMI: ST-elevation myocardial infarction; NSTEMI: non-ST-elevation myocardial infarction; UA: unstable angina; SD: standard deviation. The data presented are mean ± SD or n (percentage). The comparisons of data were done by Mann–Whitney test or Fisher’s exact test or Chi-square test. Risk factors for RA in ACS patients after PCI We then used multivariate logistic regression analysis to analyze the risk factors for RA within 24 h post PCI. The occurrence of RA (0 = no, 1 = yes) was used as the dependent variable, the detection indicators in Table 1 were used as independent variables, and the stepwise method was used to exclude irrelevant items (P > 0.05). The results showed that low LVEF (OR = 2.794, 95% CI 2.164 to 4.693, P < 0.001), high CK-MB (OR = 1.968, 95% CI 1.216 to 3.023, P = 0.002), high serum HIF-1α (OR = 1.428, 95% CI 1.119 to 1.907, P = 0.015), and high serum VEGF (OR = 1.428, 95% CI 1.119 to 1.907, P = 0.006) concentrations were risk factors for RA in ACS patients within 24 h after PCI (Table 2). Table 2. Multivariate logistic analysis of predictors for arrhythmia in acute coronary syndrome (ACS) patients after percutaneous coronary intervention (PCI). OR 95% CI P value Low LVEF (%) 2.794 2.164 to 4.693 <0.001 High CK-MB 1.968 1.216 to 3.023 0.002 High serum HIF-1α 1.428 1.119 to 1.907 0.015 High serum VEGF 1.652 1.085 to 2.273 0.006 OR: odds ratio; CI: confidence interval; LVEF: left ventricular ejection fraction; CK-MB: creatine kinase-muscle/brain; HIF-1α: hypoxia inducible factor 1α; VEGF: vascular endothelial growth factor. Correlation of serum HIF-1α and VEGF concentrations in ACS patients Compared with patients without RA, HIF-1α (Figure 1(a), P < 0.001) and VEGF (Figure 1(b), P < 0.001) were significantly increased in patients with RA. A statistical positive correlation was revealed on HIF-1α and VEGF concentrations in all ACS patients (Figure 1(c), r = 0.44, P < 0.001). Figure 1. Comparisons of serum HIF-1α (a) and VEGF (b) at admission between ACS patients with RA (RA, n = 93) and without RA (non-RA, n = 107) onset after PCI. Data were shown with median (interquartile range). The comparisons of data were done by Mann–Whitney test. (c) Spearman correlation analysis of serum HIF-1α with VEGF at admission in ACS patients (n = 200). Predictive values of serum HIF-1α, VEGF, and their combination at admission for RA onset after PCI in ACS patients We explored the predictive value of serum HIF-1α and VEGF concentrations at admission and their combined detection for the occurrence of RA within 24 h after PCI in ACS patients undergoing PCI by ROC analysis (Figure 2(a) to (c)). As shown in Table 3, the area under curve (AUC) of HIF-1α level predicting AR after PCI in ACS patients was 0.66 (95% CI, 0.59 to 0.74, sensitivity 44.09%, specificity 85.05%, Youden index 0.29, P < 0.001). The AUC of serum VEGF level predicting AR after PCI in ACS patients was 0.64 (95% CI, 0.56 to 0.72, sensitivity 68.82%, specificity 57.01%, Youden index 0.26, P < 0.001). The AUC of serum HIF-1α combined VEGF level predicting AR after PCI in ACS patients was 0.72 (95% CI, 0.65 to 0.79, sensitivity 70.97%, specificity 67.29%, Youden index 0.38, P < 0.001). Figure 2. ROC analysis of predictive values of (a) serum HIF-1α, (b) VEGF and (c) their combination at admission for RA onset after PCI in ACS patients. Table 3. Predictive values in ROC analysis. AUC (95% CI) Sensitivity (%) Specificity (%) P Youden index Serum HIF-1α 0.66 (0.59 to 0.74) 44.09 85.05 <0.001 0.29 Serum VEGF 0.64 (0.56 to 0.72) 68.82 57.01 <0.001 0.26 HIF-1α + VEGF 0.72 (0.65 to 0.79) 70.97 67.29 <0.001 0.38 ROC: receiver operating characteristic; AUC: area under curve; CI: confidence interval; HIF-1α: hypoxia inducible factor-1α; VEGF: vascular endothelial growth factor. Correlation of serum HIF-1α/VEGF concentrations and Lown grade in ACS patients following RA onset after PCI In this study, the Lown grading method was used to classify the arrhythmia of patients. The higher the grade, the more serious the arrhythmia. Therefore, we analyzed the correlation between the Lown grade and HIF-1α and VEGF concentrations at admission in 93 patients who developed RA after PCI. With the increase of Lown grade, the serum HIF-1α (Figure 3(a), r = 0.31, P = 0.002) and VEGF (Figure 3(b), r = 0.29, P = 0.005) concentrations also increased gradually at admission, showing a significant positive correlation. Figure 3. Spearman correlation analysis of Lown grade in ACS patients following RA (n = 93) onset after PCI with their (a) serum HIF-1α and (b) VEGF at admission. Discussion RA is a common complication caused by myocardial reperfusion injury during PCI in ACS patients. 17 During recanalization of diseased coronary vessels during PCI, due to the influence of energy metabolism disorder, calcium iron overload, inflammatory reaction, formation of oxygen free radicals, and other factors, myocardial damage is not improved but aggravated, which will induce RA. 18 The occurrence of RA indicates that the myocardium of ACS patients is further damaged, which predicts the poor prognosis of the patients. 19 Therefore, preventive treatment for high-risk RA patients is required before PCI to reduce the incidence of RA and ensure a good prognosis for patients. At present, there is a lack of clinical serum markers that could assess the risk of RA in patients with ACS during PCI. This study selected 200 ACS patients admitted to our hospital from December 2020 to March 2022, and evaluated the correlation between HIF-1α and VEGF concentration at admission and RA onset in ACS patients after PCI. We wanted to demonstrate the predictive effect of the two indicators and their combined detection on the occurrence of RA, so as to provide reasonable suggestions for the early intervention of RA after PCI in patients with ACS. HIF-1α is a heterodimer composed of an α subunit precisely regulated by oxygen concentration and a stably expressed β subunit. 20 The half-life of HIF-1α protein in most cells is only 5–10 min. 21 Its proline residue is hydroxylated by proline hydroxylase (PHD), which is rapidly degraded by the oxygen-dependent-ubiquitin protease pathway. 22 Therefore, HIF-1α protein is maintained at lower levels in cells under normal conditions. 23 Hypoxic environment leads to inhibition of PHD activity. 23 HIF-1α begins to express stably and migrates from the cytoplasm to the nucleus, where it combines with HIF-1β in the nucleus to form an active dimer. 24 The activated HIF-1 binds to the hypoxia response element and transcriptional coactivator p30 in the regulatory region of the target gene to form a transcription initiation complex and induce the expression of downstream genes. 24 The number of genes regulated by HIF-1α exceeds 1000, including erythropoietin, VEGF, inducible nitric oxide synthase (iNO), glucose transporter 1 (GLUT1), endothelin-1 (ET-1), and so on. 25 In terms of heart disease, HIF-1α could increase myocardial glucose uptake and transport by regulating the expression of myocardial GLUT4 and pyruvate kinase M2 isoform (PKM2) to continuously provide compensatory energy supply. 26 In addition, HIF-1α regulates the expression of GLUT1 by targeting in the process of cardiac ischemia and hypoxia, which increases the transport of peripheral blood glucose to vascular endothelial cells, and participates in the protection of mitochondrial function of vascular endothelial cells in hemorrhagic shock. 27 In this research, we found that HIF-1α was remarkably elevated in patients with RA, and in all ACS patients, HIF-1α and VEGF concentrations were significantly positively correlated, which is consistent with our description of other studies above. Innovatively, we demonstrated that HIF-1α concentrations at admission in ACS patients undergoing PCI have a statistically predictive value for the occurrence of RA within 24 h after surgery. VEGF plays important physiological functions by binding to related receptors. 28 VEGF has important roles such as inducing endothelial cell migration and proliferation, increasing vascular permeability, and regulating thrombosis. 29 Previous researches have revealed that VEGF plays a role in the occurrence and development of certain cardiovascular diseases. 30 Inhibition of VEGFs and the receptors can cause a variety of cardiovascular disease complications. 30 The more severe the lesion, the higher the level of VEGF. 30 The concentration of serum VEGF was positively correlated with myocardial oxidative damage indicators such as myeloperoxidase and advanced oxidation protein products, suggesting that VEGF can accurately reflect the degree of myocardial damage in ACS. In addition, a study indicated that the isoform of VEGF-A, VEGF-A165b, has anti-angiogenic effects, and the ratio of VEGF-A to VEGF-A may be a tool for predicting cardiovascular dysfunctions in patients with AMI after PCI. 31 VEGF and HIF-1α have been studied extensively in the context of cardiovascular disease. Both molecules are involved in the regulation of angiogenesis and are upregulated in various cardiovascular diseases. 32 Studies have shown that both molecules are involved in the regulation of cardiac remodeling and can protect the heart from ischemic injury. 33 In addition, VEGF and HIF-1α can induce cardiomyocyte proliferation and protect the heart from ischemic injury. Many studies have focused on the role of VEGF and HIF-1α in the context of cardiac disease. 34 For example, a recent study has demonstrated that VEGF and HIF-1α can induce cardiomyocyte proliferation and protect the heart from ischemic injury in a mouse model of myocardial infarction. 35 In this study, we demonstrated that patients who developed RA after PCI tended to have higher levels of serum VEGF and HIF-1α preoperatively, which is consistent with the statistical findings of previous studies in blood samples from patients with other cardiovascular diseases. More importantly, we demonstrated that with increasing Lown grade, serum VEGF and HIF-1α levels tended to increase in patients on admission, showing a significant positive correlation. Our data suggest that the higher the serum HIF-1α and VEGF concentration at admission, the more severe the RA after PCI. In addition, our results suggest that combined detection could increase the sensitivity of single HIF-1α detection and the specificity of single VEGF detection. In conclusion, we found that preoperative serum HIF-1α and VEGF levels were significantly elevated in ACS patients with RA after PCI compared with patients without arrhythmias. We demonstrated that HIF-1α and VEGF were risk factors for RA within 24 h after emergency PCI in patients with ACS. Authors’ Contributions: BL, QTF, CY, JY, XQ, XL, JNC, XX, CJY, and YJ performed the experiments, and analyzed and interpreted the data. YJ was a major contributor in writing the manuscript. All authors read and approved the final manuscript. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iD: Yan Jin https://orcid.org/0000-0003-2234-195X ==== Refs References 1 Gach O El HZ Lancellotti P. Acute coronary syndrome. Rev Med Liege 2018;73 :243–50 2 Feng H Pan H Yao W. Uric acid levels are associated with severity and mortality in patients with acute coronary syndrome. STEMedicine 2022;3 :e131 3 Makki N Brennan TM Girotra S. Acute coronary syndrome. J Intensive Care Med 2015;30 :186–200 24047692 4 Smith JN Negrelli JM Manek MB Hawes EM Viera AJ. Diagnosis and management of acute coronary syndrome: an evidence-based update. J Am Board Fam Med 2015;28 :283–93 5 Bhatt DL. Percutaneous coronary intervention in 2018. JAMA 2018; 319 :2127–8 6 Taha H Shaker MM. Percutaneous management of reperfusion arrhythmias during primary percutaneous coronary intervention: a case report. Egypt Heart J 2021;73 :30 33765245 7 Lee MS Dahodwala MQ. Percutaneous coronary intervention for acute myocardial infarction due to unprotected left main coronary artery occlusion: status update 2014. Catheter Cardiovasc Interv 2015; 85 :416–20 8 Ke Q Costa M. Hypoxia-inducible factor-1 (HIF-1). Mol Pharmacol 2006;70 :1469–80 9 Zhang FJ Luo W Lei GH. Role of HIF-1alpha and HIF-2alpha in osteoarthritis. Joint Bone Spine 2015;82 :144–7 10 Shukla SD Walters EH Simpson JL Keely S Wark PAB O’Toole RF Hansbro PM. Hypoxia-inducible factor and bacterial infections in chronic obstructive pulmonary disease. Respirology 2020;25 :53–63 31663668 11 Luo Y Teng X Zhang L Chen J Liu Z Chen X Zhao S Yang S Feng J Yan X. CD146-HIF-1alpha hypoxic reprogramming drives vascular remodeling and pulmonary arterial hypertension. Nat Commun 2019; 10 :3551 31391533 12 Warbrick I Rabkin SW. Hypoxia-inducible factor 1-alpha (HIF-1alpha) as a factor mediating the relationship between obesity and heart failure with preserved ejection fraction. Obes Rev 2019;20 :701–12 13 Fitzpatrick SF. Immunometabolism and sepsis: a role for HIF? Front Mol Biosci 2019;6 :85 31555665 14 Archid R Solass W Tempfer C Konigsrainer A Adolph M Reymond MA Wilson RB. Cachexia anorexia syndrome and associated metabolic dysfunction in peritoneal metastasis. Int J Mol Sci 2019;20 :5444 31683709 15 Claesson-Welsh L Welsh M. VEGFA and tumour angiogenesis. J Intern Med 2013;273 :114–27 16 Guo BQ Xu JB Xiao M Ding M Duan LJ. Puerarin reduces ischemia/reperfusion-induced myocardial injury in diabetic rats via upregulation of vascular endothelial growth factor A/angiotensin-1 and suppression of apoptosis. Mol Med Rep 2018;17 :7421–7 17 Ilia R Zahger D Cafri C Abu-Ful A Weinstein JM Yaroslavtsev S Gilutz H Amit G. Predicting survival with reperfusion arrhythmias during primary percutaneous coronary intervention for acute myocardial infarction. Isr Med Assoc J 2007;9 :21–3 18 Kumar S Sivagangabalan G Thiagalingam A West EB Narayan A Sadick N Ong AT Kovoor P. Effect of reperfusion time on inducible ventricular tachycardia early and spontaneous ventricular arrhythmias late after ST elevation myocardial infarction treated with primary percutaneous coronary intervention. Heart Rhythm 2011;8 :493–9 19 Angiolillo DJ Bhatt DL Cannon CP Eikelboom JW Gibson CM Goodman SG Granger CB Holmes DR Lopes RD Mehran R Moliterno DJ Price MJ Saw J Tanguay JF Faxon DP. Antithrombotic therapy in patients with atrial fibrillation treated with oral anticoagulation undergoing percutaneous coronary intervention: a North American perspective: 2021 update. Circulation 2021;143 :583–96 20 Güven Bağla A Içkin Gülen M Ercan F Aşgün F Ercan E Bakar C . Changes in kidney tissue and effects of erythropoietin after acute heart failure. Biotech Histochem 2018;93 :340–53 21 Han HJ Saeidi S Kim SJ Piao JY Lim S Guillen-Quispe YN Choi BY Surh YJ. Alternative regulation of HIF-1alpha stability through phosphorylation on Ser451. Biochem Biophys Res Commun 2021;545 :150–6 22 Jaakkola P Mole DR Tian YM Wilson MI Gielbert J Gaskell SJ von Kriegsheim A Hebestreit HF Mukherji M Schofield CJ Maxwell PH Pugh CW Ratcliffe PJ. Targeting of HIF-alpha to the von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science 2001;292 :468–72 23 Semenza GL. Hypoxia-inducible factor 1 (HIF-1) pathway. Sci STKE 2007;2007 :cm8 24 Vriend J Reiter RJ. Melatonin and the von Hippel-Lindau/HIF-1 oxygen sensing mechanism: a review. Biochim Biophys Acta 2016;1865 :176–83 25 Lee JW Bae SH Jeong JW Kim SH Kim KW. Hypoxia-inducible factor (HIF-1)alpha: its protein stability and biological functions. Exp Mol Med 2004;36 :1–12 15031665 26 Sakagami H Makino Y Mizumoto K Isoe T Takeda Y Watanabe J Fujita Y Takiyama Y Abiko A Haneda M. Loss of HIF-1alpha impairs GLUT4 translocation and glucose uptake by the skeletal muscle cells. Am J Physiol Endocrinol Metab 2014;306 :E1065–76 27 Tang W Long T Li F Peng C Zhao S Chen X Su J. HIF-1alpha may promote glycolysis in psoriasis vulgaris via upregulation of CD147 and GLUT1. Zhong Nan Da Xue Xue Bao Yi Xue Ban 2021;46 :333–44 28 Melincovici CS Boşca AB Şuşman S Mărginean M Mihu C Istrate M Moldovan IM Roman AL Mihu CM. Vascular endothelial growth factor (VEGF) – key factor in normal and pathological angiogenesis. Rom J Morphol Embryol 2018;59 :455–67 29 Porta M Striglia E. Intravitreal anti-VEGF agents and cardiovascular risk. Intern Emerg Med 2020;15 :199–210 31848994 30 Yla-Herttuala S Baker AH. Cardiovascular gene therapy: past, present, and future. Mol Ther 2017;25 :1095–106 31 Harada K Kikuchi R Ishii H Shibata Y Suzuki S Tanaka A Suzuki A Hirayama K Murohara T. Association between the ratio of anti-angiogenic isoform of VEGF-A to total VEGF-A and adverse clinical outcomes in patients after acute myocardial infarction. Int J Cardiol Heart Vasc 2018;19 :3–7 29946556 32 Xiang Y Yao X Wang X Zhao H Zou H Wang L Zhang QX. Houshiheisan promotes angiogenesis via HIF-1alpha/VEGF and SDF-1/CXCR4 pathways: in vivo and in vitro. Biosci Rep 2019;39 : BSR20191006 33 Zhang D Lv FL Wang GH. Effects of HIF-1alpha on diabetic retinopathy angiogenesis and VEGF expression. Eur Rev Med Pharmacol Sci 2018;22 :5071–6 34 Fan J Lv H Li J Che Y Xu B Tao Z Jiang W. Roles of Nrf2/HO-1 and HIF-1alpha/VEGF in lung tissue injury and repair following cerebral ischemia/reperfusion injury. J Cell Physiol 2019;234 :7695–707 35 Liu J Wang W Wang L Chen S Tian B Huang K Corrigan CJ Ying S Wang W Wang C. IL-33 initiates vascular remodelling in hypoxic pulmonary hypertension by up-regulating HIF-1alpha and VEGF expression in vascular endothelial cells. EbioMedicine 2018;33 :196–210 29921553
PMC010xxxxxx/PMC10291208.txt	==== Front JMT spjmt Journal of Music Teacher Education 1057-0837 1945-0079 SAGE Publications Sage CA: Los Angeles, CA 10.1177/10570837231178891 10.1177_10570837231178891 Research Articles Experiences of First-Year Music Teachers in Hawai‘i During the COVID-19 Pandemic: Challenges, Adaptability, and Implications for Future Music Teaching https://orcid.org/0000-0001-9674-8790 Matherne Nicholas 1 1 University of Hawai‘i at Mānoa, Honolulu, USA Nicholas Matherne, University of Melbourne, 234 St Kilda Rd, Southbank VIC 3006, Australia. Email: n.matherne@unimelb.edu.au * Nicholas Matherne is now affiliated with The University of Melbourne, Melbourne, Victoria, Australia. 23 6 2023 23 6 2023 10570837231178891© National Association for Music Education 2023 2023 MENC: The National Association for Music Education This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. The COVID-19 pandemic presented challenges to delivering music content through distance learning, adding to what is often already a stressful and challenging time for first-year educators. The purpose of this study was to explore the experiences of six first-year music teachers who began teaching in Hawai‘i during the COVID-19 pandemic. I interviewed participants twice during the summer after their first year of teaching and collected background information and teaching artifacts. Despite the year’s challenges, the teachers maintained their commitment to teaching. They did note that they felt underprepared for this experience by their undergraduate education and lack of prior teaching experience. Themes suggested the importance of teacher mentors and professional networks for the novice teachers, and that these teachers could have benefited from ongoing support from their preservice teacher education programs. In addition, the experiences of these teachers highlight the importance of adaptability in learning for both students and teachers. adaptability early career teachers mentoring professional networks edited-statecorrected-proof typesetterts1 ==== Body pmcThe first year of teaching is often a stressful and challenging experience for young educators of all disciplines (McLean et al., 2020). In particular, music teachers can feel isolated and stressed when they begin their first jobs (Conway, 2015; Stringham & Snell, 2019) in which they are often a member of a small community of music teachers or in some cases the only music teacher in the building. Furthermore, novice teachers often have unrealistic expectations of how success will look in their classrooms (Conway, 2015) and have difficulty being adaptive as teachers. Music teacher education programs are tasked with preparing preservice teachers for the realities of teaching, but inservice teachers have highlighted a disconnect between their tertiary coursework and what is needed to be successful as a teacher (Ballantyne & Packer, 2004). This disconnect suggests the need for music teacher education programs to examine the experiences of novice teachers, perhaps viewing the first year of teaching as an extension of teacher education. The restrictions and conditions during the 2020 COVID-19 outbreak increased overall stress and worries in adults, adolescents, and children (Rothe et al., 2021), and educators were no exception. As a part of the greater pandemic-induced stress, many administrators reported higher levels of teacher attrition in 2020 than in prior years, with many teachers that left the profession citing stress as one of their biggest reasons for their exit (Diliberti et al., 2021). Teachers transitioning to distance learning experienced increased workloads with new expectations, being held accountable for methods of delivery in which few had previous preparation (Kaden, 2020). During the transition to distance learning, music teachers reported challenges with “sustaining remote learning to the end of the year” and “planning appropriate instruction feasible through remote learning” (Hash, 2021, p. 389) due to the synchronous nature of traditional ensemble music instruction and the limitations of current online video conferencing software. Joseph and Lennox (2021) also reported challenges in adapting their music curriculum in their own teaching practice, as well as detriments to well-being during the initial transition to distance learning. In particular, novice teachers reported that they felt overwhelmed again with the initial transition to distance learning after finally getting comfortable in their classrooms (Mecham et al., 2021). The COVID-19 pandemic forced educators to reinvent and reimagine their teacher identities, creating dilemmas that novice teachers in particular struggled with as their professional identities were not yet fully formed (Dvir & Schatz-Oppenheimer, 2020). Within the context of music, Hash (2021) found that novice music teachers reported lower levels of student participation during the transition to distance learning. Though online engagement in learning is hardly a music-specific problem, this highlights one of the challenges that forced young music teachers to adapt and adjust as they were developing skills as novice teachers. Adaptability as a concept can be used “as a means of understanding young people’s capacity to deal with new, changing, and/or challenging situations” (Martin, 2010 as cited in Martin, 2012, p. 90). Martin asserted that adaptability is importantly different from concepts such as coping because while coping merely implies survival, adaptability’s end is a positive change. As adaptability is seen to lead to a “positive trajectory” for student learning and achievement (Martin, 2012), adaptability is important to some aspects of success in teaching. Furthermore, evidence suggests that teachers who exhibited higher levels of adaptability tended to have higher levels of well-being and were better able to meet the demands of their work (Collie & Martin, 2017). Conway and Hibbard (2019) suggested that adaptive expertise, which refers to the ability of teachers to be adaptable in the context of teaching, is one of the goals of preservice teacher preparation. At the outset of the pandemic, Orkibi (2021) proposed a related construct, creative adaptability, which “refers to one’s ability to respond creatively and adaptively to stressful situations” (p. 3), including cognitive, behavioral, and emotional responses. Creative adaptability highlights a specific way of responding to stressors and adapting in ways which are novel, but importantly emphasizes this adaptability for both work and personal challenges. As the pandemic increased stress both inside (Hash, 2021; Kaden, 2020) and outside (Rothe et al., 2021) of teaching, it is important to consider how teachers adapted to various forms of stress and how that affected intentions to continue teaching (Diliberti et al., 2021). While adaptability is listed among Teachout’s (1997) reported skills and behaviors for effective music teaching, little research specifically examines inservice music teachers’ adaptability. A study on a music activity in social work suggested that adaptability was integral to the success of the music teachers and the program’s success (Kelly & Neidorf, 2021). Roseth and Blackwell (2022) found that university music teacher educators’ adaptability in the pandemic was correlated with well-being, consistent with research exploring adaptability and well-being of K-12 teachers (Collie & Martin, 2017). In addition, music teacher educators felt that the ongoing pandemic had a substantial impact on the need to adapt their courses (Roseth & Blackwell, 2022). Researchers suggested that the challenges presented by the pandemic provided opportunities for music teachers to showcase their adaptability and to find new and creative ways of adjusting the curriculum to meet students’ needs (de Bruin, 2021; Hash, 2021; Joseph & Lennox, 2021). However, research also suggests that teachers felt underprepared for the adaptations required to shift learning to alternative formats (Parkes et al., 2021) and that the changes in expectation exacerbated the stress associated with teaching (Miksza et al., 2021). Because stress is one of the most common reasons for teachers to leave the profession (Diliberti et al., 2021), the challenges of pandemic education show the importance of addressing teacher experiences as we develop teacher education (Mecham et al., 2021). The purpose of this case study was to explore the experiences of first-year music teachers who began teaching during the COVID-19 pandemic. Specifically, the following research questions guided my inquiry: Research Question 1: How do these first-year music teachers describe their experiences of teaching during the COVID-19 pandemic? Research Question 2: How did their experiences in their first year of music teaching differ from their expectations prior to starting their first year of teaching? Research Question 3: What challenged or surprised these music teachers, what did they discover, and how did they adapt during their first year of teaching? Research Question 4: How do these first-year music teachers describe the impact of this year on their desire to continue teaching music? Method Because I was exploring the experiences of music teachers who began their careers during the COVID-19 pandemic, I employed a multiple case study design to develop “an in-depth analysis of a case” (Creswell & Creswell, 2017, p. 12) bounded by the specific time and activity. The multiple cases were bounded by their shared circumstances (both educational and societal) at a pivotal moment in each young teacher’s career, which may yield insights into the ways in which novice teachers adapt to challenges. Institutional review board (IRB) approval was granted on May 20, 2021. Participants Participants were six first-year teachers during the 2020 to 2021 school year who recently graduated from a large public university in Hawai‘i. All teachers’ student taught in the Spring of 2020 (during the transition to distance learning) or during the Fall of 2020 (during the transition from full distance learning back to hybrid teaching) and obtained a full-time music teaching job in Hawai‘i during the 2020 to 2021 school year, teaching at least one full semester prior to the study. I contacted participants by email and social media to request participation in interviews and email correspondence regarding their experiences during their first year of teaching. To contextualize their experiences, I have provided descriptions of the teachers, their backgrounds, and the position which they obtained upon graduation using pseudonyms. Martin Martin is a tuba player who began playing in middle school and continued in a well-supported urban high school program. He elected to study music over other interests because of past successes in teaching middle school low brass and high-performance achievements. His first teaching position was as a high school band teacher in a school with a large number of children of military personnel in which he taught band, guitar, and ʻukulele. His school employed a full distance learning model for Semester 1 and hybrid for Semester 2. Arthur Arthur is a clarinet player who joined band after seeing his cousin play in the local youth symphony, noting that the performance had made him want to be a conductor. He attended a local private school before attending University where he obtained a bachelor’s degree in music and then a postbaccalaureate degree in music education. His first teaching position was as a band teacher in a middle school with a reputation of high achievement in band. His school employed a full distance learning model for Semester 1 and hybrid for Semester 2. Jack Jack is a trumpet player who began playing in middle school (the same school at which he later taught) and then attended a large suburban high school with a strong music program. He served in band leadership for many years which led him to select music education as a major, though he expressed doubts about teaching during student teaching. He was a mid-year sabbatical replacement in a large middle school with a reputation of high music achievement. The school was in full distance learning in January but began a hybrid schedule in February. Hannah Hannah is a saxophone player who began playing in middle school and attended a large suburban high school with a strong music program (the same program as Jack). She knew she wanted to teach music but began university in another major and then switched to music education in her first year. She obtained a job as a middle school band and choir teacher for a new suburban middle school opening during the 2020 to 2021 school year. The school remained in full distance learning for the entire year with “pull-ins” for student academic supports. Linda Linda is an oboe player who began playing in middle school (the same school at which she now teachers) and went to a large urban high school in Honolulu. She knew before attending university that she wanted to teach, but wasn’t sure what subject. Linda was a mid-year hire for a retiring middle school orchestra, piano, and choir teacher in a school with students from low socio-economic status backgrounds. Her school employed a hybrid schedule for the second semester. Jesse Jesse is a vocalist with background in piano and guitar. She joined choir in the 10th grade at a large suburban high school with a strong music program. She realized her desire to teach music during a meeting with her freshman advisor at university and then switched into and completed the music education program. She was hired as a choir and ʻukulele teacher in a large intermediate school in the early years of reigniting their vocal music program. Her school employed a full distance learning model for Semester 1 and a hybrid model for Semester 2. Positionality As my relationships with the participants affect both the way the data were collected and my analysis of that data (Holmes, 2020), I provide a brief explanation of my positionality for this study. I completed a graduate degree in music education at the same time as these teachers were completing their undergraduate degrees and, as such, I performed in ensembles and took classes with all of the participants, serving as a graduate assistant for some of their classes. Our relationships range from friends to acquaintances, but we were all familiar with each other and the institutional environment in which they were educated. My position as a graduate student allowed me the opportunity to develop closer relationships with them while being in a mentoring role, which extended to conversations during their first year of teaching when I also returned to secondary music teaching upon completion of my degree. The shared experience of beginning in a new school during the pandemic provided me common ground with the participants; however, it also required me to carefully evaluate differences between my own experiences, which were contextualized by past teaching, and those experiences unique to the teachers who were entering the profession. This was an important distinction noted by one of the participants. Data Collection Prior to the interviews, the first-year teachers completed an online questionnaire to gather information about their background, the instructional formats employed by their schools during the 2020 to 2021 school year, the types of activities they employed in their teaching, and their general impressions of student engagement during the year. I conducted two semistructured interviews with each participant for a total of 12 interviews, the initial interviews during the second week of summer vacation and the second interviews 2 weeks later. Each interview contained between six and eight questions about their experiences, support structures, and beliefs about teaching. The second interview included questions about their perceptions of the upcoming school year. I also asked follow-up questions based on the background questionnaire and their responses during the interviews. I designed the questions to elicit a wholistic picture of their experiences teaching during the pandemic, inspired by prior literature on adaptability and coping with stressful situations (e.g., Orkibi, 2021). For example, I asked participants, “What were your personal support structures during times of increased stress?” and “How was teaching this year different from what you had been expecting prior to the pandemic?” I met with the participants online via Zoom, recorded the audio, and transcribed it using Otter.ai. An outside researcher familiar with the project confirmed the transcript accuracy. The first interviews ranged from 23min 24s to 57min 20s, and the second interviews ranged from 12min 2s to 46min 29s. In the first interview, I asked the first-year teachers if they would be willing to share professional or instructional materials from their first year of teaching. All six provided materials they felt were representative of their teaching and/or experience including teacher evaluation reflections, lesson plans, videos, and student handouts, which I used for data triangulation. Data Analysis I completed descriptive coding (Miles et al., 2014) of the interviews to identify features of the interviewees’ unique experiences. I coded the reflections and background information teachers provided for themes as well and analyzed the lesson plans provided to identify the goals and approaches employed by each teacher. An outside music teacher education specialist reviewed the materials and verified the codes (Creswell & Creswell, 2017), clarifying nuance in the interviews and codes. I then mapped out codes visually for each participant and conducted a second cycle of coding to identify patterns from interviews, both within cases and with cross-case analysis (Miles et al., 2014) from which I identified individual and cross-case themes. Trustworthiness I utilized the materials and questionnaire responses to triangulate the themes from the interviews (Creswell & Creswell, 2017) finding similar trends to those in the second cycle of coding. Once I determined the major themes in the findings, I sent a summary to the teachers to complete a member check (Williamon et al., 2021) to ensure the accuracy and authenticity of my account with regard to their personal experiences. Though these cases are undeniably unique to the pivotal educational and societal moment, the findings may resonate with individuals’ experiences, and the insights into the adaptability of these teachers provide support for existing theories of adaptability and may be transferable to future research findings (Miles et al., 2014). Findings Guided by my research questions, I present the findings as themes using the participants’ own words to summarize their experiences and reflections on their first year of teaching. These findings are divided into (a) the unique experiences of being a first-year music teacher during the 2020 to 2021 school year, (b) the different ways in which the first-year music teachers adapted to the unforeseen circumstances surrounding pandemic teaching conditions, and (c) the first-year music teachers’ beliefs about their future in teaching and desire to continue teaching music. First-Year Experiences: “We’re All in the Same Storm, but We’re Not in the Same Boat” Regarding the first and second research questions, the first-year teachers noted a large disconnect between the university education they received and the reality of teaching during the pandemic. While first-year music teachers inevitably encounter new experiences and unexpected challenges during the first year of teaching (Conway, 2015), Jack noted “we definitely weren’t trained for [teaching online] in college at all, and like, why would we be?” Jack’s comment highlights the unforeseen outcomes of moving music learning online and the focus of teacher preparation on models which are designed for in-person teaching specifically. While all teachers felt this disconnect from their prior experience throughout the pandemic, Jesse emphasized the difference that first-year teachers felt:I don’t like how people are using the phrase, “we all feel like first year teachers” because I’m like, yeah that’s true. But then for those of us who are actual first year teachers, they kind of invalidate some of those feelings. Jesse followed up on this idea with the metaphor, “We’re all in the same storm, but we’re not in the same boat.” This quote reflects what many of the first-year teachers expressed throughout the interviews: Their lack of prior experience was a major challenge during pandemic teaching, which was a major difference from their veteran teacher colleagues. Regarding the second research question about differences from their expectations from preservice teacher preparation, the first-year teachers did note that while they felt unprepared for the educational environment, classroom management was easier online in many circumstances. Students were far less disruptive with talking or behavioral issues online, and teachers felt empowered with functions like mute in online meeting platforms. However, this was, as Linda noted, accompanied by a lack of engagement in the activities of online learning: “I can talk and they’re not talking but . . . I feel like the challenge is them actually listening and me knowing that [they are listening].” These sentiments were echoed by Linda’s peers who recounted challenges in engaging students beyond the chat log of their virtual classrooms and in their ability to check for student engagement and understanding during their class time. On a personal level, many of the teachers reported feelings of lowness unmatched by any prior time in their lives. Jack confessed that due to both personal and professional challenges, this was “personally, probably the lowest I’ve ever felt, ever.” Several of the first-year teachers also described feelings of burnout by mid-year. Hannah noted that by the end of the first semester “I could slowly start to feel myself burning out,” and Martin further explained that “I lost a lot of drive this year. I feel really burnt out already.” To cope with both professional and personal challenges, the teachers relied heavily on external supports such as mentors, peers, professional organizations, friends, and family. Many of the first-year teachers noted that teacher mentors in their building were incredibly helpful in adjusting to their new positions. Many of the teachers relied on the other music teachers in the school, and some had support from academic coaches and school-assigned mentors. By contrast, Martin explained that inside his school he felt quite isolated: “I feel like professionally, I didn’t talk to a lot of teachers. Because I just kind of sat in my room trying to figure out what the heck I’m supposed to do tomorrow.” Martin recalled that professional organizations such as the local band directors association “helped a lot,” especially since he felt unsupported in his school. When talking about his peers from university, Jack mentioned that these connections were something he highly valued. His peers helped him to refocus on the positive aspects of the job. Jack recounted, “We [my peers and I] would cheer each other on, on the small things that went well or like the funny thing that a student said.” The other first-year teachers also mentioned connections within their music education cohort and with other non-music-teacher friends from university who helped them to see things from a different perspective. As is quite common in their region, many of the first-year teachers continued to live with their families throughout university and their first year of teaching. The teachers described the support they felt from their families which continued through the pandemic, Arthur noting, “From day one, they’ve [my parents] always helped me.” This support extended to helping create instrument bell covers in Arthur’s case and support navigating the Department of Education in Jesse’s case, as her mother was a veteran teacher in the same school system. Jack confided, “I won’t tell her this, but it was nice to have my mom around. Because there were some nights where I just really needed to just cry,” which helped him cope with his feelings of lowness. Adaptability: “We Had to Find Ways to Teach the Concepts We Would Normally Teach in Different Ways” Examining the third research question, regarding adaptability, all of the first-year teachers made adjustments to their teaching, but in distinct ways which reflected varying expressions of adaptability. The adjustments fell generally into three main categories: (a) retaining typical goals but employing traditional approaches with technology, (b) altering focus to emphasize empathy while teaching broader music concepts, and (c) teaching with a person-first approach and contextualizing music in unconventional ways. Traditional Approaches Martin and Arthur relied heavily on traditional techniques such as ensemble rehearsal and working from method books, which were adjusted for online mediums. They both supplied teaching artifacts which highlight their aims toward performance-based outcomes for band programs. Arthur provided a concert recording, made in-person at the end of the school year with students in masks and instrument bell covers, a typical culminating experience in music instruction with safety modifications. Similarly, the lesson plans supplied by Martin employed a traditional ensemble rehearsal structure using technology when needed, but similar to rehearsal plans which one could employ in a typical school year. In our interviews, both Martin and Arthur reflected on their approaches to the year and how little had changed in their beliefs about the way their programs should operate. Arthur recounted how he rehearsed his students online, saying, “I tried to kind of just push through it [online lag time]. I made them play anyways. You know, so I could kind of hear.” Martin reflected that he came to expect a lower level than he had before teaching but that his beliefs about teaching and learning music remained the same: “I’d say my standards have changed. Not so much my beliefs. I feel like my beliefs are very much the same.” He recalled how the year felt defeating because he was unable to rehearse at the level he wanted (both due to the actual nature of the program he was teaching and the nature of the pandemic). Though students inquired about different music or types of activities, Martin decided to continue with isolated music theory and individual method book learning, deciding, “We’re just gonna do the thing. I’m just gonna keep doing the same thing.” This was an ongoing struggle for Martin who emphasized frustration with the disconnect between his concept of band and his perceptions of his students’ motivation. One of the most unique things that I identified about these two teachers was that they expressed feeling the comparison with the previous director and how their teaching and performance outcomes would be seen by the students and community. While each teacher reflected on the comparison differently, both teachers reported comparing their own teaching to what had been done in their established band programs previously. Martin suggested that despite the pressure he felt to be successful, comparing himself to the previous teacher actually was somewhat positive because “since the last person there wasn’t super good, I didn’t feel a lot of stress.” By comparison, Arthur felt the weight of entering a program that was well-known for high achievement and the way the last director taught:It was a little bit tough. I guess this is a both little bit professional and personal. Personally, it was a little bit tough both ways. Because, you know, coming in as the new guy, you get compared a lot to him. And especially like I said, since he was so “rawrr,” very strong and everything . . . and even the kids, they said, like, “wow, you’re like a teddy bear compared to [previous director].” These concerns with comparisons to the past also extended to concerns about how they would be seen moving forward, especially with regard to enrollment numbers and public performances. Empathy and Teaching Concepts Jack and Linda both emphasized the need for empathy for their students’ circumstances due to the pandemic. Jack recounted his perspective as, “I understand your situation, and just do the best that you can to participate as much as you’re able to.” Both teachers expressed trying to meet the students where they were, both in current academic or musical level and in personal circumstances. Both Jack and Linda began fully online, slowly transitioned to some students in person, and tried to be understanding of the challenges of student learning in a hybrid environment where students may not be able to make music or have their cameras turned on. Linda sought feedback from her students to help her meet their needs in the most understanding way possible. Jack noted that the pandemic led him to try to build rapport with his students and to engage them in different ways by asking fun questions or starting with levity. He affirmed, “building that rapport and showing the students that you can have fun and want to just chill out for a second before you go into whatever it is you’re working on that day [is] really healthy for the students.” Linda also noted that she spent more time on “a lot of things that were more geared towards social emotional learning and connecting with the students” to support them. Jack and Linda both provided teaching artifacts that showed a focus on specific musical concepts which they hoped to help their students master and which could be used in both performance and nonperformance contexts. Jack found the need to teach musical concepts in brand new ways to be exciting, saying, “something that was really positive about online teaching is we had to find ways to teach the concepts we would normally teach in different was, which was cool.” Linda created several lessons which assisted students in exploring Hawaiian choral music and facilitating discourse about authenticity in musical arranging. Linda and Jack both commented that the unusual circumstances did provide some relief because they didn’t feel pressure for achievement they might have felt in a typical year. Linda noted that she felt like “since everyone is having a hard time, I shouldn’t beat myself up [for mistakes],” a sentiment that Jack also shared. Jack reflected that this year had led him to explore more about growth mindset and to incorporate this into his personal teaching philosophy, “having a mindset of this is where I’m at, what are the things that I can do to just take the students the next step . . . is more important to me than meeting a benchmark necessarily.” Linda, when discussing a unit she did on practicing, also noted, “I hope I can train them a little better, so that they will be able to practice efficiently.” This emphasis on growth and helping students to achieve based on their current ability was present throughout both of their interviews. Person First Teaching and Teaching Adaptations Similar to Jack and Linda, Jesse and Hannah emphasized empathy and understanding in their teaching, but expressed this with a commitment to what the participants called person-first teaching. Jesse noted that her experiences, “strengthened my philosophy that our leaners are people first and students second.” Jesse recounted that she placed students’ well-being above the musical content she was teaching, saying “So I didn’t teach them dotted 16th notes, but in my class today, they felt safe and they laughed. And that’s good enough for me today.” Hannah was guided by her school’s mission to know love, give love, and accept love; she affirmed its importance in her teaching, noting that by the end of the year she felt that “as long as the students learned like five things, then that was the biggest outcome for me . . . and as long as they felt like they were loved. I felt like that was also the biggest component to this year.” In a sense, both teachers expressed that when they felt the need to make a choice between teaching content and supporting well-being, they chose the latter. Jesse also noted the importance of this foundation:I think the beauty of prioritizing relationships is that once you do that, classroom management, and all that other stuff is under control, because when they know that you care about them, then they want to be invested in the class and they want to learn. Jesse’s teaching artifacts reflected her emphasis on seeing students as people. She developed a variety of activities which included instructions that both explained the necessary steps for learning and aimed to reduce stress. For instance, on a musical heritage project’s PowerPoint, she included information about the time she would provide in class to complete the work, but also included, “Breathe, friends. This should be fun.” Both in how she addressed the students and in how she designed the project to meet students with helpful support, Jesse demonstrated how she viewed the students as people. Hannah designed her course content to give students more autonomy and control over their learning. She explored creative approaches to teaching concepts but allowed for student choice and using mediums with which the students were already familiar: “So they will do TikTok songs and stuff like that and [it] allow[s] them to feel like what they’re doing now can be used educationally or gives them even more power to their own education.” In planning for teaching, both Hannah and Jesse adapted their content and approaches to teaching based on circumstances and student success. As a new teacher building a program in its infancy, Jesse struggled to assess students’ prior knowledge, but once she understood her students’ prior experiences, she redesigned her instruction to help them develop necessary skills:I was trying to start a choir online. And that was difficult. So once I identified that that was the challenge (it was they had no context of [group singing]), I was able to kind of adjust and be like, “okay, let’s rethink this.” She explained, “I kind of transitioned back to just doing more like music with a singing emphasis” because she found that teaching group singing online was not effective and that her students’ needs were better met through this approach. Like Jack, Hannah said she found the need to do things in new and creative ways actually excited her, but led her to diverge from traditional models more than Jack, even noting that she had started to feel that the “traditional music teaching setting does not work anymore.” Despite feeling overwhelmed at times, she said, “I’m driven by pressure. The pressure of knowing that I can’t do things in a traditional sense excited me.” Hannah found the fixed mindset of some peers to be quite frustrating and she eventually avoided those peers because she found the need to adapt personally energizing. She recalled, “towards the end of it, I think I started steering away from those people who were just constantly seeing it as a negative thing.” Hannah and Jesse both valued experiences that came out of the need to adapt; Jesse summed this up, reflecting, “it personally taught me flexibility in a way that I don’t think I would have learned otherwise.” Planning for the Future: “It Still Feels Like I’m Starting From Ground Zero” Despite the feelings of extreme lowness, challenges with adjusting and adapting teaching, and anxieties regarding moving into the second year, all the teachers reaffirmed their past beliefs about teaching as a profession. Also adding to my understanding of the final research question regarding intentions to continue teaching, all of the first-year teachers have continued through their second and third years. For some, there was a sense that the pandemic was temporary and that teachers would be able to employ the traditional techniques which they learned in university moving forward if they so choose. Thinking about how this year affected his future, Martin recalled thinking, “it’s a dumb year, just let it go already,” and that he was already thinking of plans for traditional concerts. Jesse also expressed how she was looking forward to getting to make music in-person with students, noting that the format caused her some concerns: “Did I question teaching? Yes. 100%. Did I question teaching music? Never.” For many of the first-year teachers, the effects of the pandemic reinvigorated their passion for teaching. Hannah affirmed, “I discovered that there’s no occupation right now that I want to be in, other than teaching music,” and Arthur asserted, “Yeah, I still fully 100,000% want to continue.” This adamant desire to continue teaching was echoed by nearly all the first-year teachers. Unlike his peers, Jack had come into his first teaching position with reservations regarding the profession; while he noted that the pandemic certainly exacerbated his worries about teaching, his specific reservations predated pandemic-related stressors. At the time of the interviews, Jack expressed that he wanted to take some time away from teaching to discern whether teaching was a vocation he wanted to pursue. Jack intentionally turned down multiple teaching positions and an offer to teach in a summer-long music camp during the 2021 summer and had planned to take time off in the next year, but was offered a position late in the summer and has since continued teaching alongside his peers in a middle school band program. All the first-year teachers expressed that they anticipated their upcoming second year of teaching would feel just like the first year of teaching. Hannah stated, “it still feels like I’m starting from ground zero,” and Arthur commented that “as a new teacher as well, I feel like in some ways, the true test for me, at least, will come when a normal year happens.” There were several concerns about lack of experience from student teaching as all of the first-year teachers had done all or part of their student teaching in a fully distanced learning environment, echoing similar reports about feeling unprepared for teaching from other students who student taught during the pandemic (VanLone et al., 2022). Many of them felt concerned that they had never sequenced a concert cycle from start to finish; Jesse noted, “The fact that we didn’t get to finish student teaching. That was also traumatic . . . So we had never, and to this day, I have not actualized starting with the group, preparing music, and having a concert.” Along with concerns for these new expectations were concerns about classroom management; Linda commented,I think classroom management is something that I wanted to learn a lot during student teaching. But through student teaching, it was all online for me for my secondary . . . So we didn’t really get the in-person students causing disruptions or trying to handle that. Despite their anxieties, and the shared sentiment that the second year would feel like a restart because of how different it would be, the teachers did reflect that they felt the benefit of having at least some experience moving into the second year. Hannah noted that the different schedule from distance learning had given her the space to learn “to be more reflective . . . I feel like I had a lot of time last year to be more reflective” which she hoped to carry into her second year. Arthur reiterated several times that “You can’t beat experience. Nothing can beat experience.” Jesse also commented that she felt she could plan better because she had a general idea of what to expect with students now, saying, “you at least know what’s gonna happen.” Discussion All the first-year teachers found the conditions teaching in the COVID-19 pandemic to be quite different from their prior expectations and university teacher preparation. While this is often true when teachers compare their experiences with inservice teaching and the preservice coursework completed in tertiary education (e.g., Ballantyne & Packer, 2004), teachers noted the disconnect specifically with regard to pandemic conditions. Despite intense feelings of lowness and burnout which many teachers felt during this year (Mecham et al., 2021), these teachers largely viewed the conditions to be temporary and expressed that their overall intentions to continue teaching had not been altered by the pandemic. The teachers found comfort in friends and family who supported them through both personal and professional challenges. They also noted the importance of their peer connections and mentorship through schools and professional organizations; the importance of mentoring echoes past research suggesting the importance of positive mentoring in novice teachers’ success (Schmidt, 2008). As the first-year teachers reflected on their experiences, I noted two main approaches to managing online and hybrid teaching: adapting to unexpected circumstances and surviving until things returned to normal. Adaptability differs from coping with challenges in that coping prioritizes survival, whereas adaptability indicates adjustments which yield positive results (Martin, 2012). Thus, while all of these teachers coped with and survived the challenges of teaching during the pandemic, not all were adaptable in ways that benefited their personal life or teaching practices. Martin and Arthur both focused on adjusting their traditional expectations to be delivered in an online format, but did not fundamentally change their teaching behaviors to meet the challenges presented. Perhaps, this is connected to the comparisons they consistently made to the previous band teachers in their schools who ran what might be considered traditional performance programs. By contrast, Jack and Linda both adapted their delivery and selection of content to focus on those skills best taught online with empathy for students’ situations; demonstrating even larger changes, Hannah and Jesse adapted their teaching approaches substantially and shifted the focus of their classrooms to meet the person-level needs of their students. Those that showed the most apparent adaptive expertise in their teaching also expressed or demonstrated aspects of a growth mindset which is linked to engagement and motivation in teaching (Nalipay et al., 2021). Jack specifically mentioned how he liked that he was forced to think of new ways to teach content and expressed a newfound belief in the importance of having a growth mindset for both himself and his students. Similarly, Hannah referenced excitement about having to adapt her teaching and appreciation for the time to reflect on her teaching. The connection between overall adaptability and a growth mindset suggests cognitive creative adaptability (Orkibi, 2021) in some of the teachers, which led to positive teaching outcomes. Teachers who demonstrated adaptive expertise through the pandemic expressed specific changes they made to their teaching while teachers who continued with traditional techniques referenced concerns about how they were viewed and compared with past teachers based on typical criteria. Implications for Novice Teachers and Teacher Education The first-year teachers’ experiences highlight the importance of personal and professional networks and connections during the first year of teaching. Several of the teachers emphasized the importance of help they received from their in-school teacher mentors, which Schmidt (2008) also indicated was an important contribution to success in novice music teaching. By contrast to his peers with strong mentors, Martin felt disconnected from his school and did not have a strong teacher mentor, which he associated with his feelings of burnout and not knowing what to do. This underscores the importance of having strong mentors who are able to provide consistent support for young teachers. While there are time constraints, perhaps universities could provide postgraduation mentorship for novice teachers as a part of faculty teaching loads, leveraging the relationships forged with students in preservice preparation. Participants also referenced the benefits of informal music teacher mentors and professional organizations to support their continued professional development. The support they felt from their graduation cohort helped them to contextualize their experience and find things to keep them motivated, highlighting how peer-mentoring and peer networks can support novice teachers (Draves, 2012). Moreover, fostering strong relationships with students seems to have helped these early career teachers maintain a positive disposition toward teaching in a highly stressful period in their careers. The experiences of these first-year teachers remind us of the importance of teacher–student relationships and fostering a positive learning climate within music classrooms. Similar to other teachers during the COVID-19 pandemic (de Bruin, 2021), these teachers largely found that rapport developed with their students was an important factor in maintaining classroom management and engagement. This skill is an important part of effective teaching and should be specifically addressed in teacher education to help preservice teachers understand how to develop and maintain appropriate relationships with students. Adaptability is an important aide for adjusting pedagogy based on student needs (Loughland & Alonzo, 2018) and helps yield positive learning outcomes for students (Collie & Martin, 2017), as evidenced by the reported experiences from these first-year teachers. Granziera and colleagues (2019) suggested that the capacity for adaptability can and should be fostered in preservice teacher education. To that end, music teacher preparation programs might explore opportunities for microteaching or scenario-based learning (Granziera et al., 2019), varied types of field placements, and extended experiences with the same students. Extended field experiences can feel more authentic to preservice teachers because they can try different approaches with the same students (Blackwell et al., 2022), which allows them develop their adaptive expertise in a setting which more closely matches inservice teaching. The experiences of these first-year teachers, particularly the experiences of the teachers who felt the direct comparison with prior teachers, indicate a need to develop adaptive expertise in future music teachers within their preservice music teacher education programs. In particular, helping preservice teachers to develop a positive mindset with regard to how they are viewed, rather than making comparisons to other educators, may provide space for students to explore personal style and develop adaptability, thereby helping them to become engaged and capable music teachers. Supplemental Material sj-docx-1-jmt-10.1177_10570837231178891 – Supplemental material for Experiences of First-Year Music Teachers in Hawai‘i During the COVID-19 Pandemic: Challenges, Adaptability, and Implications for Future Music Teaching Click here for additional data file. Supplemental material, sj-docx-1-jmt-10.1177_10570837231178891 for Experiences of First-Year Music Teachers in Hawai‘i During the COVID-19 Pandemic: Challenges, Adaptability, and Implications for Future Music Teaching by Nicholas Matherne in Journal of Music Teacher Education The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iD: Nicholas Matherne https://orcid.org/0000-0001-9674-8790 Supplemental Material: Supplemental material for this article is available online. ==== Refs References Ballantyne J. Packer J. (2004). Effectiveness of preservice music teacher education programs: Perceptions of early-career music teachers. Music Education Research, 6 (3 ), 299–312. 10.1080/1461380042000281749 Blackwell J. Matherne N. Momohara-Ho C. (2022). Preservice music teachers perceptions of teaching and learning popular music. Journal of Music Teacher Education, 31 (3 ), 49–65. 10.1177/10570837211067215 Collie R. J. Martin A. J. (2017). Teachers’ sense of adaptability: Examining links with perceived autonomy support, teachers’ psychological functioning, and students’ numeracy achievement. Learning and Individual Differences, 55 , 29–39. 10.1016/j.lindif.2017.03.003 Conway C. M. (2015). The experiences of first-year music teachers: A literature review. Update: Applications of Research in Music Education, 33 (2 ), 65–72. 10.1177/8755123314547911 Conway C. M. Hibbard S. (2019). Pushing the boundaries from the inside. In Conway C. M. Pellegrino K. Stanley A. M. West C. (Eds.), The Oxford handbook of preservice music teacher education in the United States (pp. 2–22). Oxford University Press. 10.1093/oxfordhb/9780190671402.013.1 Creswell J. W. Creswell J. D. (2017). Research design: Qualitative, quantitative, and mixed methods approaches (5th ed.). Sage. de Bruin L. R . (2021). Instrumental music educators in a COVID landscape: A reassertion of relationality and connection in teaching practice. Frontiers in Psychology, 11 , Article 624717. 10.3389/fpsyg.2020.624717 Diliberti M. Schwartz H. Grant D. (2021). Stress topped the reasons why public school teachers quit, even before COVID-19 (RR-A1121-2). RAND Corporation. https://www.rand.org/pubs/research_reports/RRA1121-2.html Draves T. J. (2012). Second-stage music teachers’ professional identities. Contributions to Music Education, 39 , 31–51. https://www.jstor.org/stable/24127243 Dvir N. Schatz-Oppenheimer O. (2020). Novice teachers in a changing reality. European Journal of Teacher Education, 43 (4 ), 639–656. 10.1080/02619768.2020.1821360 Granziera H. Collie R. J. Martin A. J. (2019). Adaptability: An important capacity to cultivate among pre-service teachers in teacher education programmes. Psychology Teaching Review, 25 (1 ), 60–66. https://files.eric.ed.gov/fulltext/EJ1216443.pdf Hash P. (2021). Remote learning in school bands during the COVID-19 shutdown. Journal of Research in Music Education, 68 (4 ), 381–397. 10.1177/0022429420967008 35769178 Holmes A. G. D. (2020). Researcher positionality: A consideration of its influence and place in qualitative research—A new researcher guide. Shanlax International Journal of Education, 8 (4 ), 1–10. 10.34293/education.v8i4.3232 Joseph D. Lennox L. (2021). Twists, turns and thrills during COVID-19: Music teaching and practice in Australia. Music Education Research, 23 (2 ), 241–255. 10.1080/14613808.2021.1906852 Kaden U. (2020). COVID-19 school closure-related changes to the professional life of a K–12 teacher. Education Sciences, 10 (6 ), 165. 10.3390/educsci10060165 Kelly B. L. Neidorf J. (2021). Teaching artists’ adaptability in group-based music education residencies. Social Work with Groups, 45 , 228–243. 10.1080/01609513.2021.1896165 Loughland T. Alonzo D. (2018). Teacher adaptive practices: Examining links with teacher self-efficacy, perceived autonomy support and teachers’ sense of adaptability. Educational Practice and Theory, 40 (2 ), 55–70. 10.7459/ept/40.2.04 Martin A. J. (2010). Building classroom success eliminating academic fear and failure. Continuum International Publishing Group. Martin A. J. (2012). Adaptability and learning. In Seel N. M. (Ed.), Encyclopedia of the sciences of learning (pp. 90–92). Springer. McLean L. Abry T. Taylor M. Gaias L. (2020). The influence of adverse classroom and school experiences on first year teachers’ mental health and career optimism. Teaching and Teacher Education, 87 , 102956. 10.1016/j.tate.2019.102956 Mecham E. Newell E. J. Reina L. J. Stewart C. (2021). Navigating pandemic schooling for novice teachers. Educational Research: Theory and Practice, 32 (1 ), 90–96. Miksza P. Parkes K. Russell J. A. Bauer W. (2021). The well-being of music educators during the pandemic Spring of 2020. Psychology of Music, 50 , 1152–1168. 10.1177/03057356211042086 Miles M. B. Huberman A. M. Saldana J. (2014). Qualitative data analysis: A methods sourcebook (3rd ed.). Sage. Nalipay M. J. N. King R. B. Mordeno I. G. Chai C.-S. Jong M. S. (2021). Teachers with a growth mindset are motivated and engaged: The relationships among mindsets, motivation, and engagement in teaching. Social Psychology of Education: An International Journal, 24 (6 ), 1663–1684. 10.1007/s11218-021-09661-8 Orkibi H. (2021). Creative adaptability: Conceptual framework, measurement, and outcomes in times of crisis. Frontiers in Psychology, 11 , Article 588172. 10.3389/fpsyg.2020.588172 Parkes K. A. Russell J. A. Bauer W. I. Miksza P. (2021). The well-being and instructional experiences of K-12 music educators: Starting a new school year during a pandemic. Frontiers in Psychology, 12 , Article 2837. 10.3389/fpsyg.2021.701189 Roseth N. E. Blackwell J. (2022). Relationships between well-being and teaching adaptability among music teacher educators: A snapshot of the 2020–2021 academic year. Journal of Music Teacher Education, 32 , 86–100. 10.1177/10570837221120762 Rothe J. Buse J. Uhlmann A. Bluschke A. Roessner V. (2021). Changes in emotions and worries during the Covid-19 pandemic: An online-survey with children and adults with and without mental health conditions. Child and Adolescent Psychiatry and Mental Health, 15 (1 ), 11. 10.1186/s13034-021-00363-9 33612122 Schmidt M. (2008). Mentoring and being mentored: The story of a novice music teacher’s success. Teaching and Teacher Education, 24 (3 ), 635–648. 10.1016/j.tate.2006.11.015 Stringham D. Snell A. (2019). “CONSIDERABLE STRESS and MISERY”: A first-year music teacher’s experiences. Research Studies in Music Education, 41 (1 ), 81–98. 10.1177/1321103X18773100 VanLone J. Pansé-Barone C. Long K. (2022). Teacher preparation and the COVID-19 disruption: Understanding the impact and implications for novice teachers. International. Journal of Educational Research Open, 3 , 100120. 10.1016/j.ijedro.2021.100120 Williamon A. Ginsborg J. Perkins R. Waddell G. (2021). Performing music research: Methods in music education, psychology, and performance science. Oxford University Press.
PMC010xxxxxx/PMC10291209.txt	==== Front POL sppol Politics 0263-3957 1467-9256 SAGE Publications Sage UK: London, England 10.1177/02633957231178526 10.1177_02633957231178526 Special Issue Article Economic and mobility repercussions of the COVID-19 pandemic on the Chile–Bolivia border Liberona Nanette https://orcid.org/0000-0002-4771-3345 Piñones-Rivera Carlos Universidad de Tarapacá, Chile Carlos Piñones-Rivera, Departamento de Ciencias Sociales, Universidad de Tarapacá, Calle 2 #4521, Iquique 1100000, Chile. Email: carlospinonesrivera@gmail.com 23 6 2023 23 6 2023 0263395723117852615 7 2022 25 4 2023 27 4 2023 © The Author(s) 2023 2023 Political Studies Association This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. This article analyzes how the pandemic caused by the coronavirus (COVID-19) has impacted international migration. In particular, we compare the mobility and economic repercussions faced by Bolivian and Venezuelan migrants. We conducted 16 semi-structured interviews with migrants who requested legal and social support and advice provided by the Open Assembly of Migrants and Pro-Migrants of Tarapacá, Chile (AMPRO), an organisation dedicated to defending migrant rights. The Bolivian interviewees worked in Chile before the pandemic in the city of Iquique (close to the Bolivian border). The Venezuelan interviewees are undocumented people in transit who entered Chile during the pandemic. Through this comparison, we describe the economic repercussions on the everyday life, mobility, and survival strategies of people in transit, transboundary workers, and migrants with transnational families, and reveal a realignment of Chile’s border regime that benefits post-pandemic capitalism. Furthermore, we clarify how the health restrictions implemented due to the pandemic have favoured the reconfiguration of the border regime imposed in Chile, through a racist immigration policy based on the control and management of migration, leading to a greater irregularization of migration. border regime irregular transit migrant labour migration pandemic Fondo Nacional de Desarrollo Científico y Tecnológico https://doi.org/10.13039/501100002850 1210602 edited-statecorrected-proof typesetterts1 ==== Body pmcIntroduction For the general population, 2020 was a challenging year due to the pandemic’s multiple impacts on their lives, which have included restrictions on transnational mobility and within cities of residence, along with a global economic crisis that has increased poverty and dependence on informal work (ECLAC, 2020). The pandemic has produced one of the greatest global economic crises ever known, comparable to that of 2008 and the Great Depression of 1929 (Foladori and Delgado, 2020). This situation has become even worse and more profound for the migrant population (Blanco and Cuervo, 2021) because of the consolidation of migratory policies influenced by the global migratory regime, given the international approach in the control and management of migration (Mezzadra, 2005; Pécoud, 2018). In Chile, in particular, 2020 will be remembered as the year the Congress debated and voted for the new Law of Migration, without the participation of the people it concerns (Liberona et al., 2022a, 2022b). At the beginning of the pandemic, President Piñera requested extreme urgency for a parliamentary discussion on the migration bill that he had presented at the beginning of his mandate in 2018. Thus, the parliamentary vote for the bill took place amid a state of exception 1 and was implemented by a decree in response to the pandemic. Moreover, the parliamentary debate was dominated by a government executive, and a repetitive and misleading discourse that constantly alluded to the need for ‘orderly, safe, and regular’ migration, but in truth, was being used only to criminalise any form of irregular migration (Dufraix et al., 2020; Pécoud, 2018; Piñones-Rivera et al., 2022). The year 2021 commenced with the country’s borderlands stained by the deaths of mostly Venezuelan nationals. Between November 2020 and November 2021, 19 people died, including a 9-month-old baby, as they attempted to enter Chile from Bolivia along the border near the village of Colchane in the Chilean highlands. In this settlement, a 4-h drive from Iquique (the capital of the Tarapacá region in northern Chile), Venezuelan migrants, identified as the second-largest group worldwide in terms of forced displacement after Syria (United Nations High Commissioner for Refugees – International Organisation for Migration (UNHCR-IOM), 2021), experienced horrific state neglect and social rejection (Mayorga, 2021). Thus, these borderlands were overwhelmed by a huge migratory flow, consisting of families with children of different ages and older adults. Moreover, the health of many people was seriously affected by the extreme climatic conditions of the Andean highlands, as well as the long journey, often by foot, which they had endured across the continent (Liberona et al., 2022b). In this article, we present the economic and mobility repercussions experienced by two groups of migrants in the context of the pandemic, addressing two different types of mobility. The first is the Bolivian group which has been subject to a long-standing phenomenon of cross-border migration, and who enter and leave the country in search of work and have strong transnational networks in Chile; the second is the Venezuelan group who, instead, enter Chile after being forcibly displaced by a complex humanitarian emergency, 2 and with the hope of becoming long-term migrants. Our research question is as follows: How do these particularities produce differences and similarities in the economic and mobility repercussions? Our goal is to compare the experiences of both groups and show their differences and similarities, according to each group’s specific characteristics and determinations and their relationship with Tarapacá’s territory. Considering that Iquique is a border city and is affected by international border flows in multidimensional aspects (economic, demographic, cultural, etc.) (Fondecyt Project 1210602), we are interested in understanding the importance of the territorial and political context, as well as how both economic and mobility repercussions are defined. Importantly, Tarapacá is a region characterised by social, cultural, economic, and demographic exchanges with neighbouring countries (Guizardi et al., 2015). Moreover, in recent years, the region has witnessed significant migration from other South American and Caribbean countries (INE-DEM, 2021). Although people from both nationalities entered Chile before and after the pandemic, we focused our analysis on the Bolivian migrant population present in Iquique up to 2020, and the Venezuelan population that entered during the pandemic. This is because we wanted to observe the application of migratory and border policies that have been differentiated towards certain groups, following Heyman’s (2012) approach to unequal mobility. This proposal is based on recognising the border as a valve that allows mobility flows to be controlled to establish a subtle balance and thus produce an enormous unauthorised migrant labour force. This approach allowed us to observe differentiated border policies for both groups and their consequences. Based on the inter-actionist approach proposed by Fredrik Barth ([1976] 1969: 18), who states that ‘stable inter-ethnic relationships presuppose a similar structure of interaction’, we believe, on the hand, that there is a type of historical inter-ethnic relationship with neighbouring countries, especially regarding the Bolivian migrant population of indigenous origin, which has influenced the pandemic’s role in restructuring capitalist exploitation. On the other hand, we consider that Chilean immigration policies, under the influence of migratory governance in South America (Domenech, 2017) have focused on controlling Venezuelan migration because of the huge numbers of people involved, taking advantage of the context of sanitary border closures to push the vast majority of this population into a condition of illegality. Both realities highlight how capitalism relies on such restrictions on mobility caused by the pandemic to exploit migrant labour. Thus, this study enhances our knowledge on how capitalism profits by sophisticating the global regime for migration control in South America in the context of the COVID-19 pandemic. Border regimes and the pandemic If the borders, as defined by Heyman (2011), represent a system of differentiated flows, and the approach to these flows is what differentiates them at the time of people crossing borders, the pandemic, and the consequent border closures for health reasons, simply intensified the differentiation of such flows. This has also led to a change in cross-border mobility practices, which can be observed in a comparative analysis as unequal. Heyman states that there is a differentiated form of mobility across borders based on ‘a complex series of social, cultural and economic inequalities’ (Heyman, 2011: 82). This refers to a movement in space that can be considered a form of unequal mobility displayed within contemporary capitalism, the main characteristic of which is the segregation of the poor, from whom economic participation is demanded but who are socially excluded (Heyman, 2012). This is the case for South American and Caribbean migrant populations in Chile, who have been racialised by policies enacted for immigration control (Tijoux and Mandiola, 2014). However, parallelly, they are viewed as useful for the system’s reproduction; the more precarious their lives, the more they can be exploited, which is achieved by forcing them into a condition of irregularization (De Genova, 2002). This case study forms part of the discussion regarding border regimes (Mezzadra and Neilson, 2013) in Latin America, in which migratory dynamics and their control, since 2015, have experienced the so-called punitive turn (Domenech, 2017) with the consolidation of neoliberal policies (Velasco et al., 2021). Following Llavaneras (2022), border regimes refer to the various actors and interactions through which borders operate, with human (in)mobilities being an integral part of these regulatory systems of inclusion and exclusion that determine what and who cross the borders of nation-states, and under what conditions. This concept was extensively discussed in a collective work whose Anglo-Saxon authors propose a ‘radically constructivist approach to border studies’: ‘By speaking of a ‘border regime’ we signal an epistemological, conceptual and methodological shift in the way we think, conceive and research borders’ (Casas-Cortés et al., 2015: 28). The study of border regimes incorporates the agency of migrants as a constitutive element of their configuration, as their practices and bodies are at the centre of conflict and negotiation. The intensifying state violence towards migrants takes place within the context of important changes in migratory patterns. The agency of migrants in these changes is seen in the consolidation of corridors from the Andean−Central America−Mexico region and the formation of the Andean−Southern Cone corridor (Velasco et al., 2021: 22); these trends adapt to the creation of a ‘global regime for the control of migration’, characterised by ‘the emergence of new ways of thinking and acting on migrations, such as migration management’, the purpose of which is the regulation of international migrations within a neoliberal global framework. These policies have focused on preventing, eliminating, and regulating ‘migratory pressures’. One of the culminating moments of this global regime is the Global Compact for Migration (2018), which sets out that migration should be ‘safe, orderly, and regular’. Domenech (2017: 24) states that ‘the idea of’ orderly migration ‘constitutes one of its fundamental components and, consequently, establishes “illegal” or “irregular” migration as a worldwide problem that entails multiple risks, making its “prevention” and “combat” the focus of a specific intervention strategy’. In a highly developed capitalistic context (Foladori and Delgado, 2020), the COVID-19 pandemic has disrupted the capital−work relationship irreversibly. In this case, we see how the conditions of confinement, along with the restriction and control of cross-border mobility, favour the exploitation and the precariousness of migrant labour due to the massive irregularization and consequent deportability they are subjected to, caused by collective expulsions (De Genova, 2002). Furthermore, this is implemented by incorporating elements of global trends absent in other countries of the region, such as the privatisation of deportations and the adoption of humanitarianism to manage forced migration (Casas-Cortés et al., 2015). Despite this, there are similar forms of control in other Latin American countries, as in the case of Mexico’s border with the United States, as studied by Castro (2021), with the sanitised management of migrations, a process that adds another marker of difference to the migrant, as a carrier of the virus, thus becoming part of the pathogenic element themselves. In the following section, we present the working methodology and collaborative ethnography on which this study was based. We then address the mobility of the Bolivian and Venezuelan migrants interviewed, who have experienced restrictions and challenges caused by the sanitary and political closure of the border. Furthermore, we present details of the economic repercussions that the COVID-19 crisis has generated in the everyday life, mobility, and survival strategies of the two groups of migrants mentioned above. Finally, we analyse how these restrictions impacted the reconfiguration of the border regime that accompanied the COVID-19 pandemic, while simultaneously favouring capitalist restructuring. Methodology This investigation has resulted from 6 years of monitoring and collaboration with the organisation Open Assembly of Migrants and Pro-Migrants of Tarapacá, Chile (AMPRO), which is dedicated to defending migrant rights (Piñones-Rivera et al., 2023). Our collaborative efforts included producing documents, studies, and communications, thereby contributing to AMPRO’s work. We emphasise that this participation has not made us the protagonists of AMPRO’s work, but simply some among the organisation’s collaborators. In addition, particularly for this investigation, we conducted 16 semi-structured interviews with migrants who requested legal and social support and advice provided by AMPRO in their office in Iquique and during the organisation’s field trips. All the interviewees were adults whom we subsequently consulted regarding their interest in participating in the investigation. They included three Venezuelan women who recently arrived in Chile and were homeless; a Venezuelan woman who had just entered the labour market; and three Venezuelan men, one who had recently arrived in Chile and was unemployed, while the other two had been in the country for a while and had a more stable level of labour and social integration. The Bolivian nationals were four women and four men with different degrees of social and labour participation, but they had worked in Iquique for several years. We reviewed reports in the Chilean press between March 2020 and August 2021, categorising the news based on issues related to our objectives. This was how we selected news articles about the pandemic, borders, and migrations. The information collated contextualised the investigation and provided references for some of the events described in this article. Mobility repercussions of the pandemic on two groups of migrants in Tarapacá. Between 2017 and 2018, there was a significant increase in Venezuelan migration to Chile, which became the largest immigrant group in the country, representing 23% of all migrants (Stefoni et al., 2019). Until 2019, Venezuelans in Chile stood out as one of the migrant groups with the highest levels of qualifications and best workforce participation, although this did not necessarily translate into better jobs due to the difficulties in obtaining recognition for their qualifications. After the first few migratory waves, other Venezuelans began to arrive, for family reunification and seeking asylum in many cases. In fact, in 2018, the largest number of asylum applications globally was made by Venezuelans (Servicio Jesuita a Migrantes (SJM), 2020). By 2019, the Venezuelan diaspora was firmly consolidated in Chile as the largest immigrant group, representing 30.5% of all resident migrants (INE-DEM, 2021), while in the Tarapacá region in 2020, it represented 6.6% of the total migrants (Instituto Nacional de Estadísticas (INE), 2021). In the same year, 16,748 asylum applications were made at the Chilean border by Venezuelan nationals (SJM, 2020). According to some surveys, this population still has a good level of workforce participation because of their high levels of qualifications and professional skills, although they are mostly included in the informal job market (Centro Nacional de Estudios Migratorios – United Nations High Commissioner for Refugees (CENEM-UNHCR), 2021). In 2021, in the northern zone of Chile, almost 50% of this group was undocumented, while 68.5% indicated having entered the country through unauthorised border crossings (CENEM-UNHCR, 2021). Furthermore, at the beginning of the pandemic, the presence of a significant Bolivian migrant population in the country was evident, with people affected by the border closure. This prevented Bolivians from returning to their country, leaving many in conditions of extreme vulnerability, first in the country’s centre, then in the city of Iquique, and finally at the border. The cross-border migration of Bolivians in the Tarapacá region has a long history (Tapia, 2015), in addition to being an important practice in communities of indigenous origin, particularly the Aymara (Tapia, 2018). This means that the place of such people had already become pre-established in the social, racial, and sexual organisation of work, taking up the most lowly appreciated and precarious jobs, such as domestic, agricultural, and construction work (Fernández, 2015; Leiva and Ross, 2016; Stefoni et al., 2017; Tapia, 2015). However, the 21st century witnessed a significant increase in the number of traders, truck drivers, and service workers (Tapia and Chacón, 2016). In 2020, the Bolivian diaspora was the main group of foreign nationals in Tarapacá, representing 45.7% of the total migrants (INE-DEM, 2021). The historical presence of Bolivians in the region has created a consolidated social fabric, expressed through social and family networks (Guizardi and Garcés, 2013) that have maintained profound and extensive transnational links, aided by circular migration (Leiva and Ross, 2016). Therefore, the entry of Bolivian migrants into Chile since the onset of the pandemic has not been considered newsworthy. However, support networks have played an essential role in the reception of this population. The sanitary closure of the border has had wide-ranging repercussions on the mobility of these two migrant groups, although this does not mean that there has been no mobility. In this case, we can observe an increase in the number of people entering via non-authorised border crossings, accompanied by a rise in migrant smuggling and collective expulsions (Liberona et al., 2022b). This is the first element of the reconfiguration of the border regime. According to the analysis carried out by Mayorga (2021), the sanitary closure of the borders responds to the historical role played by the state to protect public health by limiting international mobility. He thus concludes that these limitations established by states in any country in the world must be based on strict epidemiological criteria; otherwise, they can affect the fundamental rights of people who migrate, which has happened when other questionable reasons are considered. In the case of Chile, Mayorga analysed the link between immigration control and public health, which began with the suspension of the free movement of people within the framework of the state of exception of catastrophe, a decree issued in March 2020. According to the author, international experience has shown that when technical concepts of medical science are applied to restrict human mobility, they tend to merge with other kinds of arguments, such as those pertaining to ‘public morality’, or ‘national security’. This is precisely what has happened in Chile, where restrictive migratory policies have been justified: The response of the State has been to militarise the border with Peru and Bolivia, authorising the Armed Forces to support the control of the illegal traffic of migrants and human trafficking [. . .] but while this has been going on, Santiago’s International Airport has resumed operations to reactivate international tourism. (Mayorga, 2021: 218) In other words, to protect the health of the national population, a sanitised management of migrations is carried out (Castro, 2021), associating migrants with the pathogenic element and with transnational crimes that must be combated. With this introduction, we wish to comment on the repercussions experienced by Bolivian and Venezuelan migrants in terms of their mobility, caused by the changes in migration policies in the context of the COVID-19 pandemic. This allows us to explain how these changes in the border regime have intensified unequal mobility (with reference to Heyman) between the two groups studied. Repercussions on the mobility of Bolivian nationals Border closure has brought repercussions for the Bolivian group in two ways: first, in the impossibility of exercising their lives as cross-border workers, given the obstruction to the migratory circularity system based on the possibility of tourist visas (Tapia and Chacón, 2016), and the regular practice of entering and leaving Chile every 90 days. Second, it has affected transnational families who have not been able to be physically together since the beginning of the pandemic: I had travel plans, but it was no longer possible. Because I had to leave before my visa period ended again. So I just had to stay here (in Chile). (Carmen, Iquique, interviewed in April 2021) In these cases, many Bolivian nationals preferred not to leave Chile via unauthorised border crossings to avoid exposing themselves to migratory controls and sanctions, as the arrival of the armed forces led to increased violence on the border. In other cases, border closure has not prevented continuous cross-border mobility, as commented by Ana, referring to women who work informally in the dressmaking sector during periods of temporary border opening and the restrictions imposed by the Chilean government: And these women come from other countries to work in dressmaking, and they come anyway. I know the case of three women, who just when we entered phase two (of quarantine restrictions by the Chilean government) they came over [. . .], but just as we went back to phase one, they left. They obviously had to expose themselves to using an authorized border crossing, without any permission, they travelled without permission, and the employer does not take any responsibility for that, and they did not have a written employment contract anyway. (Ana, Iquique, interviewed in April 2021) This last account illustrates how cross-border mobility continued between Bolivia and Chile via non-authorised border crossings, which the Chilean media did not report or question. However, this undoubtedly favoured the exploitation of this population segment. Consequently, it can be observed how interaction has been maintained despite the border closure, reconfiguring the border regime, which underlines stable inter-ethnic relations, marked by the existence of a cross-border territory. This can be corroborated by the fact that Iquique’s municipality set up shelters for Bolivians trying to return to their country and were stranded in the city due to border closure. 3 Repercussions on the mobility of Venezuelan nationals The situation of the Venezuelan population has been quite different, as they have become the most rejected national group in the continent. The border closures in response to COVID-19 simply reinforced the administrative measures that had already been implemented to prevent the entry of Venezuelan immigrants into Chile, such as the Democratic Accountability Visa (DAV) and the Consular Tourist Visa (CTV), both of which are very difficult to obtain (Liberona et al., 2022b). Instead of resorting to applying through the Law of Asylum, considering the humanitarian emergency from which they are fleeing, they simply prolonged the humanitarian emergency at the borders of Chile with Peru and Bolivia in 2019. With the sanitary closure of borders, the processing of this visa was suspended, but people continued to migrate to Chile, accompanied by the formation of new migratory patterns. Towards the end of 2020, this dramatic situation worsened as the number of families reaching the border, closed because of the pandemic, began building up in border towns and cities. This led to unrest and panic, as well as an increase in racism within the local indigenous community and in the local population in general, resulting in public demonstrations, 4 reports in the media, 5 and key decisions taken by local and national authorities (including the militarization of the border). Migratory control, carried out with the support of the Armed Forces, employed sanitary control to associate the new migratory patterns with organised crime. In 2021, Venezuelans became the largest population group in the world requesting asylum (IOM). In this context, the Chilean State granted asylum to only five Venezuelan nationals. In 2020, only seven nationals from this country were recognised as refugees, after processing 1629 applications for asylum (SJM, 2021). In 2021, the processing of asylum applications dropped dramatically due to a policy of disincentives, along with barriers to making such applications, as revealed by the Comptroller’s Report on freezing this administrative process by the country’s border authorities for the year in question. Through Oficio (official instruction) 7.196 issued in 2021, the order was given to certain provincial governments to suspend the reception of asylum applications, in accordance with Law 20,430. Furthermore, it has been confirmed that ‘State agents have hindered the possibility to apply for international protection and to claim the right to asylum’ (Pascual, 2020: 408). This same situation has been documented at the U.S.−Mexico border, revealing a similar reconfiguration of border regimes at the regional level (Paris, 2022). In February of the same year, the first of a series of collective expulsions occurred, when 138 foreign nationals were expelled from Chile, 100 of whom were Venezuelans. The people concerned were then staying at a quarantine health residence 6 in Iquique and were notified of their expulsion less than a day in advance. The expulsion was carried out without the legal guarantee of the right to due process and a legal defence and was therefore completely illegal (Liberona et al., 2022a; Piñones-Rivera et al., 2022). Other similar operations were carried out following this event, ignoring the rights of this population, violating the protection of their health – recognised as an inherent right of migrants within the framework of international migration law – and violating the prohibition on collective expulsions. Although, in this respect, the case of the collective expulsions of Cubans in Ecuador represents a precedent in South America (Álvarez, 2020), how these operations were carried out indicates a reconfiguration of the border regime that is unique in the context of the pandemic. In particular, there is evidence of an alignment with global trends for the control of irregular migration which, according to López-Sala and Godenau (2017), include coordinated management between States; the use of technologies for monitoring, surveillance and data management; the strengthening of internal control mechanisms; and the participation of various types of public and private actors. A specific example was the hiring of the private company Sky Airlines 7 to carry out the expulsions, involving private sector actors in a form of outsourcing of state responsibilities, as is the case with border externalisation (Casas-Cortés et al., 2015). Furthermore, expulsion/deportation policies have the functionality of producing deportability (De Genova, 2002), that is, the generation of a disciplined and docile labour force (Álvarez, 2020). Thus, the deportation of a limited group of migrants could have the effect that precarious migrant labour increasingly benefits post-pandemic capitalism. Instead of applying the law of asylum to a population that certainly met the requirements, the creation of the respective DAV and CVT visas for specific national groups represented bureaucratic strategies for the deprivation of rights. Likewise, the implementation of questionable acts of mass expulsions during a pandemic reveals a reconfiguration of Chile’s border regime, based on the production of deportability. This reconfiguration was supported by the ‘Colchane Plan’, the ‘Protected Borders Plan’, and numerous administrative actions of the State, such as an official instruction that prevents the processing of asylum applications. Although these measures had repercussions for Bolivian nationals, as obstacles to historical mobility, their knowledge of the cross-border territory and the transnational networks of this population permitted them to continue their mobility with certain limitations. However, for the Venezuelan population, the repercussions of mobility represented significant denials of their rights, which seriously exacerbated their vulnerabilities. 2. Economic repercussions of the pandemic on two groups of migrants in Tarapacá In this section, we compare the economic repercussions faced by Bolivian migrants who had been working in Chile before the pandemic, with those of Venezuelan nationals, especially those in a state of irregular transit, and those who had entered Chile in search of asylum. The economic conditions we present here focus on their employment situation, the labour or economic sectors in which they are engaged, and the relevant experiences indicated in the interviews. Economic repercussions experienced by Bolivian nationals For Bolivian nationals residing in Iquique, the problem that most affected them during the pandemic was related to mass dismissals, causing them consequent difficulty in paying their rent for housing. Moreover, the slowness of administrative processing in the Department of Immigration, for issuing or renewing visas and identity documents, had impacts on other areas of their lives (Liberona et al., 2022a). The demand for an identity card was maintained as a requirement for different procedures and benefits, forcing people into exploitative jobs if they did not have this document. In addition, the health control measures imposed by the government were described by one interviewee as ‘indeterminate and destabilizing’, which directly affected access to work because such measures were constantly changing. Furthermore, when we asked people in the interviews about their economic situation, the answer was clear: ‘It’s worsened a lot, now I only earn enough for food, no more than that’ (Elda, Iquique, interviewed in May 2021). It was clear that people’s income had fallen, thus restricting the possibility of sending remittances back home: Some of my girlfriends are in the same boat, others are unemployed, others only have a half salary that doesn’t cover their needs, because they pay rent and apart from that have to send money back to Bolivia, so it’s just not enough . . . (Natalia, Iquique, interviewed in May 2021) When analysing these repercussions in terms of people’s contractual situation, it was clear that the loss of employment had been the most destabilising factor, as well as having to re-enter the informal sector, which many people had been able to abandon once they had managed to normalise their immigration status. In comparison, at the national level, the informal occupation rate of the migrant population in Chile rose to 28.9% in 2021, representing a year-on-year growth of 6.4 percentage points. The number of informally employed migrants grew by 52.4% during this period (SNM-INE, 2021). 8 However, informal work only enables day-to-day survival. For example, in the hotel and food sector, which depended on the lifting of restrictions to open up to the public, there was no advanced warning of such conditions. This meant that staff were not being hired, creating more harsh conditions of exploitation and uncertainty. For self-employed migrant workers, the situations have been diverse, and in some cases, quite complex, as seen from accounts of two interviewees (Wilfredo and Elda). They told us that before the pandemic, they had a hairdressing business, but they had to close their salon because of restrictions and military control. However, they had to continue paying rent for the salon while working from home: I had to run a clandestine hairdressing salon to be able to work [. . .] to keep up on the rent of the house and the business premises, and to buy food. Thankfully, we only have one child; otherwise, it would have been much more difficult. (Wilfredo, Iquique, interviewed in May 2021) Fernando has an agri-food sales business that allowed him to maintain his economic situation during the pandemic, and even to open another similar businesses, although he pointed out that the job only provides him with daily sustenance: ‘if one day you don’t work, it’s a day you don’t eat’ (Fernando, Iquique, interviewed in May 2021). In some labour sectors, it was practically only the migrant community that continued to work: ‘I’m a truck driver, so I’ve been working; in other words, the pandemic opened up more opportunities for me to work given the sector I’m in’. (Roly, Iquique, interviewed in May 2021). This was also the case for Antenor, who claimed to have his own business, all legalised – he confirmed – which has been doing well during the pandemic, mainly due to the increase in delivery work in the sale of prepared meals, added to which was the 10% withdrawal permitted from people’s pension funds, a national policy to reactivate the economy. Observing the situation for contractual labour insertion, many of the jobs occupied by this population segment were provided by small businesses, which stopped hiring staff, especially in the service sector. This was because public service was suspended as a health protection measure. Thus, many Bolivian nationals preferred to return to their country instead of attempting to survive under such precarious conditions. 9 In the agricultural sector, various irregularities could be identified, such as the falsification of contracts to obtain interregional travel permits. Generally, in this sector, contracts consist of verbal agreements, but these types of irregularities have an economic impact on workers, who also have to pay expenses and deal with the associated risks. Another employment sector that is clearly precarious is dressmaking, with the women in this sector being able to work only sporadically, when the sanitary control system established by the Chilean government in response to the pandemic allows a temporary opening for domestic mobility. That is, when it passed from phase 1 (equivalent to total confinement) to phase 2, mobility was allowed during working weekdays. In the latter case, Bolivian women entered the country to work, but on the resumption of phase 1, they had to return home to Bolivia. In addition, it was also pointed out that the payment was per day worked or garment manufactured, in which case, they received only one third of the product price. In the area of domestic work, there has been a greater level of exploitation owing to confinement. Working days off has been a normal practice, thus preventing other activities and even complementary jobs, as Carmen shared. In her case, she could not take her days off and leave the house where she worked, and had to work those days without receiving any extra payment. To address the economic crisis caused by the COVID-19 pandemic, the Chilean government began to provide support through food boxes and emergency benefit payments, in addition to allowing people to withdraw 10% of their pension funds. All those contributing to such funds were entitled to these withdrawals on three occasions, irrespective of their nationality. However, much of the migrant population did not receive this support, especially those whose residency in Chile was still being processed by the authorities. We also verified that access to state aid to deal with the economic crisis was differentiated according to immigration status and duration of permanence in the country, although the employment areas in which people worked had a significant influence on whether they received such support. For those who had a long-term residence status in Chile, it was confirmed that they received both the food boxes and benefit payments provided by the government, and it was possible for them to withdraw 10% of the pension funds. In some cases, they were even eligible to receive the Emergency Family Income. A permanent status in the country of more than 5 years in Fernando’s case, 8 years in Antenor’s case, and 20 years in Role’s case enabled them to enjoy a certain level of job stability, even though they were self-employed workers whose income depended on their daily activities. In the case of the four women interviewed, it was confirmed that they had limited access to state aid given their circular cross-border way of life, which left them in a situation of irregularity (with the Chilean authorities). Friendships and solidarity between peers has been their main survival strategy when ‘you don’t have enough brass’ (Janeth, Iquique, interviewed in 2021). When asked about receiving financial aid, the answer was quite blunt: Nothing, everything’s come straight out of my pocket, so we were just supporting each other, us Bolivian women, and among my three or four friends. I spent about four months without a job, with nothing, and during that time I was staying in a friend’s apartment. There were three of us, and she gave us so much support during those three or four months when I was without a job. (Natalia, Iquique, interviewed in May 2021) We noted that the economic repercussions experienced by the Bolivian people interviewed have been significant, and due, to a large extent, to the position they occupy in the social hierarchy historically imposed on such people. Likewise, it is due to the impossibility of maintaining a migratory circulatory system between Chile and Bolivia, which keeps them trapped in a state of irregularity. Furthermore, we can confirm that there is a restructuring of the capitalist economy through exploitation, the hyper-flexibility of labour, and a trend to use migrant labour resources for jobs where people are over-exposed to the spread of the coronavirus. The situation of the women interviewed is particularly concerning, especially when considering information published in the Migrant Employment Bulletin, which indicates that in the August−October 2021 3-month period, the unemployment rate of migrant women at the national level was 8.4%, whereas that of men was 6.1% (INE-SNM 2021). This underlines the differences in economic repercussions due to gender inequality in Chile, which has been previously documented (Liberona et al., 2022a). Economic repercussions experienced by Venezuelan nationals The situation is much worse for Venezuelans who entered Chile during the pandemic using unauthorised border-crossing routes, many of whom then found themselves on the streets and homeless, and with the type of job that this situation allows, such as guarding and washing parked cars. This was the case for Haydée, who studied journalism and completed her graduation in broadcasting studies in Venezuela. Women were particularly affected: It’s really difficult because we are homeless, we are not eating well, cannot cover our basic needs, and as women, we have many needs. (Angélica, Iquique, interviewed in May 2021) Moreover, many also have to care for their young children, and when working as street sellers, they must take them along, as they have no one to look after them. They are then criticised for exposing their children to street work, although such critics seldom consider that leaving their offspring alone or in improvised squatter camps in squares and beaches is not an option because of the dangers to which the children are then exposed. Nevertheless, street work is also very difficult for Venezuelan men who arrived recently in Chile, such as for Carlos, especially to make enough money to send remittances back home. Carlos’s main concern is that there are days when he has no money to provide anything to his children, apart from paying for some form of accommodation: ‘After paying for the hostel, I often have nothing left . . . and that makes me so sad’ (Carlos, Iquique, interviewed in April 2021). In addition to these types of jobs that they can access fairly easily, these individuals are still looking for other options, offering their labour for whatever might be needed. Few homeless people are able to escape this situation, however, as not many employers agree to hire them, even informally. Thus, the economic repercussions of the pandemic for these people are quite profound, particularly as they are combined with the Chilean government’s immigration policies, which have made it impossible to normalise their migratory situation after entering the country through unauthorised border crossings, leaving most of them relegated to a state of social indigence. Once again, only social solidarity has allowed such people to survive on a daily basis, with interviewees telling us how they received support from certain organisations and people, mainly in the form of food. This sets in motion the externalisation of the control and management of adversities through humanitarianism, representing another specific element of the Chilean border regime’s reconfiguration that followed the pandemic. Despite the fact that precariousness is constant, we observed specific differences in how the pandemic affected each national group economically. First, we noticed that each group had its own characteristics, which were distinguished by their link with the adopted territory and by their length of stay in Chile. In the case of the Bolivians, we can highlight the long-standing transnational networks fostered by circular migration, which favours both permanence in the country and their return to Bolivia. Similarly, the cross-border mobility of the territory facilitates access to certain employment sectors, such as the transport and dressmaking industries. Length of stay in Chile is another differentiating factor, given that those in the country for several years (e.g. the three Bolivian men interviewed) reported a certain level of job stability and being eligible to receive emergency state aid. Likewise, gender was identified as a differentiating factor, with women of both nationalities being more affected by the type of sector in which they work (such as domestic services), and by the needs and gender mandates imposed by a patriarchal society. However, the Venezuelans who entered Chile during the pandemic by non-authorised border crossings have been the most economically affected, living on the street, receiving humanitarian support, and with few opportunities for workforce participation. Through forced migration, a characteristic of contemporary capitalism (Foladori and Delgado, 2020), such nationals have to deal with unequal mobility, exclusion, and segregation, while at the same time, they are part of the reconfiguration of the border regime by crossing borders despite their closure, in order to become resident migrants. Conclusion The main economic repercussions that the pandemic brought to the migrant population and those that we identified based on our interviews with Bolivian migrants in Tarapacá were the increase in unemployment, informal work, and labour exploitation. In both the groups interviewed, we also identified issues such as greater job insecurity and the risks associated with labour informality and working clandestinely in certain job sectors, including the irregular crossing of borders to carry out their work, or the violation of restrictions on internal mobility. The intensification of unequal mobility, for its part, generated greater economic repercussions on undocumented Venezuelans due to their forced displacement during the pandemic. Among the main repercussions of the pandemic on mobility, a forced immobility was identified in Bolivian nationals that had, in the first place, an impact on transnational economies and families, due to the impossibility of generating remittances and travelling for reasons of family reunification. Moreover, it led to workers’ irregular border crossings and restricted circular migration, as they were prevented from obtaining tourist visas. In addition, the deployment of clandestine border crossings was identified in those periods when internal mobility was temporarily permitted, thus assuming greater risks of being exposed to migratory and health control, while also demonstrating the stability of inter-ethnic relations and the constitutive role of migrants in border regimes. Thus, the main contribution of this study lies in its analysis of how economic precariousness is closely linked to migration irregularization and border securitization. Furthermore, the implementation of a racist immigration policy exacerbated the vulnerability of Venezuelan migrants, while also generating the reconfiguration of the border regime, mainly in terms of a series of factors: the denial of international protection to a population in a condition of forced displacement; the installation of humanitarianism; the militarization of the borderlands; the use of quarantine health residences to implement collective expulsions; and the use of private sector actors to implement these expulsions. The latter reflects a key element in understanding capitalist restructuring, through the privatisation of border control as it intensifies the neoliberal model. This immigration policy, which focused on the control of migrations and their governance, has been reflected in the promulgation of the new Law of Migration (Law 21.325 of 2021) and in the non-application of the Law of Asylum (Law 20.430 of 2010). Chile’s Law of Migration is based on mobility control. However, the cross-border mobility of Bolivian nationals during the pandemic brings this capacity into question. Furthermore, the policy since 2019 of refusing access to requests for asylum at the border by Venezuelan migrants has failed to recognise that this population is in a state of internationally recognised forced displacement. Despite this refusal, Venezuelans continue to cross this border. Therefore, the border, more than ever before, represents the scenario of the legal irregularization of certain migrant populations (De Genova, 2002). This comparative case study allows us to demonstrate that the economic and mobility repercussions produced by the pandemic are different according to the type of mobility. Thus, Bolivians’ long-standing cross-border mobility and migratory circularity is favoured by the historical link with the territory of Tarapacá. However, the forced mobility of Venezuelans is criminalised, producing a greater state of deportability, clearly playing a role in restructuring post-pandemic capitalism, but above all, exacerbating their state of vulnerability. Author biographies Nanette Liberona Concha has a degree in Ethnology from the University Paris 8 and a PhD in Anthropology and Sociology from the University Paris 7. She is an Assistant Professor of the Department of Anthropology and faculty of the doctoral programme in Social Sciences at Universidad de Tarapacá-Chile. She is currently a researcher in charge of the regular FONDECYT project N° 1210602, ‘Refuge in Chile and transit density: production of corporealities and impact on the health of bodies in mobility’. Her lines of research are migration, borders, racism, corporeality, migrant health and irregular cross-border transit. Carlos Piñones-Rivera has a PhD in Anthropology from the Universitat Rovira i Virgili. He is currently an Assistant Professor of the Department of Social Sciences and faculty of doctoral programmes in Psychology and Social Sciences at Universidad de Tarapacá-Chile. He is a researcher in charge of FONDECYT N° 11230358 ‘SARNAQAÑA’, which addresses mobility, economy and rituality of Andean medical knowledge. He has been a guest editor of international journals such as Health and Human Rights and Global Public Health and visiting scholar at the Berkeley Centre for Social Medicine. Lines of research: health of indigenous peoples, intercultural health, migrant health. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article has been funded by ANID through the project FONDECYT 1210602 ‘Refugio en Chile y densidad del tránsito. Producción de corporalidades e impacto en la salud de los cuerpos en movilidad’. ORCID iD: Carlos Piñones-Rivera https://orcid.org/0000-0002-4771-3345 1. Through Supreme Decree 104 of March 18, 2020, issued by the Ministry of the Interior and Public Security, the President of the Republic declared a state of constitutional exception of catastrophe throughout the national territory, due to a public disaster and for a period of 90 days, a measure that was then extended until October 1, 2021. 2. https://www.hrw.org/es/world-report/2023/country-chapters/venezuela 3. https://www.radiopaulina.cl/2020/04/20/450-bolivianos-albergados-en-iquique-vuelven-a-su-pais/ 4. https://iquiquevision.cl/2020/11/14/movilizacion-en-huara-por-migrantes/ 5. https://www.colegiodeperiodistas.cl/2022/01/declaracion-publica-ante-la-violencia.html 6. https://iquiquetv.cl/2020/10/08/autoridades-habilitaran-exestadio-cavancha-como-residencia-sanitaria-transitoria/ 7. https://www.latercera.com/la-tercera-pm/noticia/1500-millones-15-vuelos-y-180-pasajeros-el-contrato-del-ministerio-del-interior-con-sky-airlines-para-expulsar 8. Chilean government National Migrations Service 9. https://www.chvnoticias.cl/nacional/bolivianos-peruanos-acampan-fuera-consulados_20200712/ ==== Refs References Álvarez A (2020) Deportación, tránsito y refugio. El caso de los Cubanos de El Arbolito en Ecuador. Periplos, Revista de Investigación sobre Migraciones 3 (2 ): 52–88. Barth F ([1976] 1969) Los Grupos Étnicos y sus Fronteras. México City, México: Fondo de Cultura Económica. Blanco ML Cuervo MG (2021) The pandemic as a portal: Policy transformations disputing the new normal. DAWN Discussion Paper no. 32, February. Suva, Fiji: Dawn. Casas-Cortés M Cobarrubias S De Genova N , et al . (2015) New keywords: Migration and borders. Cultural Studies 29 (1 ): 1–33. Castro Neira Y (2021) “Viejas y nuevas formas de control migratorio en tiempos de COVID-19. El caso de la frontera México-EE.UU” PERIPLOS. Revista de Investigación sobre Migraciones 5 (1 ): 28–52. Centro Nacional de Estudios Migratorios – United Nations High Commissioner for Refugees (CENEM-UNHCR) (2021) Caracterización sociolaboral de la población refugiada y migrante venezolana en Chile. Resultados por zonas de Chile. Centro Nacional de Estudios Migratorios (CENEM) Universidad de Talca y la Agencia de la ONU Para Los Refugiados (ACNUR). Available at: http://www.cenem.utalca.cl/docs/pdf/Estudio%20Caracterizacion%20Sociolaboral%20de%20la%20Poblacion%20Refugiada%20y%20Migrante%20Venezolana%20por%20Zonas%20de%20Chile%20con%20la%20colaboracion%20de%20ACNUR.pdf (accessed 13 December 2021). De Genova N (2002) Migrant “illegality” and deportability in everyday life. Annual Review of Anthropology 31 (1 ): 419–447. Domenech E (2017) Las políticas de migración en Sudamérica: Elementos para el análisis crítico del control migratorio y fronterizo. Terceiro Milênio: Revista Crítica de Sociologia e Política 8 (1 ): 19–48. Dufraix R Ramos R Quinteros D (2020) ‘Ordenar la casa’: Securitización y producción de irregularidad en el norte de Chile. Sociologias 22 (55 ): 172–196. ECLAC (2020) Los efectos del COVID 19: Una oportunidad para reafirmar la centralidad de los derechos humanos de las personas migrantes en el desarrollo sostenible. Fernández P (2015) La otredad incivilizada en el mundo del salitre: El caso de indígenas bolivianos e inmigrantes asiáticos en Tarapacá, 1900–1910. Polis 14 (42 ): 79–96. Foladori G Delgado R (2020) Para comprender el impacto disruptivo de la COVID-19, un análisis desde la crítica de la economía política. Migración Y Desarrollo 18 (34 ): 161–178. Guizardi M Garcés A (2013) Circuitos migrantes: Itinerarios y formación de redes migratorias entre Perú, Bolivia, Chile y Argentina en el norte grande chileno. Papeles De Población 19 (78 ): 65–110. Guizardi M Penna C Vicuña J , et al . (2015) Claves conceptuales e históricas para comprender la frontera norte de Chile y la migración en Arica y Parinacota. In: Vicuña J Rojas T (eds) Migración en Arica y Parinacota: Panoramas y Tendencias de una Región Fronteriza. Antofagasta, Chile: SJM, pp.19–36. Heyman J (2011) Cuatro temas en los estudios de la frontera contemporánea. In: Ribas N (ed.) El Río Bravo Mediterráneo: Las Regiones Fronterizas en la Época de la Globalización. New York: Bellaterra, pp.81–97. Heyman J (2012) Capitalismo, movilidad desigual y la gobernanza de la frontera México-Estados Unidos. In: Aquino A Varela A Décasse F (eds) Desafiando Fronteras. Control de la Movilidad y Experiencias Migratorias en el Contexto Capitalista. Oaxaca: Sur Plus, pp.25–40. Instituto Nacional de Estadísticas (INE) (2021) Empleo población extranjera n◦1 boletín estadístico del instituto Nacional de Estadísticas y del Servicio Nacional de Migraciones (Chile). Available at: https://www.ine.gob.cl/docs/default-source/ocupacion-y-desocupacion/boletines/2021/poblaci%C3%B3n-extranjera/boletin-empleo-poblacion-extranjera-trimestre-movil-agosto-2021—septiembre-2021—octubre-2021.pdf (accessed 3 May 2023). Llavaneras M (2022) The intimate-mobility entanglement: Subaltern trajectories in the Haitian-Dominican borderlands. Environment and Planning C: Politics and Space 40 (5 ): 1032–1047. 10.1177/2399654420930727 Leiva S Ross C (2016) Migración circular y trabajo de cuidado: Fragmentación de trayectorias laborales de migrantes bolivianas en Tarapacá. Psicoperspectivas. Individuo y Sociedad 15 (3 ): 101–111. Liberona N Stefoni C Salinas S (2022a) Chile. Collective care to confront the pandemic and the criminalisation of migration. DAWN Discussion Paper no. 39, February. Suva, Fiji: Dawn. Liberona N Romero Quezada M Salinas SG , et al . (2022b) Tráfico de migrantes en las fronteras del norte de Chile: Irregularización migratoria y sus resistencias. Derecho PUCP 9 (89 ): 36. López-Sala A Godenau D (2017) Estados de Contención, Estados de Detención. El Control de la Inmigración Irregular en España. Barcelona: Anthropos. Mayorga R (2021) Control migratorio y salud pública: Restricciones a la movilidad humana por razones médicas o sanitarias en tiempos del Covid-19. Revista de Derecho 34 (2 ): 203–223. Mezzadra S (2005) Derecho de Fuga. Migraciones, Ciudadanía y Globalización. Madrid: Traficantes de sueños; Buenos Aires: Tinta limón. Mezzadra S Neilson B (2013) Border as Method or the Multiplication of Labor. Durham, NC; London: Duke University Press. Pascual TI (2020) La [des]protección de los derechos humanos en contextos de movilidad humana en Chile: Expulsiones administrativas y solicitudes de protección internacional. Anuario de Derechos Humanos 16 (2 ): 381. Pécoud A (2018) Una nueva «gobernanza» de la migración? Lo que dicen las organizaciones internacionales. Migración y Desarrollo 16 (30 ): 3143. Piñones-Rivera C Liberona N Arancibia R , et al . (2022) Indigenous border migrants and (im) mobility policies in Chile in times of COVID-19. International Journal of Environmental Research and Public Health 19 (15 ): 9728.35955083 Piñones-Rivera C Liberona N Jiménez V , et al . (2023) Beyond the classroom: The development of collective structural competency in pro-migrant activism. Global Public Health 18 (1 ): 2203732.37078756 Paris Pombo M (2022) Régime de frontières et politiques migratoires dans le Nord du Mexique (2018–2020). Amérique Latine 1 (1 ): 31–60. Available at: https://www.cairn.info/revue-amerique-latine-2022-1-page-31.htm (accessed 3 May 2023). Servicio Jesuita a Migrantes (SJM) (2020) Migración en Chile. Anuario 2019, un análisis multisectorial. Santiago, Chile. Available at: https://www.migracionenchile.cl/publicaciones (accessed 3 May 2023). Servicio Jesuita a Migrantes (SJM) (2021) Migración en Chile. Anuario 2020. Medidas Migratorias, vulnerabilidad y oportunidades en un año de pandemia (N°2). Santiago, Chile. Available at: https://www.migracionenchile.cl/publicaciones (accessed 3 May 2023). Stefoni C Leiva S Bonhomme M (2017) Migración internacional y precariedad laboral. El caso de la industria de la construcción en Chile. REMHU: Revista Interdisciplinar da Mobilidade Humana 25 (49 ): 95–112. Stefoni C Silva C Brito S (2019) Migración venezolana en Chile. La (des)esperanza de los jóvenes. In: Gandini L Lozano F Prieto V (eds) Crisis y Migración de Población Venezolana. Entre la Desprotección y la Seguridad Jurídica en Latinoamérica. México City, México: Universidad Nacional Autónoma de México, pp. 259–284. Tapia M (2015) Frontera, movilidad y circulación reciente de peruanos y bolivianos en el norte de Chile. Estudios Atacameños 1 (50 ): 195–213. Tapia M (2018) Prácticas sociales fronterizas entre Chile y Bolivia, movilidad, circulación y migración. Siglos XX Y XXI. Intus – Legere Historia 12 (1 ): 6686. Tapia M Chacón F (2016) Vínculos transfronterizos: Vida, movilidad y comercio en el barrio boliviano de Iquique, Chile. REMHU: Revista Interdisciplinar da Mobilidade Humana 24 (47 ): 131–152. Tijoux ME Mandiola SP (2015) Aproximaciones teóricas para el estudio de procesos de racialización y sexualización en los fenómenos migratorios de Chile. Polis. Revista Latinoamericana 14 (42 ): 247–275. United Nations High Commissioner for Refugees – International Organization for Migration (UNHCR-IOM) (2021) Plan de respuesta regional para refugiados y migrantes de venezuela (enero-diciembre de 2021), plataforma de coordinación interagencial para la situación de venezuela. Available at: https://www.r4v.info/sites/default/files/2021-06/R4V%20Plan%20Regional%20de%20Respuesta%20para%20Refugiados%20y%20Migrantes.pdf (accessed 3 May 2023). Velasco AS Pedone CG Miranda B (2021) Movilidades, control y disputa espacial: La formación y transformación de corredores migratorios en las Américas. Périplos 5 (1 ): 4–27.
PMC010xxxxxx/PMC10291210.txt	==== Front Am J Health Promot Am J Health Promot spahp AHP American Journal of Health Promotion 0890-1171 2168-6602 SAGE Publications Sage CA: Los Angeles, CA 37349879 10.1177_08901171231185765 10.1177/08901171231185765 Revised Submission from Allen Track Exploring the Impact of the COVID-19 Pandemic on Employees’ Workplace Health Promotion Preferences https://orcid.org/0000-0002-5933-1360 Hammerback Kristen MS 1 Strait Michelle MSW, MPH 1 Kohn Marlana J. MPH 1 Garcia Cinthya MPH 1 https://orcid.org/0000-0001-8728-7195 Harris Jeffrey R. MD, MBA, MPH 1 Hannon Peggy A. PhD, MPH 1 1 Health Promotion Research Center, Department of Health Systems and Population Health, 7284 University of Washington , Seattle, WA, USA Kristen Hammerback, Health Promotion Research Center, Department of Health Systems and Population Health, University of Washington, 3980 15th Avenue NE, Seattle, WA 98195, USA. Email: khammerb@uw.edu 22 6 2023 22 6 2023 08901171231185765© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Purpose The COVID-19 pandemic has led to profound changes in the workplace as well as increases in stress, missed preventive care, and other health concerns. There is limited research since the onset of the pandemic on employees’ primary health concerns and their willingness to engage with workplace health promotion (WHP) programs to address these needs. We conducted this survey about employees’ current health priorities as a first step to exploring whether WHP programs need to evolve to be responsive to employees’ needs at this stage of the pandemic. Design National cross-sectional survey. Setting United States, April 29-May 5, 2022. Subjects 2053 Americans employed part or full time. Measures 17-item online survey assessing demographics, health priorities, and impact of the pandemic on health. Analysis Descriptive statistics, SPSS Version 19. Results Employees’ most common health concerns included work/life balance and stress (each cited by 55%). Nearly half (46%) said their health or well-being was affected by the pandemic; within this group, the most common concerns were stress (66%), anxiety (61%), sleep (49%), and depression (48%). Almost all (94%) indicated they would be open to receiving support from their employers. Conclusion This research is a first step in learning about employees’ current health priorities and how they may have changed. WHP researchers and practitioners can determine how their programs align with current priorities. Our future research will explore employees’ preferences, heath behaviors, and their current workplace environments in more depth. workplace health promotion worksite mental health weight management interventions COVID-19 Division of Cancer Prevention, National Cancer Institute https://doi.org/10.13039/100007316 5R01CA160217 edited-statecorrected-proof typesetterts10 ==== Body pmcPurpose There are 164 million employed Americans. 1 The workplace is a uniquely effective place to reach adults “where they are” with evidence-based health promotion interventions. Concerns about health, and mental health in particular, have increased among employees since the onset of the COVID-19 pandemic and its consequences for society (hereafter “pandemic”).2-4 Employees earning lower than average wages were particularly susceptible to declines in both physical and mental health during the pandemic.5,6 While prior research suggests that employees prioritize healthy eating and physical activity in workplace health promotion (WHP),7,8 we were unable to find any studies that address employees’ health concerns since the pandemic began affecting workplaces in March 2020, including what role WHP might play in addressing them. As little is currently known, this study is intended as a first-look at whether and how preferences may have shifted, and whether there were meaningful differences based on income or other demographic variables. This applied research brief explores employees’ current health priorities, how the pandemic has impacted those priorities, and how open they are to receiving WHP that addresses their priorities. Future research will be needed to go deeper into how and why shifts have occurred and how best to meet changing needs. Methods The University of Washington Institutional Review Board determined that this study was exempt from formal review. Design A national cross-sectional survey was fielded online by Drive Research Incorporated, a market research company based in Syracuse, New York. Sample The survey was fielded with a sample pool balanced on the outgo (ie, respondents sent survey invitations) on six demographic variables (age, race, gender identity, household income, education, region) to reflect the U.S. population. Sample was drawn from Drive Research’s panel of over 15 million Americans who agree to periodically take surveys in exchange for incentives. Respondents were eligible to participate in the survey if they were employed full or part-time and worked for an organization that employed at least 10 people. Measures The 17-item survey covered respondents’ age, current employment status, type and size of the industries for which they work, previous and/or current participation in a workplace wellness program, types of wellness program offerings they have participated in, current concerns related to physical and mental health, the impact of the pandemic on their physical and mental health, and interest in WHP to address their pandemic-related concerns. All questions were multiple choice, and the survey took less than 10 minutes to complete. Respondents’ state of residence, gender identity, race, level of education, and annual household income were collected via Drive Research’s panel pre-screening. Analysis Descriptive statistics were calculated using SPSS Version 19. We analyzed data by total respondents and compared respondents with annual household incomes of <$75,000 and $75,000+ (categories chosen because many state and federal poverty levels are set at $55,500 or slightly higher for a family of four) and by age. We also analyzed data for a subset of respondents who reported their health was directly impacted by the pandemic. Results The survey was completed by 2053 employed Americans. An abbreviated set of demographics are listed below. Full demographics available upon request of author: Age 46% 18-41 54% 42+ 21% 58+ Gender Identification 53% Female 47% Male < .5% non-binary or other identification Race* 4% Asian 13% Black 10% LatinX <1% Native Hawaiian or Other Pacific Islander 76% White 1% Other Racial Identity * Percentages for Race >100% because respondents could select more than one category. Household Income 45% <$75k 55% $75k+ When asked about health topics they consider to be most important to them, employees were most likely to cite work/life balance and stress (each cited by 55%), followed by increasing physical activity (45%), better nutrition (44%), improving sleep (43%), and reducing anxiety (43%). Topics were similar across household income and age. Among the subgroup of employees (46%) who reported that their health and well-being were directly affected by the pandemic, the two most cited concerns were stress (66%) and anxiety (61%), followed by sleep (49%) and depression (48%). Physical activity and work/life balance were each mentioned by 36% of these employees. The younger the respondent, the more likely they were to say their health was affected by the pandemic (55% among Millennials/Gen Z, 43% among Gen X, 31% among Silent Gen/Baby Boomers), but the concerns cited were similar across age groups. While there were differences by income, none were large (>10%). Ninety-four percent of employees directly affected by the pandemic said they would be open to receiving WHP related to the topics they selected. Figure 1 presents a fuller set of findings from these two groups.Figure 1. Health among Americans employed part/full-time: Total respondents vs Subgroup with health issues directly related to the COVID-19 pandemic. Discussion Most of what we know about WHP, including what employees need and want WHP to address, is based on data gathered before March 2020. Previous research suggests that physical activity and nutrition were top priorities.7,8 Given the dramatic changes to both the working situations of many Americans and priorities around health unique to the pandemic era, it is vital for public health researchers and practitioners to figure out whether and how existing WHP needs to adapt. Over half of all employees indicated that stress and work/life balance were among their top health priorities, regardless of income, and these topics were the top concerns among the subset who reported that the pandemic directly affected their health. These results are consistent with health concerns reported by the broader American population during the pandemic.9,10 We are struck by the relative lack of differences between household income groups. Respondents across income groups generally selected the same health priorities. Consistent with other findings, younger employees were more likely to report a direct impact of the pandemic on their health and well-being. 9 Almost all employees whose health was impacted by the pandemic were open to receiving WHP related to their priorities; this did not differ by household income or age. Future research should explore employee WHP needs and preferences in more depth, and employers’ capacity to address stress and work/life balance, especially in smaller or low-wage worksites that generally have less capacity for WHP.11,12 Summary Our findings reveal that the primary health concerns of American employees include both physical and mental health, with work/life balance and stress of particular concern. For those whose health was directly impacted by the pandemic, priorities focused primarily on mental health issues, including stress, anxiety, sleep (which often negatively reflects increased levels of stress, anxiety and depression that people experience13,14) and depression. Most employees whose health declined during the pandemic are open to receiving support from their employers with their health concerns. Limitations This study was conducted via online survey, which does not allow probing on questions that might have benefited from it. There may be differences between individuals who agree to participate in market research panel surveys and those who do not. Questions addressing openness to workplace health promotion did not include the 54% of employees who indicated their health was not directly affected by the pandemic. Significance Since the onset of the pandemic, there has been little information on the WHP needs and wants of American employees. This research is a first step toward identifying what employees care about, how likely they would be to accept WHP on those topics from their employers, and how existing WHP might need to evolve to meet the challenges experienced by employees since March 2020.So What? What is already known on this topic? While some is known about the WHP wants and needs of employees, including those earning low-wages, surprisingly little research has been conducted with employees, including those earning low wages, who have been impacted most since the onset of the pandemic. What does this article add? Our research adds vitally important information about what American employees currently want and need from WHP. The survey results, conducted among a demographically balanced group of employees, suggest that traditional WHP may need to be adjusted given workplace and health realities since March 2020. What are the implications for health promotion practice or research? Health promotion researchers in the field of workplace-based approaches to addressing health issues need to consider how the programs they promote might need to evolve given the effects of the pandemic. Acknowledgments The authors would like to thank Drive Research for their excellent data collection. ORCID iDs Kristen Hammerback https://orcid.org/0000-0002-5933-1360 Jeffrey R. Harris https://orcid.org/0000-0001-8728-7195 The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by grant 5R01CA160217 from the National Cancer Institute. ==== Refs References 1 U.S. Census Bureau . Current population survey. Available at: https://www.bls.gov/cps/lfcharacteristics.htm#emp https://www.bls.gov/cps/lfcharacteristics.htm#emp. Accessed December 4, 2022. 2 Adams-Prassl A Boneva T Golin M Rauh C . The Impact of the Coronavirus Lockdown on Mental Health: Evidence from the US; 2020. 3 Hamouche S . COVID-19, physical distancing in the workplace and employees’ mental health: implications and insights for organizational interventions - narrative review. Psychiatr Danub. 2021;33 (2 ):202-208.34185751 4 Barkley JE Lepp A Glickman E , et al. The acute effects of the COVID-19 pandemic on physical activity and sedentary behavior in university students and employees. Int J Exerc Sci. 2020;13 (5 ):1326-1339.33042377 5 Cubrich M . On the frontlines: protecting low-wage workers during COVID-19. Psychol Trauma. Aug 2020;12 (S1 ):S186-s187.32551757 6 Peters S Dennerlein TJ Wagner GR Sorensen G . Work and worker health in the post-pandemic world: a public health perspective. The Lancet Public Health. Feb 2022;7 (2 ):e188-e194. 7 Hammerback K Hannon PA Harris JR Clegg-Thorp C Kohn M Parrish A . Perspectives on workplace health promotion among employees in low-wage industries. Am J Health Promot. 2015;29 (6 ):384-392.25162321 8 Parrish AT Hammerback K Hannon PA Mason C Wilkie MN Harris JR . Supporting the health of low socioeconomic status employees: qualitative perspectives from employees and large companies. J Occup Environ Med. Jul 2018;60 (7 ):577-583.29538272 9 American Psychological Association . Stress in America 2020: a national mental health crisis. Available at: https://www.apa.org/news/press/releases/stress/2020/report-october https://www.apa.org/news/press/releases/stress/2020/report-october. Accessed March 23, 2022. 10 Cooke JE Eirich R Racine N Madigan S . Prevalence of posttraumatic and general psychological stress during COVID-19: a rapid review and meta-analysis. Psychiatry Res. Oct 2020;292 :113347.32763477 11 McPeck W Ryan M Chapman LS . Bringing wellness to the small employer. Am J Health Promot. 2009;23 (5 ):1-10, iii. 12 McCoy K Stinson K Scott K Tenney L Newman LS . Health promotion in small business: a systematic review of factors influencing adoption and effectiveness of worksite wellness programs. J Occup Environ Med. Jun 2014;56 (6 ):579-587.24905421 13 Nutt D Wilson S Paterson L . Sleep disorders as core symptoms of depression. Dialogues Clin Neurosci. 2008;10 (3 ):329-336.18979946 14 Khurshid KA . Comorbid Insomnia and Psychiatric Disorders: An Update. Innov Clin Neurosci. Apr 01 2018;15 (3-4 ):28-32.
PMC010xxxxxx/PMC10291211.txt	==== Front Int J Soc Psychiatry Int J Soc Psychiatry ISP spisp The International Journal of Social Psychiatry 0020-7640 1741-2854 SAGE Publications Sage UK: London, England 37353961 10.1177/00207640231183913 10.1177_00207640231183913 Original Article Anxiety, depression and social support of LGBTIQ during COVID-19 in Kerala, India https://orcid.org/0009-0001-1958-5622 Das Harisankar Kannankott 1 https://orcid.org/0000-0002-1324-5387 Govindappa Lakshmana 2 1 National Institute of Mental Health and Neurosciences, Bangalore, Karnataka, India 2 Department of Social Work, School of Social and Behavioral Science, Central University of Karnataka, Kalaburagi, Karnataka, India Lakshmana Govindappa, Department of Social Work, School of Social and Behavioral Science, Central University of Karnataka, Kalaburagi, Karnataka 585367, India. Email: lakshmanagsagar@gmail.com 23 6 2023 23 6 2023 00207640231183913© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background: It is reported that the marginalised and underprivileged sections suffer bitter consequences in the event of calamities and pandemics. The present study aims at assessing the level of anxiety, depression and social support of the LGBTIQ communities during the COVID-19. Since the ‘LGBTIQ’ community is an integral part of society, it is necessary to study these psychological dimensions in the face of multiple waves of the pandemic in the country. Aim: The study aims to measure the anxiety, depression and social support of LGBTIQ during COVID-19 in Kerala. Method: The study followed descriptive research design and using snowball sampling, total of 106 respondents were interviewed from the urban and rural areas of Kerala. The researchers used the ‘DASS21’ to assess anxiety and depression and the Multidimensional Scale of Perceived Social Support to assess social support. Results: Approximately half (44.3%) of the participants were experiencing severe or extremely severe levels of depression. At the same time, many of them had episodes of anxiety disorder at much higher levels (41.5%) than the other members of society. Perceived social support was negatively correlated with depression, anxiety and stress, while depression, anxiety and stress showed a positive correlation with each other. Conclusion: Common mental health problems such as anxiety, depression and stress were largely prevalent in the LGBTIQ community during COVID-19, who found the social support inadequate and suffered from other social and economic problems. There is a need to address these issues among this population. Anxiety depression social support LGBTIQ COVID-19 edited-statecorrected-proof typesetterts1 ==== Body pmcIntroduction Sexual minorities are one of the marginalised and underprivileged sections in society. The emotional and physical trauma persists due to the problems of geniality, desolation, negligence, inadequate support from family and inadequate employment possibilities, added to the pre-existing conditions of marginalisation and social exclusion. A study conducted in 174 countries revealed that the social acceptance of the Lesbian, Gay, Bisexual and Transgender (LGBT) population increased by 75% over the period from 1981 to 2017 (Flores, 2019). However, LGBT population in schools and other educational institutions faces bullying from peers and other professional staff (Watson & Miller, 2012), and the degree of mental health issues increased among the LGBT youth. The lifetime prevalence of depression and anxiety is more than 2.5 times among LGBTQ when compared to heterosexuals (Bennett, 2020). During the COVID-19 pandemic, 72% of LGBTQ youth experienced anxiety disorder symptoms, 62% experienced symptoms of depressive disorder in the US and 70% reported that the pandemic had affected their mental health (Trevour, 2021). LGBTIQ includes Lesbian, Gay, Bisexual, Transgender, Intersex and Queer. They account for approximately 8% of India’s population, as per the scientific estimates cited by the Supreme Court of India while decriminalising same-sex relationships (Sahu, 2016). The social acceptance of the LGBTIQ community has risen since 2010 after a slight decline in the prior decade (Flores, 2019). The need for social awareness about the concept of the LGBTQ community and the mentoring needs of the parents to be more subservient and acquiescent to their children’s sexual orientation are essentially vital to improving the familial and social acceptance along with the mental health of the sexual minorities in India (Tripathi & Talwar, 2022). Methodology The study has used a quantitative research design followed by a qualitative interview. A total of 106 LGBTIQ individuals (20 Lesbian, 10 Gay, 14 Bisexual and 62 transgender) were recruited for the study from different parts of Kerala via snowball sampling. Kerala is a progressive state in the Indian Union, has high literacy rate, has a history of sexuality and gender-related debates and government has taken various initiatives to empower gender and sexual minority by providing social security, healthcare, education and employment opportunities (Thampy, 2022; Tharayil, 2014). Hence, Kerala was chosen as a place to conduct the study. All individuals of the LGBTIQ community who have completed the age of 18 years from all socio-economic status, cultural and religious backgrounds, those who can read and understand English or Malayalam and whether or not socially disclosed their sexual orientation, were invited to participate in the study. The data collection was conducted in March and April 2020. The participants were recruited based on individual contacts via sexual minority supportive organisations and social media. The community members interested in the study were asked to give informed consent before participating in the study without compensation. The data were collected from a single situation using the questionnaire method. The socio-demographic schedule and the DASS-21 (Depression, Anxiety, Stress Scale-21) developed by the University of the New South Wales was used (Lovibond & Lovibond, 1995). DASS-21 scale had Cronbach’s alpha scores of .91 for depression, .84 anxiety and .90 stress. The Multidimensional Scale of Perceived Social Support (MSPSS), developed by Zim, was used to assess the level of social support. These self-reported questionnaires were shared among the population using pen and paper and Google forms. The individuals who had access to smartphones filled the questionnaires through Google form, while others utilised the pen and paper method. The collected data were analysed using IBM SPSS (25) by applying frequency analysis, independent sample test, One-way ANOVA test, Pearson Correlation test and Regression analysis. The study was approved by the Departments ethics committee. Result Socio-demographic details About 19% were lesbians, 9% were gays, 13% were bisexuals and 59% were transgenders. As far as the religion is concerned, 69% follow the Hindu religion, 11% follow Christianity, 17% follow Islam and 3% others. About 44% of the participants were graduates or post-graduates, one fifth were unemployed, 11% were daily wage workers, 32% had a private job, 10% were sex workers, 18% were self-employed and 9% were professionals. About 56% earned less than 1 lakh, and 3% earned between 1 and 2.5 lakhs. It was observed that 60% belonged to the below poverty line category, 65% did not socially disclose their sexual orientation. The mean age of the respondents was 33.35 years (SD ± 9.36) with the minimum age being 19 years and maximum, 66 years. It was also observed that the lowest duration of stay at the current location of the respondents was 1 year, and the highest time of stay was 61 years with the mean duration of stay being 18.60 years (SD ± 14.65). Depression, anxiety and stress About 12%, 25%, 15%, and 29% experienced mild, moderate, severe and extremely severe levels of depression respectively. Regarding anxiety, 28%, 29%, 7% and 6% experienced mild, moderate, severe and extremely severe anxiety levels respectively (Table 1). About 30% and 21% reported severe and extremely severe stress, respectively. Table 1. Depression, anxiety & stress of the respondent. Variables Categories N (%) Depression Normal 20 (18.9) Mild 13 (12.3) Moderate 26 (24.5) Severe 16 (15.1) Extremely severe 31 (29.2) Anxiety Normal 32 (30.2) Mild 30 (28.3) Moderate 31 (29.2) Severe 07 (6.6) Extremely severe 06 (5.7) Stress Normal 17 (16.0) Mild 05 (4.7) Moderate 30 (28.3) Severe 32 (30.2) Extremely severe 22 (20.8) About 19%, 73% and 53% received high support from family, friends and significant others respectively. The total perceived social support shows that 8%, 51%, 42% received low, moderate and high social support, respectively (Table 2). Table 2. Perceived social support of the respondents. Variables Categories N (%) Support from the family Low 46 (43.4) Moderate 40 (37.7) High 20 (18.9) Support from friends Low 6 (5.7) Moderate 23 (21.7) High 77 (72.6) Support from significant others Low 24 (22.6) Moderate 26 (24.5) High 56 (52.8) Total perceived social support Low 8 (7.5) Moderate 54 (50.9) High 44 (41.5) The Below Poverty Line (BPL) is an economic benchmark standard of the government of India related to the threshold income. It helps to identify the financially weaker sections of the population who need government support. Above Poverty Line (APL) mentions the individuals or households not coming under the Below Poverty Line category. There was a significant difference in mean support from significant others between the families of Above Poverty Line (APL) and Below Poverty Line (BPL; t104 = 2.874, p < .05; M ± SD = 20.85 ± 8.50). There was no significant difference between APL and BPL families in the support received from family, friends, perceived total support, depression, anxiety and stress (p > .05; Table 3). Table 3. Test of difference between the socio-economic status of respondents and research variables. Dependent variables Socio-economic status of the participant N M SD t df Sig. (two-tailed) Support from significant others Above the poverty line 42 20.85 8.50 2.874 104 .006 Below the poverty line 64 16.15 8.49 Support from family Above the poverty line 42 12.14 7.66 0.105 104 .917 Below the poverty line 64 11.98 7.60 Support from friends Above the poverty line 42 21.78 6.19 0.508 104 .612 Below the poverty line 64 21.14 6.51 Perceived total support Above the poverty line 42 54.78 16.44 1.668 104 .098 Below the poverty line 64 49.28 16.73 Rate of depression Above the poverty line 42 17.47 11.14 −1.395 104 .166 Below the poverty line 64 20.53 10.95 Rate of anxiety Above the poverty line 42 11.47 8.02 −1.394 104 .166 Below the poverty line 64 13.75 8.33 Rate of stress Above the poverty line 42 19.00 8.82 −0.878 104 .382 Below the poverty line 64 20.65 9.92 ANOVA results using sexual orientation as the predictor variable, revealed that the support from the significant others was significant (F(3,346) = 0.020, p < .05), and the participants belonging to Lesbian sexual orientation had more support from significant others (M ± SD = 22.60 ± 6.90) than from other groups. Sexual orientation did not have a significant bearing on the support received from family, and friends, total perceived social support, and depression, anxiety and stress experienced by the respondents (p > .05; Table 4). Table 4. ANOVA results using sexual orientation as the predictor variable. Criterion Predictor N M SD Sig. Values Support from significant others Lesbian 20 22.60 6.90 SS = 6.416, MS = 247.212, df = 3, F = 3.436, p = .020* Gay 10 21.00 8.45 Bisexual 14 17.92 9.86 Transgender 62 16.08 8.61 Support received from family Lesbian 20 11.85 7.81 SS = 741.637, MS = 2.139, df = 3, F = 0.036, p = .991 Gay 10 11.80 7.65 Bisexual 14 12.64 5.74 Transgender 62 12.01 8.02 Support received from friends Lesbian 20 23.30 4.52 SS = 109.496, MS = 36.499, df = 3, F = 89.8, p = .445 Gay 10 21.10 7.75 Bisexual 14 19.85 7.00 Transgender 62 21.17 6.50 Perceived total support Lesbian 20 57.75 13.58 SS = 1161.932, MS = 387.311, df = 3, F = 1.394, p = .249 Gay 10 53.90 18.18 Bisexual 14 50.42 16.16 Transgender 62 49.27 17.39 Experienced depression Lesbian 20 22.80 10.03 SS = 408.702, MS = 136.234, df = 3, F = 1.114, p = .347 Gay 10 15.40 9.28 Bisexual 14 18.71 13.71 Transgender 62 18.96 10.97 Experienced anxiety Lesbian 20 13.90 8.06 SS = 354.885, MS = 118.295, df = 3, F = 1.775, p = .157 Gay 10 11.20 6.94 Bisexual 14 8.57 8.13 Transgender 62 13.74 8.36 Experienced stress Lesbian 20 20.90 8.14 SS = 165.116, MS = 55.039, df = 3, F = 0.604, p = .614 Gay 10 19.20 8.75 Bisexual 14 17.00 10.12 Transgender 62 20.51 9.93 ANOVA results using the annual income as the predictor variable expressed that the support from the significant others was significant (F(3,566) = 0.017, p < .05), and the participants who have an annual income between 2.5 lakh and 5 lakhs had more support from significant others (M ± SD = 26.66 ± 2.23) than the other groups followed by the participants with an annual income between five lakhs and eight lakhs (M ± SD = 21.00 ± 9.89). The annual income did not have a significant association with the support received from family, and friends, total perceived social support; and depression, anxiety and stress experienced by the respondents (p > .05; Table 5) Table 5. ANOVA results using annual income as the predictor variable. Criterion Predictor N M SD Sig. Values Support from significant others Below INR100,000 59 17.01 08.56 SS = 766.979, MS = 255.660, df = 3, F = 3.566, p = .017 Between INR100,000 and 250,000 36 17.33 09.13 Between INR 250,000 and 500,000 9 26.66 02.23 Between INR 500,000 and 800,000 2 21.00 09.89 Support from family Below INR100,000 59 10.52 07.11 SS = 319.302, MS = 106.434, df = 3, F = 1.895, p = .135 Between INR100,000 and 250,000 36 13.75 08.21 Between INR 250,000 and 500,000 9 14.33 07.33 Between INR 500,000 and 800,000 2 16.00 0.00 Support from friends Below INR100,000 59 20.64 06.99 SS = 113.611, MS = 37.870, df = 3, F = 0.932, p = .428 Between INR100,000 and 250,000 36 22.77 05.36 Between INR 250,000 and 500,000 9 21.33 04.09 Between INR 500,000 and 800,000 2 19.00 12.72 Perceived total support Below INR100,000 59 48.18 17.35 SS = 1945.094, MS = 648.365, df = 3, F = 2.400, p = .072 Between INR100,000 and 250,000 36 53.86 15.80 Between INR250,000 and 500,000 9 62.33 10.14 Between INR 500,000 and 800,000 2 56.00 22.62 Depression experienced Below INR100,000 59 19.52 10.98 SS = 430.605, MS = 143.535, df = 3, F = 1.175, p = .323 Between INR100,000 and 250,000 36 20.55 10.51 Between INR250,000 and 500,000 9 12.88 13.19 Between INR 500,000 and 800,000 2 20.00 14.14 Anxiety experienced Below INR100,000 59 13.52 08.25 SS = 305.873, MS = 101.958, df = 3, F = 1.519, p = .214 Between INR100,000 and 250,000 36 13.16 08.46 Between INR250,000 and 500,000 9 07.33 06.92 Between INR 500,000 and 800,000 2 12.00 0.00 Stress experienced Below INR100,000 59 20.84 10.32 SS = 197.151, MS = 65.717, df = 3, F = 0.723, p = .540 Between INR100,000 and 250,000 36 19.72 08.52 Between INR250,000 and 500,000 9 16.00 08.24 Between INR 500,000 and 800,000 2 18.00 0.00 The Pearson correlation test between the independent and dependent variables revealed that age negatively correlates with depression (r = −.251, p < .01). and postively corelated with duration of staty (r = 0.315, p < .001). The support from the significant others showed direct relation with support from family (r = 0.434, p < .01), support from friends (r = 0.376, p < .01); and negatively correlated with depression (r = −0.418, p < .01), and anxiety (r = −0.205, p < .05). The support from family was negatively correlated with depression (r = −0.523, p < .01), anxiety (r = −0.253, p < .01) and the stress experienced by the respondents (r = −0.299, p < .01). The total perceived social support had a significant level of negative correlation with the rate of depression (r = −0.431, p < .01). The depression had a significant level of direct relationship with the anxiety (r = 0.635, p < .01) and stress (r = 0.695, p < .01). The anxiety had a direct relationship with the stress (r = 0.756, p < .01; Table 6). Table 6. Details of Pearson correlation test between the independent variables and dependent variables. Age Duration of stay Support from Perceived total support Depression Anxiety Stress Significant others Family Friends Age r 1 p N 106 Duration of stay r .315 1 p .001 N 106 106 Support from significant others r .155 .189 1 p 114 .053 N 106 106 106 Support from family r .032 .129 .434 1 p 741 .189 .000 N 106 106 106 106 Support from friends r .075 .074 .376 .062 1 p .443 .448 .000 .530 N 106 106 106 106 106 Perceived total support r .124 .185 .863 .704 .605 1 p .205 .057 .000 .000 .000 N 106 106 106 106 106 106 Depression r −.251 −.188 −.418 −.523 .064 −.431 1 p .010 .054 .000 .000 .518 .000 N 106 106 106 106 106 106 106 Anxiety r −.167 −.072 −.205 * −.253 .094 −.186 .635 1 p .087 .464 .035 .009 .340 .056 .000 N 106 106 106 106 106 106 106 106 Stress r .043 −.080 −.126 −.299 .170 −.137 .695 .756 1 p .661 .417 .196 .002 .081 .162 .000 .000 N 106 106 106 106 106 106 106 106 106 Correlation is significant at the .01 level (two-tailed). * Correlation is significant at the .05 level (two-tailed). ANOVA results showed that there was no significant difference between social disclosure of sexual orientation and study variables (p > .05) among various religious groups on study variables (p > .05) and among participants of various educational qualifications and study variables (p > .05). Regression analysis for the dependent variable social support, with other independent variables such as duration of stay, educational qualification, annual income, sexual orientation and age of the participants showed that these variables predicted 17% (R2 = .174), and the ANOVA expressed that F = 4.207 and p = .002 (Table 7). Table 7. Regression analysis for dependent variable social support. Variable Unstd β Unstd SE Std β t Sig. R 2 F (Constant) 1.784 0.383 4.662 .000 .174 4.207 Age of the participant .014 0.007 .219 2.112 .037 Sexual orientation of the participant −.119 0.053 −.231 −2.241 .027 Educational qualification of the participant .052 0.055 .103 0.943 .348 Annual income of the participant .107 0.084 .128 1.275 .205 Duration of stay in the current location .006 0.004 .150 1.557 .123 Discussion The findings of the present study revealed that the participants with higher level of social support reported lower levels of anxiety and depression, and the support from their family, friends and significant others positively influenced the respondents’ mental health. It also revealed that social support could improve the mental health of individuals. The current study was in line with the existing findings. A similar study carried out in Canada has identified that those who received higher levels of social support experienced lower levels of mental health issues/distress, and effects of social support had four times more effects among sexual and gender minorities (Jacmin-Park et al., 2022). The perceived discrimination and harassment itself were an idol of low social support and social acceptance and negatively affect the mental health and quality of life of sexual minorities (Pandey, 2018). A systematic review reported that the LGBT community was using increased precautions and avoiding public transportation in comparison with the general public due to the fear of pandemics and lack of community support (Konnoth, 2020). The significant challenges of the LGBTQ were: being isolated from unsupportive families, losing in-person identity-based socialisation and social support and reduced access to individual services (Fish et al., 2020). The socially transitioned transgender children backed up their gender status had average levels of depression and only a minimal rise in anxiety (Olson et al., 2016). The results showed that nearly half the participants underwent either severe or extremely severe rate of depression during the pandemic, which was remarkably high for the community. The statistics illustrate that the national average of depression in Indian society was 7.93% (Gururaj et al., 2016) which points that the risk of severe depression among the LGBTIQ population was six times as high as that in the typical population. The results are also on par with the findings of the census bureau of the United States that 38.2% of the LGBTIQ respondents experienced depression for most of the days in the week (Anderson et al., 2021) and experienced higher levels of anxiety, depression and problematic level of alcohol intake during the times of pandemic when compared with the cisgender counterparts (Akré et al., 2021). Among general population, the severity level of anxiety in the LGBTIQ community was two times higher compared to national average which was 5.54% (Murthy, 2017). It was found that more than half of the respondents experienced severe and extremely severe stress levels. Studies conducted on the mental health needs of sexual and gender minorities during COVID-19 identified that 60% of the participants experienced a higher level of anxiety, depression and other psychological distress during the outbreak (Clark et al., 2022; Gonzales et al., 2020), which is going in line with the finding of the current study. Another study reported higher levels of difference in the depression and anxiety during the lockdown and before the lockdown in Belgium among the sexual and gender minorities (Reyniers et al., 2022). It was proclaimed that more mental health problems such as depression, anxiety and stress are recognised among sexual minorities than sexual majorities (Peterson et al., 2021), which is in line with the finding of the current study. The aberrant effects of pandemic solicitudes on psychological symptoms such as distress could be curbed by individual resilience, which distinguished that the LGBTIQ community experienced higher physiological and psychological issues throughout the pandemic (Goldbach et al., 2021), and the same can be inferred from the results of the current study. As an effect of the pandemic situations, young transgender persons might experience gender dysphoria and aerial anxiety, depression and suicidal behaviour (Salerno et al., 2020), which was confirmed by the present findings that depression is higher in lower age group than higher age. A study reported that austerity of depression and anxiety was disproportionally very high among the sexual and gender minorities, and sexual minorities exhibited a greater ubiquity of psychological distress than cisgender and heterosexuals (Moore et al., 2021). It was reported that the risk for anxiety and depression disorders was 1.5 times higher in lesbian, gay and bisexual people, along with increased dependence on alcohol and substance (King et al., 2008). The current study’s findings also implied that the LGBTIQ community faced discrimination based on the socioeconomic status. Among the participants, the transgender community was the most vulnerable and received the lowest levels of social support from the significant others, which colours the social scenario that most transgender people in the country are engaged in sex work to meet their needs (Dabas, 2016). However, the condition of the transgender community had significant improvement in Kerala when compared with that in the other parts of the country, which can be inferred from the data that only 10% were engaged in sex work for their livelihood. Two studies stated that higher levels of perceived social support lowered the likelihood of increased symptoms among black transgender women (Bukowski et al., 2019). The youth with higher perceived social support expressed lower depressive symptoms along with increased self-esteem and coping mechanisms towards the situations (Wilkerson et al., 2017), and the current study described that there exists a negative correlation between anxiety, depression, stress; and support received from family, friends and significant others. Limitations Even though a roughly estimated 25,000 LGBTIQ communities are living in Kerala, only 106 people could be identified due to the COVID-19 outbreak and travel restrictions study. The less visibility of the LGBTIQ community, along with the pre-existing conditions, forced the researchers to limit the scope of their study and could not explore the other contributing factors which had a stake in the mental health conditions of the populations. The lack of reliable data on the prevalence, experiences, needs of the gender minorities in the state as well as the scope and nature of sexual orientation, gender identity and expression change efforts, which are aimed to suppress or deny the identities, made things difficult for the researchers. The study could not focus on other social and physical issues faced by the respondents, which may also have contributed to their stress and anxiety. The participation of other stakeholders was considerably limited in the study due to the pandemic and lockdown. There were more than 70+ identified categories of sexual minorities. In an inclusive study of all of them, the results may vary. Recommendations The future research in this dimension should try to include the other members of the queer spectrum, and more samples should be included in the study. The study should cover dimensions of mental health other than the ones used in the present study. Future studies should also develop an effective intervention plan to improve the target population’s mental health and social support status. Conclusion The legislative and judicial involvements; and organisational interventions generated a considerably notable level of improvement in the life of the LGBTIQ community in the past decades. Significant improvement in the visibility of the LGBTIQ community members was noted in society. In such a situation, it is very relevant to consider the life situation of the LGBTIQ community within society. Even though the livelihood and standard of living of the LGBTIQ population improved, the health concerns of the community still pose a major challenge. During COVID-19, people from all over the country experienced severe mental and emotional issues due to social isolation, alienation and lack of emotional support. So, it is vital to check and understand the physical, mental and emotional well-being of the LGBTIQ community during COVID-19, especially in the domains of anxiety, depression and social support. The study revealed that the life and livelihood of the LGBTIQ community during COVID-19 were miserable in every aspect. In the physical aspect, most of them had lost their jobs, faced an economic crisis due to unemployment, and they struggled to fulfil their basic needs. All the after-effects of these physical struggles were reflected in their emotional well-being. Around half the population were experiencing severe and extremely severe levels of depression besides anxiety disorders experienced by a considerably larger number and, in addition, lack of physical and emotional support from family and significant others. The study was able to clearly understand the problems of the LGBTIQ community while assessing the emotional and psychosocial well-being of the participants. The results clearly indicated the severity of the issue while considering the situation of the respondents’ unawareness of their problems while taking the attitude of mental health hospitals towards the LGBTIQ population into account. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was conducted as part of the partial fulfilment of the Masters in Social Work from the Central University of Karnataka, Kalaburagi. ORCID iDs: Harisankar Kannankott Das https://orcid.org/0009-0001-1958-5622 Lakshmana Govindappa https://orcid.org/0000-0002-1324-5387 ==== Refs References Akré E. R. Anderson A. Stojanovski K. Chung K. W. VanKim N. A. Chae D. H. (2021). Depression, anxiety, and alcohol use among LGBTQ+ people during the COVID-19 pandemic. American Journal of Public Health, 111 (9 ), 1610–1619. 10.2105/AJPH.2021.306394 34410817 Anderson L. File T. Marshall J. McElrath K. Scherer Z. (2021). New household pulse survey data reveals differences between LGBT and Non-LGBT respondents during COVID-19 pandemic. https://www.census.gov/library/stories/2021/11/census-bureau-survey-explores-sexual-orientation-and-gender-identity.html Bennett T. (2020, May 29). Mental health and illness statistics among LGBTQ+: The prevalence of mental illness, risk of suicide, and mental health help. Thriveworks. https://thriveworks.com/blog/mental-health-and-illness-statistics-lgbtq/ Bukowski L. A. Hampton M. C. Escobar-Viera C. G. Sang J. M. Chandler C. J. Henderson E. Creasy S. L. Stall R. D. (2019). Intimate partner violence and depression among black transgender women in the USA: The potential suppressive effect of perceived social support. Journal of Urban Health, 96 (5 ), 760–771. 10.1007/s11524-019-00355-3.31037482 Clark K. D. Lunn M. R. Sherman A. D. F. Bosley H. G. Lubensky M. E. Obedin-Maliver J. Dastur Z. Flentje A. (2022). COVID-19 news and its association with the mental health of sexual and gender minority adults: Cross-sectional study. JMIR Public Health and Surveillance, 8 (5 ), e34710. 10.2196/34710. Dabas M. (2016, July 10). LGTBQ: 90% of Transgenders are Forced into Prostitution, But Healthcare Still Remains A Distant Dream. India Times. https://www.indiatimes.com/news/india/lgtbq-90-of-transgenders-are-into-prostitution-but-healthcare-still-remains-a-distant-dream-257526.html Fish J. N. McInroy L. B. Paceley M. S. Williams N. D. Henderson S. Levine D. S. Edsall R. N. (2020). “I’m kinda stuck at home with unsupportive parents right now”: LGBTQ youths’ experiences with COVID-19 and the importance of online support. Journal of Adolescent Health, 67 (3 ), 450–452. 10.1016/j.jadohealth.2020.06.002 Flores A. R. (2019). UCLA other recent work title social acceptance of LGBT people in 174 countries permalink publication date data availability social acceptance of LGBT people in 174 countries. https://escholarship.org/uc/item/5qs218xd Goldbach C. Knutson D. Milton D. C. (2021). LGBTQ+ people and COVID-19: The importance of resilience during a pandemic. Psychology of Sexual Orientation and Gender Diversity, 8 (2 ), 123–132. 10.1037/sgd0000463 Gonzales G. Loret de Mola E. Gavulic K. A. McKay T. Purcell C. (2020). Mental health needs among Lesbian, Gay, Bisexual, and Transgender college students during the COVID-19 pandemic. Journal of Adolescent Health, 67 (5 ), 645–648. 10.1016/j.jadohealth.2020.08.006 Gururaj G. Varghese M. Benegal V. Rao G. N. Pathak K. Singh L. K. Mehta R. Y. Ram D. Shibukumar T. M. Kokane A. Lenin Singh R. K. Chavan B. S. Sharma P. Ramasubramanian C. Dalal P. K. Saha P. K. Deuri S. P. Giri A. K. kavishvar A. B. Sinha V. K. , . . . Misra R. (2016). National Mental Health Survey of India, 2015-16: Prevalence, Pattern and Outcomes. NIMHANS Publication. Jacmin-Park S. Rossi M. Dumont L. Lupien S. J. Juster R. P. (2022). Mental health and social support of sexual and gender diverse people from Québec, Canada during the COVID-19 Crisis. LGBT Health, 9 (3 ), 151–160. 10.1089/lgbt.2021.0255 35290754 King M. Semlyen J. Tai S. S. Killaspy H. Osborn D. Popelyuk D. Nazareth I. (2008). A systematic review of mental disorder, suicide, and deliberate self harm in lesbian, gay and bisexual people. BMC Psychiatry, 8 , 70. 10.1186/1471-244X-8-70 18706118 Konnoth C. (2020). Supporting LGBT communities in the COVID-19 pandemic. SSRN Electronic Journal. Advance online publication. 10.2139/ssrn.3675915 Lovibond S. H. Lovibond P. F. (1995). Manual for the depression anxiety stress scales. Psychology Foundation of Australia. Moore S. E. Wierenga K. L. Prince D. M. Gillani B. Mintz L. J. (2021). Disproportionate impact of the COVID-19 pandemic on perceived social support, mental health and somatic symptoms in sexual and gender minority populations. Journal of Homosexuality, 68 (4 ), 577–591. 10.1080/00918369.2020.1868184 33399504 Murthy R. S. (2017). National mental health survey of India 2015-2016. Indian Journal of Psychiatry, 59 (1 ), 21–26. 10.4103/psychiatry.IndianJPsychiatry_102_17 28529357 Olson K. R. Durwood L. Demeules M. McLaughlin K. A. (2016). Mental health of transgender children who are supported in their identities. Pediatrics, 137 (3 ), e20153223. 10.1542/peds.2015-3223 Pandey M. K. (2018). Assessing the mental health and quality of life of transgenders: The role of perceived discrimination and harassment. Indian Journal of Mental Health, 5 (3 ), 324–335. 10.30877/ijmh.5.3.2018.324-335 Peterson Z. D. Vaughan E. L. Carver D. N. (2021). Sexual identity and psychological reactions to COVID-19. Traumatology, 27 (1 ), 6–13. 10.1037/trm0000283 Reyniers T. Buffel V. Thunnissen E. Vuylsteke B. Siegel M. Nöstlinger C. Wouters E. (2022). Increased Anxiety and depression among Belgian sexual minority groups during the first COVID-19 lockdown—results from an online survey. Frontiers in Public Health, 10 , 797093. 10.3389/fpubh.2022.797093 35480568 Sahu M. K. (2016). National legal services authority v. Union of India & ors.(AIR 2014 SC 1863): a ray of hope for the LGBT community. BRICS LJ, 3 , 164. Salerno J. P. Williams N. D. Gattamorta K. A. (2020). LGBTQ populations: Psychologically vulnerable communities in the COVID-19 pandemic. Psychological Trauma: Theory, Research, Practice, and Policy, 12 , 239–242. 10.1037/tra0000837 Thampy L. (2022). Social reformation in Kerala; a quest to modernity. Muktshabd, 11 (12 ), 353–362. Tharayil M. (2014). Shifting paradigms: Gender and sexuality debates in Kerala. Economic and Political Weekly, 49 (17 ), 70–78. http://www.jstor.org/stable/24480122 Trevour. (2021). National survey on LGBTQ youth mental health 2021. National Action Alliance for Suicide Prevention. Tripathi A. Talwar T. (2022). Parental acceptance and challenges faced by LGBTQ youth in India and their mental health. In Mahajan V. Chowdhury A. Kaushal U. Jariwala N. Bong A. S. (Eds.), Gender equity; challenges and opportunities (pp. 23–35). Springer. Watson S. Miller T. (2012). LGBT oppression. Multicultural Education, 19 (4 ), 2–7. Wilkerson J. M. Schick V. R. Romijnders K. A. Bauldry J. Butame S. A. (2017). Social support, depression, self-esteem, and coping among LGBTQ adolescents participating in hatch youth. Health Promotion Practice, 18 (3 ), 358–365. 10.1177/1524839916654461 27338277
PMC010xxxxxx/PMC10291215.txt	==== Front Acta Radiol Acta Radiol ACR spacr Acta Radiologica (Stockholm, Sweden : 1987) 0284-1851 1600-0455 SAGE Publications Sage UK: London, England 35068179 10.1177/02841851211070491 10.1177_02841851211070491 Chest Imaging The association between the CT severity index and the pulmonary artery area in COVID-19 pneumonia https://orcid.org/0000-0002-7348-0813 Ongen Gokhan 1 https://orcid.org/0000-0002-3682-2474 Gokalp Gokhan 1 https://orcid.org/0000-0001-6211-4191 Nas Omer Fatih 1 https://orcid.org/0000-0001-6649-9287 Ozpar Rifat 1 https://orcid.org/0000-0002-4742-296X Candan Selman 2 1 School of Medicine, Department of Radiology, 37523 Bursa Uludag University , Bursa, Turkey 2 Ministry of Health, Department of Radiology, Kelkit State Hospital, Gumushane, Turkey Gokhan Ongen, Department of Radiology, Bursa Uludag University Hospital, Bursa, Turkey. Email: drgokhanongen@gmail.com 24 1 2022 7 2023 24 1 2022 64 7 22382244 26 9 2021 13 12 2021 © The Foundation Acta Radiologica 2021 2021 The Foundation Acta Radiologica This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background The pulmonary artery area (PAA) is a valuable non-invasive method for the diagnosis of pulmonary hypertension. Purpose To compare the change in PAA in patients with COVID-19 with the computed tomography (CT) severity index using follow-up imaging. Material and Methods A total of 81 patients who were followed up and underwent CT assessment more than once at our hospital‘s pandemic department were evaluated retrospectively. Patients with progression were separated into three groups: progression ranging from mild-to-mild infiltration (Group A, CT severity index of 0–2); progression from mild to severe infiltration (Group B, CT severity index of 0-2 to 3–5); and progression from severe-to-severe infiltration (Group C, CT severity index of 3–5). The PAAs were calculated separately. Results The mean age was 56 ± 12 years. In terms of those patients showing progression in the CT images, the number of patients in Groups A, B, and C was 29, 40, and 12 in the right lung; 32, 45, and 4 in the left lung; 23, 45, and 13 on both lungs, respectively. There was no significant difference between the main, right, and left PAAs in Group A (P > 0.05). In Group B, there were significant increases in the areas of the main, right, and left PAAs (P < 0.05). There were also significant increases in the areas of the right and main pulmonary arteries in Group C (P < 0.05). Conclusion PAAs increase as disease involvement advances in cases with COVID-19 pneumonia, which is thought to be correlated with progression. COVID-19 pulmonary artery area computed tomography severity index typesetterts19 ==== Body pmcIntroduction COVID-19 pneumonia is a pandemic viral infection that is transmitted between individuals by droplets and direct contact. It was first reported in the Wuhan district of China in December 2019, and by the end of January 2020, its effects had been felt globally (1–4). Since the virus directly affects the respiratory tract and causes infection in the lungs, radiological modalities, primarily computed tomography (CT), play a significant role in diagnosis and follow-up. Bilateral peripheral and subpleural ground-glass views are frequently seen at the onset of the disease, but consolidations, inverted halo view, and crazy-paving pattern appearances are more frequently reported in advanced stages of the disease. Examples of common complications include acute respiratory distress syndrome (ARDS), acute cardiac injury, and secondary infection (5–10). Respiratory tract diseases, infectious and non-infectious diseases that affect the lung parenchyma, may cause increased pulmonary artery diameter and hypertension, depending on the severity of involvement (11). Early diagnosis of pulmonary hypertension in the initial periods of the disease impacts the prognosis and may necessitate different treatment procedures (11,12). Pulmonary hypertension can be diagnosed non-invasively via echocardiography, CT, and magnetic resonance imaging (MRI). A pulmonary arterial diameter >29 mm has a sensitivity of 87% and specificity of 89% for pulmonary hypertension (13). It was previously reported that deteriorations in the coagulation parameters and increased D-dimer levels were observed during the follow-up of cases with COVID-19 pneumonia (5). Moreover, ARDS, the most frequent cause of mortality in these cases, may also be associated with pulmonary arterial micro emboli and hypoxemia (14). Thus, the changes in the pulmonary artery measurements in the follow-up imaging of these patients have particular importance. In patients with COVID-19 pneumonia showing the progression of chest CT findings, pulmonary artery areas may be affected due to the various reasons mentioned above. The aim of the present study was to retrospectively evaluate the association between the CT severity index based on the pulmonary involvement degree and the pulmonary artery area. Material and Methods Study population A total of 102 patients were evaluated retrospectively, who were followed up and underwent CT assessment more than once at our hospital‘s pandemic department between the first reported COVID-19 case in our country, 11 March 2020, to 2 December 2020. The inclusion criteria in our study were as follows: a positive reverse transcriptase-polymerase chain reaction (RT-PCR) test for COVID-19 pneumonia; at least one initial and follow-up chest CT exam; and progression on subsequent CT images compared to the initial CT findings. A total of 21 patients with a secondary disorder, including chronic obstructive pulmonary disease, emphysema, lung cancer, heart failure, cardiomegaly in chest imaging, were excluded. In addition, the patients with frequent pulmonary findings of COVID-19 pneumonia but negative RT-PCR test results were excluded from the study. The study was completed with 81 patients. The RT-PCR tests were positive for all patients. The study was approved by the local ethical committee (2020-23/3). The investigation conforms to the principles outlined in the Declaration of Helsinki. CT technique The CT images of the patients were obtained with a 16-section device (Somatom; Siemens, Erlangen, Germany) with a slice thickness of 1.5 mm in multiplanar reformatted (MPR) images (120 kV, 110–540 mAs). To minimize motion artifacts, patients were instructed to hold their breath at full inspiration. The CT images were evaluated on the Picture Archiving Communication Systems (PACS) by a chest radiologist who had 15 years of experience (R1) and a specialist radiologist who had eight years of experience (R2). The CT findings were evaluated at two timelines (the initial CT and the last follow-up CT) and the CT severity index scores were calculated at each timeline. Two radiologists determined the final score with consensus. To determine the CT severity index, we evaluated ground-glass opacity (GGO), crazy-paving pattern, and consolidation described by Pan et al. (15). The score was calculated for each of the lungs considering the extent of anatomic involvement, as follows: 0 = no involvement; 1 = <5% involvement; 2 = 25% involvement; 3 = 26%–49% involvement; 4 = 50%–75% involvement; and 5 = >75% involvement (15). In this study, we categorized the patients with a CT severity index of 0–2 as a mild infiltration and 3–5 as a severe infiltration. The right pulmonary artery area was calculated as 1.5 cm distally to the pulmonary artery bifurcation, the left pulmonary artery area was calculated as 1 cm distally to the pulmonary artery bifurcation, and the main pulmonary artery area was calculated as 1.5 proximally to the pulmonary artery bifurcation (Fig. 1) (RadiAnt DICOM Viewer 64 bit 2020). The measurements of the pulmonary artery area were performed manually by the same person (R2). Fig. 1. Pulmonary artery measurement method. Initial and follow-up exam images are placed side by side and evaluated according to adjacent anatomical landmarks. Statistical analysis The categorical variables were presented using frequency and percent; parametric continuous variables were presented with mean ± standard deviation; and non-parametric continuous variables were presented with median (interquartile range [IQR]). Patients with progression were separated into three groups for right, left, and both lungs: progression from mild-to-mild infiltration (Group A, CT severity index of 0–2); the progression from mild to severe infiltration (Group B, CT severity index of 0–2 to 3–5) (Fig. 2); and progression from severe-to-severe infiltration (Group C, CT severity index of 3–5). The normal distributions of the right, left, and main pulmonary artery areas were assessed with the Shapiro–Wilk test. The number of cases in each group was >30 and distributed normally. The initial and follow-up values of the pulmonary artery areas of right, left, and main pulmonary artery were compared in these groups of CT severity index (Groups A, B, and C) for right, left, and both lungs, respectively. In addition, we compared the percentage differences of the pulmonary artery areas between the deceased and the surviving patients. The paired t-test was used for comparison if variables were normally distributed, and the Wilcoxon signed-ranks test was used for non-parametric variables after removing outliers. The percentage differences were calculated by the following equation: (r2−r1)/r1×100 (r1: initial area, initial CT severity index; r2: follow-up area, follow-up CT severity index). The association between changes of CT severity index and pulmonary artery areas was assessed using correlation analyses. P < 0.05 was considered statistically significant. In correlation analysis, attention was focused on the R-value if the P value was <0.05. Those with an R-value of <0.40 were interpreted as poor correlation, 0.40–0.60 as mild correlation, and 0.60–0.80 as strong correlation. SPSS version 23.0 (IBM Corp., Armonk, NY, USA) was used for the statistical analyses. Fig. 2. A case with progressive infiltration in both lungs (Group B, CT severity index of 0–2 to 3–5). (a, c) Initial CT images; (b, d) CT images obtained two days later. (a) Bilateral peripheral and subpleural ground-glass opacity was seen, (b) two days later, severe ground-glass progression was obtained. (c, d) The main pulmonary artery diameter was increased from 26 mm to 31 mm, right pulmonary artery diameter was increased from 20 mm to 25.7 mm, left pulmonary artery diameter was increased from 21 mm to 25.8 mm. (e, f) The main pulmonary artery area was increased from 6.88 cm2 to 7.62 cm2. The interval between the two images was two days, and the patient died. Results A total of 81 patients (44 men, 37 women; mean age = 56ׅ ± 12 years) were included. Regarding the patients showing progression in the CT images, the number of patients in Groups A, B, and C was 29, 40, and 12 on the right lung, and 32, 45, and 4 on the left lung, and 23, 45, and 13 on both lungs, respectively. The time between CT exams was in the range of 2–38 days (median = 9 days). In Group A, the right pulmonary artery area was increased from 4.39 (1.63) cm2 to 4.46 (1.85) cm2; the left pulmonary artery area was increased from 4.27 ± 1.19 cm2 to 4.43 ± 1.26 cm2, but the main pulmonary artery area was similar (6.02 (2.11) cm2 to 6.02 (1.16) cm2). There was no significant difference between the main, right, and left pulmonary artery areas in Group A (P > 0.05). In Group B, the right pulmonary artery area was increased from 4.7 ± 1.42 cm2 to 5.26 ± 1.42 cm2, and the left pulmonary artery area was increased from 4.31 ± 1.07 cm2 to 4.76 ± 1.21 cm2. The main pulmonary artery area was increased from 6.73 ± 1.6 cm2 to 7.74 ± 1.68 cm2. In Group B, there were significant increases in the diameters of main, right, and left pulmonary arteries (P < 0.05). In Group C, the right pulmonary artery area was increased from 4.7 (1.29) cm2 to 5.3 (0.98) cm2 and left pulmonary artery area was increased from 4.11 (2.88) cm2 to 5.02 (3.42) cm2, and the main pulmonary artery area was increased from 6.68 (2.64) cm2 to 8.12 (3.42) cm2. There were also significant increases in the diameters of the right and main pulmonary arteries in Group C (P < 0.05). The comparisons between the initial and follow-up values of pulmonary artery areas are presented in Table 1, and the relevant box-plot graphs for percent changes in the pulmonary artery areas are presented in Fig. 3. Fig. 3. Percentage differences in the right, left, and main pulmonary artery areas in each group. Table 1. The initial and follow-up values of the pulmonary artery area in each group. Group Initial area (cm2) Follow-up area (cm2) P Right pulmonary artery A 4.39 (1.63) 4.46 (1.85) 0.21 B 4.7 ± 1.42 5.26 ± 1.42 <0.001 C 4.7 (1.29) 5.3 (0.98) 0.005 Left pulmonary artery A 4.27 ± 1.19 4.43 ± 1.26 0.14 B 4.31 ± 1.07 4.76 ± 1.21 <0.001 C 4.11 (2.88) 5.02 (3.42) 0.068 Main pulmonary artery A 6.02 (2.11) 6.02 (1.16) 0.08 B 6.73 ± 1.6 7.74 ± 1.68 <0.001 C 6.68 (2.64) 8.12 (3.42) 0.001 Values are given as mean ± SD or median (interquartile range). The correlation between the change in CT severity index and area of the pulmonary artery was analyzed using linear regression analysis, which showed that the CT severity index of both lungs was positively but poorly correlated with the pulmonary artery area in progressed cases (P < 0.05). There was no significant association between the remaining measurements. The results of the linear regression are presented in Table 2. Table 2. R and P values for correlation analyses of percentage change of pulmonary artery areas and CT severity index change. Percentage difference in CT severity index Right lung Left lung Both lungs R P R P R P Percentage difference in pulmonary artery area Right pulmonary artery 0.034 0.762 0.15 0.181 0.108 0.338 Left pulmonary artery 0.107 0.344 0.097 0.391 0.148 0.189 Main pulmonary artery 0.158 0.16 0.216 0.052 0.248 0.025 Comparing between the deceased (n = 8) and the surviving patients (n = 73), in the surviving patients, a percentage difference of the main pulmonary artery area was 10.4 ± 14.64, the right pulmonary artery area was 5.72 ± 13.5, and the left pulmonary artery was 7.03 ± 17.33. In the deceased patients, the percentage difference of the main pulmonary artery was 24.86 ± 42.47, the right pulmonary artery was 14.25 ± 23.85, and the left pulmonary artery was 14.2 ± 13.6. There was a significant increase in the main pulmonary artery area in deceased cases compared to survivors (P = 0.011) (Table 3). Table 3. The pulmonary artery areas percentage differences between deceased patients and survivors. Survivors Deceased patients P Right pulmonary artery 5.72 (13.5) 14.25 (23.85) 0.15 Left pulmonary artery 7.03 (17.33) 14.2 (13.6) 0.1 Main pulmonary artery 10.4 (14.64) 24.86 (42.47) 0.011 Values are given as median (interquartile range). Discussion Pulmonary arteries may be affected secondary to infectious and non-infectious diseases. In these cases, increases in pulmonary artery diameters, pulmonary thromboembolism, or pulmonary hypertension may develop (11). COVID-19 pneumonia causes diffuse pulmonary parenchymal involvement by causing vascular endothelial injury, which subsequently advances with vascular diameter changes (9). The most frequently underlying mechanism is progression of ARDS due to thromboembolism and viral pneumonia that develops secondary to endothelial injury and deteriorations in the coagulation cascade (16). Moreover, the frequent involvement of the pulmonary vascular structures may be associated with the thinness of the muscular layer in the pulmonary arteries and veins (17). Menter et al. reported their findings in 21 post-mortem cases in Switzerland, which showed that 76% of cases had a diffuse exudative alveolar disease, 38% of cases diffuse proliferative alveolar disease, massive capillary congestion, and microthrombi in the alveolar capillaries (5/11,45%), and 19% of cases had a pulmonary embolism (18). All these findings were also suspected to be caused by edema and exudation in the intercellular distance associated with secretion of proinflammatory cytokines (tumor necrosis factor alpha, interleukine-1, etc.) against inflammation by the organism (5,18,19). Marini et al. reported that endothelial injury associated with inflammatory changes might result in reflex vasoconstriction and ventilation-perfusion injury as an underlying mechanism (20). The severity of vasoconstriction in pulmonary parenchyma is mainly thought to be associated with increased pulmonary artery diameter and a cause of hypertension. Moreover, Yang et al. reported in their study on SARS in 2005 that autoantibodies against the virus may cause vascular endothelial injury (21). The changes in the coagulation parameters in patients’ follow-up for COVID-19 pneumonia are associated with poor prognosis, such as increases in D-dimer and troponin, and associated with prolonged hospitalization and death rates (22). In addition, in two different studies, 23%–30% of patients with COVID-19 pneumonia were reported to have had at least one episode of pulmonary embolism (23,24). Besides thromboembolic complications, COVID-19 pneumonia can cause cardiac damage as SARS-CoV-2 can infect the myocardium. Moreover, hypoxemia, which develops in severe patients, is among the other causes of cardiac failure. These factors may be responsible for the increasing diameter of the pulmonary artery (25). Pulmonary artery diameters can be measured by several methods using a chest CT. The pulmonary artery is often measured transversely in the axial section (26). There are two studies on the measurement of pulmonary artery diameters in the literature (16,27). In both studies, pulmonary artery diameters were compared with ascending aorta, and measurements were made only in the axial plane. Instead of transverse measurement of the diameter, the measurement on the plane perpendicular to the arterial axis in oblique reformatted images in all three planes is a more objective approach to the measurement of pulmonary artery diameters. In this study, in addition to the measurement of the main pulmonary artery area, we compared the areas of the right and left pulmonary artery areas with the CT severity index because the main pulmonary artery and its main branches may be affected differently by the severity of disease involvement. Yildiz et al. reported that in patients with severe infiltration, pulmonary artery diameters were larger than in patients with mild infiltration (27). In our study, there was no significant difference between changes of main and right, left pulmonary arteries in Group A (mild to mild) (P > 0.05). However, all areas in Group B (mild to severe) and the main and right areas in Group C (severe to severe) significantly increased (P < 0.05). The significant difference between the pulmonary artery area in lung involvement in Group B and Group C suggest that severe pulmonary artery area changes occur in severe disease. In the deceased patients, the percentage difference in the right, left, and main pulmonary artery areas was increased compared to that of the surviving patients. However, only the increase in the main pulmonary artery area was statistically significant (P = 0.011). Although there was a rise in the right, left, and main pulmonary arteries in Groups B and C, the main pulmonary artery change was only statistically significant in the deceased group, from which it can be assumed that the deceased group had a small number. A poor meaningful positive correlation was found between the CT severity index of both lungs and the main pulmonary artery area. The correlation itself is a good indicator of the relationship between these two factors despite it being poor, which shows an increase in CT severity index results with an increase in pulmonary artery area. Since the pulmonary artery area is affected by many other factors, we do not expect a linear relationship between the pulmonary artery area and severity index. The present study has some limitations. These include the low number of cases and the retrospective design of the study. We did not evaluate inter-observer variability and use of intravenous contrast medium in the imaging studies. In addition, we did not know the cardiac compliance of patients, laboratory findings, and if there was a presence of pulmonary artery embolism. In conclusion, the area of pulmonary arteries increases as the disease involvement advances in cases with COVID-19 pneumonia and are thought to be associated with hypoxemia, infiltration, and maybe reflex vasoconstriction. These changes in the pulmonary artery area were remarkable in the deceased patients. For this reason, an increase in the pulmonary artery area should be assessed by the radiologists along with the CT severity index. Radiological findings that appeared from respiratory and cardiac systems highlighted in this study should be supported by further clinical studies. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship and/or publication of this article. ORCID iDs: Gokhan Ongen https://orcid.org/0000-0002-7348-0813 Gokhan Gokalp https://orcid.org/0000-0002-3682-2474 Omer Fatih Nas https://orcid.org/0000-0001-6211-4191 Rifat Ozpar https://orcid.org/0000-0001-6649-9287 Selman Candan https://orcid.org/0000-0002-4742-296X ==== Refs References 1 World Health Organization. Novel Coronavirus (2019-nCoV). Available at: https://www.who.int/docs/default-source/coronaviruse/situation-reports/20200121-sitrep-1-2019-ncov.pdf (accessed 15 June 2021). 2 World Health Organization. Coronavirus disease (COVID-19) technical guidance: The Unity Studies: Early Investigation Protocols. Available at: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/technical-guidance/early-investigations (accessed 15 June 2021). 3 Johns Hopkins Whiting School of Engineering website. Coronavirus COVID-19 global cases. Available at: https://Gisanddata.maps.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6 (accessed 15 June 2021). 4 Chan JF Yuan S Kok KH , et al. A familial cluster of pneumonia associated with the 2019 novel corona virus indicating person to-person transmission: a study of a family cluster. Lancet 2020;395 :514–523.31986261 5 Huang C Wang Y Li X , et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet 2020;395 :497–506.31986264 6 Salehi S Abedi A Balakrishnan S , et al. Coronavirus disease 2019 (COVID-19): a systematic review of imaging findings in 919 patients. AJR Am J Roentgenol 2020;215 :87–93.32174129 7 Bernheim A Mei X Huang M , et al. Chest CT findings in coronavirus disease-19 (COVID-19): relationship to duration of infection. Radiology 2020;295 :200463.32077789 8 Raptis CA Hammer MM Short RG , et al. Chest CT and coronavirus disease (COVID-19): a critical review of the literature to date. AJR Am J Roentgenol 2020;215 :839–842.32298149 9 Zhao W Zhong Z Xie X , et al. Relation between chest CT findings and clinical conditions of coronavirus disease (COVID-19) pneumonia: a multicenter study. AJR Am J Roentgenol 2020;214 :1072–1077.32125873 10 Xu R Du M Li L , et al. CT Imaging of one extended family cluster of corona virus disease 2019 (COVID-19) including adolescent patients and “silent infection”. Quant Imaging Med Surg 2020;10 :800–804.32269939 11 Oswald-Mammosser M Weitzenblum E Quoix E , et al. Prognostic factors in COPD patients receiving long-term oxygen therapy: importance of pulmonary artery pressure. Chest 1995;107 :1193–1198.7750305 12 Olschewski H Simonneau G Galiè N , et al. The aerosolized iloprost randomized study group inhaled iloprost for severe pulmonary hypertension. N Engl J Med 2002;347 :322–329.12151469 13 Tan RT Kuzo R Goodman LR , et al. Utility of CT scan evaluation for predicting pulmonary hypertension in patients with parenchymal lung disease. Chest 1998;113 :1250–1256.9596302 14 Marongiu F Grandone E Barcellona D . Pulmonary thrombosis in 2019-nCoV pneumonia? J Thromb Haemost 2020;18 :1511–1513.32293083 15 Pan F Ye T Sun P , et al. Time course of lung changes on chest CT during recovery from 2019 novel coronavirus (COVID19) pneumonia. Radiology 2020;295 :715–721.32053470 16 Spagnolo P Cozzi A Foà RA , et al. CT-derived pulmonary vascular metrics and clinical outcome in COVID-19 patients. Quant Imaging Med Surg 2020;10 :1325–1333.32550141 17 Tekcan Şanlı DE Yıldırım D . A new imaging sign in COVID-19 pneumonia: vascular changes and their correlation with clinical severity of the disease. Diagn Interv Radiol 2021;27 :172–180.33044171 18 Menter T Haslbauer JD Nienhold R , et al. Postmortem examination of COVID-19 patients reveals diffuse alveolar damage with severe capillary congestion and variegated findings in lungs and other organs suggesting vascular dysfunction. Histopathology 2020;77 :198–209.32364264 19 Rankin JA . Biological mediators of acute inflammation. AACN Clin Issues 2004;15 :3–17.14767362 20 Marini JJ , et al. Management of COVID-19 respiratory distress. JAMA 2020;323 :2329–2330.32329799 21 Yang YH Huang YH Chuang YH , et al. Autoantibodies against human epithelial cells and endothelial cells after severe acute respiratory syndrome (SARS)-associated coronavirus infection. J Med Virol 2005;77 :1–7.16032747 22 Tang N Li D Wang X , et al. Abnormal coagulation parameters are associated with poor prognosis in patients with novel coronavirus pneumonia. J Thromb Haemost 2020;18 :844–847.32073213 23 Léonard-Lorant I Delabranche X Séverac F , et al. Acute pulmonary embolism in patients with COVID-19 at CT angiography and relationship to d-dimer levels. Radiology 2020;296 :189–191.32293223 24 Grillet F Behr J Calame P , et al. Acute pulmonary embolism associated with COVID-19 pneumonia detected with pulmonary CT angiography. Radiology 2020;296 :186–188. 25 Ruan Q Yang K Wang W , et al. Clinical predictors of mortality due to COVID-19 based on an analysis of data of 150 patients from Wuhan, China. Intensive Care Med 2020;46 :846–848.32125452 26 Ieki H Nagatomo Y Tsugu M , et al. Impact of pulmonary artery-to-aorta ratio by CT on the clinical outcome in heart failure. J Card Fail 2019;25 :886–883.31100468 27 Yildiz M Yadigar S BŞ Y , et al. Evaluation of the relationship between COVID-19 pneumonia severity and pulmonary artery diameter measurement. Herz 2021;46 :56–62.33433652
PMC010xxxxxx/PMC10291216.txt	==== Front spepc EPC Environment and Planning. C, Politics and Space 2399-6544 2399-6552 SAGE Publications Sage UK: London, England 10.1177_23996544231180462 10.1177/23996544231180462 Lockdown and the Intimate Under lockdown: Remaking “home” through infrastructures of care during COVID-19 https://orcid.org/0000-0001-5353-184X Muñoz Solange Department of Geography and Sustainability, 4292 University of Tennessee , Knoxville, TN, USA Clark Jordin * Department of Rhetoric, 2719 Wabash College , Crawfordsville, IN, USA Auerbach Jeremy School of Geography, 8797 University College Dublin , Ireland Hardwig Lily Department of English and Global Studies, 4292 University of Tennessee , Knoxville, TN, USA Solange Muñoz, Department of Geography and Sustainability, University of Tennessee, 410 Burchfiel Building, Knoxville, TN 37996-4519, USA. Email: imunoz@utk.edu * Current affiliation: Department of Communication and Media, West Chester University, West Chester, PA 22 6 2023 22 6 2023 23996544231180462© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. This paper examines how poor urban residents in the Sun Valley public housing community in Denver, CO (US) experienced the pandemic during the first few months of the crisis. Employing a framework that focuses on people, community, housing and home as potential spaces and possibilities of “infrastructures of care”, this research examines the strategies and practices that emerged during the pandemic to address the immediate needs and concerns of the Sun Valley residents. We consider how these practices and the pandemic itself have potentially led to new imaginings and understandings of home and community, both at the intimate and collective scales. Using qualitative methods and photo-voice techniques, we documented residents’ experience during lockdown. Their narratives reveal the many ways the COVID-19 pandemic has affected people’s lives and highlight how community support, services and home are necessary for ensuring that residents can develop resilient infrastructures of care that allow them to overcome public health crises. COVID-19 care infrastructures house and home edited-statecorrected-proof typesetterts10 ==== Body pmcIntroduction To mitigate and control the spread of COVID-19, governments across the world imposed various restrictions on economic activities and social interactions. Although the lockdown may have decreased the overall number of COVID-19 cases and hospitalizations during those early months, it also upended people’s lives and lead to an unprecedented rise in unemployment (Svaldi, 2021), food insecurity (Nicholson, 2020), evictions (Root Policy Research, 2020), mental and physical health issues (Griego and Greene, 2020), and domestic violence (McCrary and Sanga, 2021), all the while isolating people in their houses and homes, away from family, friends, and community support. The COVID-19 pandemic and lockdown also underscored the significance of housing and home as a primary space from which individuals, families, and communities structure, provide, and perform relationships and dynamics of care, often under extreme conditions of precarity, poverty, or violence. At the same time, the impacts of the lockdown demonstrated the lack of government preparedness and infrastructures in place to support individuals and communities to adequately address the crisis (Silver and Hyman, 2020). During COVID-19, many scholars pointed to the uneven role of government in exacerbating structural inequalities, and social and spatial marginalization of families and communities, especially in minority and poor neighborhoods already struggling even before the pandemic (Chen et al., 2021; Tai et al., 2021). Put simply, the lockdown further isolated and marginalized many families and individuals, not just by keeping them in their homes, but also through people’s isolation and struggle to meet all of their daily professional, domestic, and financial responsibilities amidst extreme confusion and uncertainty regarding the virus and its effects. Yet, the COVID-19 pandemic and lockdown also gave rise to examples of collective and community action across the globe—friends, families, and communities were able to come together to find ways to collectively address the new reality and demands under the COVID-19 pandemic and lockdown. In some neighborhoods, families organized around pods, made up of parents and their children who would play and do school activities together as a form of community or shared child-care in the context of social distancing (Horn, 2021). Some urban communities came together to fight against eviction from homes and buildings, since many families fell behind on rent or mortgages during the pandemic (Massarenti, 2020; Mendes, 2020; Vilenica et al., 2020). In other neighborhoods, the community created soup kitchens to help feed people who had lost their jobs and were unable to provide sufficient food for themselves and their families (Mendes, 2020; Vilenica et al., 2020). These examples highlight the simultaneous hardships, and the ways that people and communities were able to collectively organize to address the challenges of this new COVID reality. Of course, community practices are not new. Communities have always responded to crisis through local collective care practices, often in reaction to the state’s absence or failure to provide adequate services and support (Lancione, 2020; Simone, 2004). However, the intensity and the urgency of the COVID-19 pandemic and lockdown meant that people and communities around the world found themselves scrambling and improvising ways to address the multiple challenges and uncertainties the pandemic suddenly posed. From literature that employs the concepts of infrastructures and care practices (Alam and Houston, 2020; Hobart and Kneese, 2020; Power and Mee, 2020), we examine the ways that residents developed strategies to access needs and resources, and identify and discuss the community-based infrastructures of care that emerged as a response to crisis and uncertainty. We begin at the scale of the house and home in part because of its significance in the COVID-19 pandemic as a state-mandated refuge from the virus, as well as an intimate space in which people were ultimately forced to conduct so many of their daily activities and responsibilities. We employ house and home interchangeably to consider the myriad emotional, social, and material meanings of these mutually imbricated spaces and imaginaries (Blunt and Dowling, 2006; Easthope et al., 2020; Handel, 2019; Power and Mee, 2020). Unlike more traditional understandings of house/home as a private and separate domain outside of the broader public suburban or urban realm (Altman and Werner, 1985; Dovey, 1985; Wardhaugh, 1999), in this paper house/home is conceptualized as part of multi-scalar networks of intimate, social, economic, and material relationships that often begin at the house/home, but that are not exclusive to it (Blunt and Dowling, 2006; Handel, 2019; Power and Mee, 2020). Plainly stated, the house/home is understood to constitute a space from which people organize their daily lives, their relationships, and their access to resources, spaces, and services within a localized geographical area and socio-political context on which they rely in order to achieve and maintain some degree of individual and collective wellness. Building on the works from scholars like Emma Power Power and Mee, 2020, Katherine Mee Alam and Houston (2020) among others, we draw on the concepts of house/home, care, and infrastructure to examine the ways in which a broadening and reimagining of these concepts can offer a framework for understanding and building on community practices of care, solidarity, and networks that emerged from the COVID-19 pandemic and lockdown. Specifically, this paper asks the following research questions: how has the Covid-19 pandemic and lockdown created opportunities for infrastructures of care? What is the social and spatial significance of house/home in the production of collective forms of care? Finally, what can we learn from community produced infrastructures of care to build more sustainable, just, and resilient societies? In the following pages we first present the theoretical framework for our case study. We engage in a discussion that seeks to both define the concepts of infrastructure, care, and house/home in their more traditional forms and then expand on the definitions through current and critical approaches. We then briefly discuss the impact of the COVID-19 pandemic as a care crisis in the context of the house/home. Third, we present our case study and methods as an example of community care practices and infrastructures that emerged in the context of the COVID-19 pandemic and lockdown. We employ our findings to consider the ways in which the COVID-19 moment has provided opportunities for the emergence of different kinds of relationships and care practices that have the potential to resist neoliberal structures, and to produce care infrastructures that promote relationships of mutuality and solidarity. Finally, we end the paper by discussing our findings and further positioning them within a broader theoretical analysis of the significance and potential of community-based care infrastructures as response and resistance to current, neoliberal economic structures that prioritize the individual, monetize care work and practices, isolate workers, and communities. Changing the infrastructures paradigm Recent studies in Geography and other disciplines have focused attention on the concept of “infrastructure” as a term that explains not only how societies are patterned by large state-funded public works, but also how the actions and organizations of communities and individuals can also pattern societies in similar ways (Alam and Houston, 2020; Amin, 2014; Berlant, 2016; Power et al., 2022; Simone, 2004). This has led to a broadening of the definition of infrastructure and to debates regarding what infrastructure is and should be—debates that reflect those that recently occurred inside the halls of the U.S. government (Hamilton, 2021). In this research, we apply an infrastructure framework to care practices and house/home in order to explore how spatial and social patternings of people’s domestic everyday practices, relationships, and spaces constitute infrastructural regimes. From state to shadow care infrastructures Infrastructures organize society through spatial applications that are unevenly distributed among people and communities. As everyday forms of social and spatial organization, infrastructures both represent and reproduce people’s varied accessibility to resources and the inherent power dynamics on which interrelationships are built (Graham and McFarlane, 2014). In their most extreme form, state organized infrastructures reinforce extreme inequality within a society, what Graham and McFarlane (2014) define as infrastructural violence, coded into the everyday through uneven access and opportunities. For whom infrastructure projects are developed, who accesses infrastructure, who benefits from infrastructure projects and who does not are all representations and reproductions of the inherent inequality of society and experienced socially, spatially, and materially. These technologies essentially “pattern social life and identify the values that are selectively coded into infrastructures, (re)producing social difference through use” (Power and Mee, 2020: 485). Although infrastructures are inherently power-laden and uneven, producing and reinforcing social and spatial inequality and marginalization, they are usually normalized in the backdrop of daily life, as are their uneven impacts on space and society. Scholars like Amin (2014), Graham and McFarlane (2014), and Simone (2004) acknowledge the normalizing and invisible character of infrastructure, explaining that infrastructure’s invisibility is inherent to its functionality. Graham and Marvin (2001) argue that it is only in times of crisis, when infrastructures fail and people are no longer able to successfully attend to their daily needs that they become visible. These moments of crisis and failure are often (forced) opportunities for people and communities to produce new social and technical infrastructures (Berlant, 2016). Simone (2004) emphasizes that when infrastructure regimes stop serving societies, they become a vector through which people and communities find ways to reconstitute the spaces and structures under which they live through local and dynamic everyday practices and social relationships. Infrastructure scholars have increasingly begun to focus more on the localized and social origins of infrastructures regimes, employing concepts like people as infrastructure (Simone, 2004) alternate infrastructures of care (Alam and Houston, 2020), bottom-up infrastructures (Clark et al., 2022), infrastructure as commons (Berlant, 2016), and lively infrastructures (Amin, 2014). This scholarship works to broaden the conceptual terrain of how we understand the power and agency of what are often traditionally marginalized and poor communities that, in many cases, have remained outside of or marginal to state-based infrastructure and welfare regimes, but that have none the less developed other kinds of infrastructures and welfare regimes. Within this growing body of literature on infrastructure, the concept of care is still somewhat obscure. Yet, we find the concept of care to be a powerful one in understanding not only the ways in which people and communities organize, but also in understanding the material impacts and the broader significance of care practices for resistance and social change (Lynch, 2022). Care, in the context of infrastructure, can be understood as encompassing practices and activities in which people engage in the everyday in order to make ends meet and get things done. Broadly conceived, care practices rely on both individual and collective practices, relationships, and interdependence that together make up infrastructure regimes on which people and communities rely in order to meet basic needs and, in some cases, ensure survival. Building on the literature on infrastructure (mentioned above) and care (described further in the following section), Power et al. (2022) have recently proposed the concept of shadow care infrastructures to “trace a more comprehensive set of relations through which life is sustained, across different domains and sectors of provisioning in post welfare cities” (p. 1166). Focusing on poor and marginalized communities, Power et al. (2022) describe a reality in which residents rely on multiple and often disparate infrastructures which may include state run infrastructures and welfare regimes, in addition to many other care strategies and arrangements necessary to make ends meet (Power et al., 2022). Similar to other conceptualizations of care and infrastructure, the concept of shadow geographies refers to those care infrastructures that are often invisible or unrecognized under more institutionalized understandings of infrastructure and welfare regimes. Importantly, the authors acknowledge that these practices operate not outside of state-based infrastructures but in relation and in reaction to them (Power et al., 2022). This reconceptualization provides a comprehensive and integrative framework that recognizes the myriad forms of care infrastructures and the conditions and spaces in which they emerge. Important to this research are the ways in which the authors seek to operationalize the study of diverse care infrastructures. Infrastructures of care Much of the literature on care and care infrastructures employs Tronto’s (1993) feminist work on care ethics, which understands care as relational as well as an activity and a moral practice. These scholars build on Tronto’s work by employing both infrastructure and care to highlight the socio-spatial and material technologies that are also representative of how care is produced, maintained, and how it is imagined. Power and Bergan (2019) argue that relational care hinges around two central concepts: that everyone is vulnerable and therefore in need of care in some way; and that all humans are interdependent and therefore rely on each other at some point in their lives (Power and Bergan, 2019; Tronto, 2013). This understanding of care and society challenge capitalist logics and neoliberal norms that “produces a social ‘nature’ in humans that is characterized by coldness and indifference in politics and public life” (Lynch, 2022: 27). Lynch argues that neoliberalism is not simply an analytical model from which to understand economic structures and social realities, but rather, “it is a normative framework [that] prescribes who you should be and how you should be” (Lynch, 2022: 27). When applied to an ethics of care, neoliberalism’s logic works to destroy or severely limit potential connections and conditions of interdependence, instead pitting people against each other and thus, isolating them from mutual care regimes like those described above. Employing a spatial analysis to this conundrum, Alam and Houston (2020) argue that we need to untangle care from the private sphere in order to allow it to reemerge as a public function that can exist communally and outside of fragmented dominant imaginaries of domestic individualism produced and reinforced by neoliberal structures. Instead, Alam and Houston (2020) claim that it is the everyday, non-institutional, relational, and infrastructural spaces that function as a vector through which life is both enabled and constrained. It is in these everyday spaces and practices, that interrelations within and across communities can create other kinds of infrastructures of care outside of capitalist structures (Alam and Houston, 2020). Employing the concept of “caring with”, which they explain “recognizes care as a dynamic relation” between both caregiver and care-receiver (Alam and Houston, 2020; see also Power, 2019), they also argue that this leads to greater trust as well as responsibility and capacity (Tronto, 2013 in Alam and Houston, 2020: 3). Alam and Houston (2020) continue their argument by employing the concept of “care collectives” to illustrate the mutual relationship between community members in which care is mutually constitutive between those who provide and those who receive it. This relational, interdependent and multi-scalar model of care is part of what Alam and Houston (2020) call a “participatory urban flourishing” in which mutual care collectives build their own infrastructures of collective care that address local needs and demands, while also extending outward to other spaces and practices (p. 8). Hobart and Kneese (2020) call these care collectives “coalitions,” and explain that they can “inspire people to work together across class, race, ethnicity, religious, and state boundaries toward a common cause,” furthering the idea of collective, community flourishing across scales (Hobart and Kneese, 2020: 9). Although many of these authors mentioned offer what might be conceived as simplistic uses of concepts like care, many recognize that care practices and infrastructures can also be problematic, abusive, and exploitative (Lynch, 2022; Power et al., 2022). As Power et al. (2022) state, “The motives and effects of care are not necessarily benevolent or beneficial. […] care practices are not always caring (p, 1174: see also: Lawson, 2007; Williams, 2017). At the same time however, these approaches reconceptualize how infrastructures and care can be experienced, understood, and applied through practices, relationships, spaces, and infrastructures that are grounded and formed in community agency, relationality, interdependence as well as plural ontologies and collectivities. These applications of infrastructure and care are not happenstance nor trite, but rather are part of critical approaches that challenge neoliberal regimes and structures while also reimagining and promoting more egalitarian and care-full societies and ways of life (Power and Bergan, 2019). These reimaginings offer a “roadmap for an otherwise” and allows for new life-world understandings that escape the individualism and fragmentation of neoliberal housing and care regimes (Hobart and Kneese, 2020: 13). Infrastructures of house and home In this paper, housing/home is also understood through an infrastructure of care framework, positioning it at the center of care practices through particular understandings of what constitutes the domestic, relationships, responsibilities, and community. Despite scholarship by feminists that have challenged traditional understandings of house/home as representative exclusively of the private realm, it remains undertheorized in terms of its myriad collective and multi-scalar significance as an infrastructure and technology for the organization of people, communities, and care practices (Power and Mee, 2020). At the same time, the house/home continues to be imagined as the “central location for care work” to happen, as well as the primary space in which care work is defined, “[informing] the performance of care” (Power and Mee, 2020). In the United States and other wealthy countries where, in addition to care regimes that are defined by capitalist structures that seek to monetize care relationships in the name of profit, the house/home constitutes a space that is often rigidly separated from public and collective spaces through zoning laws, cultural norms, and lack of investment in public services like transportation. In this context, the naming of the domestic sphere as the primary location for care work reinforces its private, fragmented, and exclusive significance for much of society. More generally, the way that care work is socially and spatially constituted in the US and other wealthy nations often isolates care workers from one another, devalues care labor as outside the realm of “real work,” “obscures relational caring practices,” and prevents collectivization between care workers locally and internationally (Blunt and Dowling 2006; Lopez and Neely, 2021). Research like Blunt and Dowling’s (2006) work on home, and Power and Mee’s (2020) analysis of home as an infrastructure of care in which they explore how infrastructural forms shape the possibility of care, and how care is translated through housing present a more nuanced, complex, and critical understanding of house/home as infrastructure. Through a multi-scalar analysis of the significance of housing regimes Power and Mee (2020) illustrate the ways in which care is valued, organized, and circulated through space. Specifically, they argue that the ways in which housing regimes are produced and understood through design, markets, and governance patterns how care is performed, valued, and spatialized. Importantly for this research, their analysis highlights possibilities for the production of other kinds of care regimes by recognizing and reevaluating housing and the house/home as an infrastructure of care. Many feminist scholars argue that for women and minorities in particular, the house and home are representative of multiple lives, jobs, spaces, and experiences that overlap in myriad ways (Blunt and Dowling, 2006; Easthope et al., 2020; Hooks, 1990). These include working from home, or working in other people’s homes, or the ways in which domestic responsibilities and care do not remain inside the house and home, but rather move through local, national, and transnational spaces (Muñoz, 2021). These and other conceptualizations position house/home as dynamic, multi-scalar sites and infrastructures from which individuals and communities engage in a series of relationships and activities that move through different spaces, but that are experienced and interconnected in dynamic ways. Building from these understandings, we turn to COVID-19 as a context wherein house/home were essential infrastructures of care that reached beyond the domestic, private home. We argue that as dynamic and multiscalar, people’s care practices within and through the house/home crafted alternative forms of care that expanded into collective spaces and reimagined community practices. The home in a COVID-19 world The COVID-19 pandemic and lockdown posed numerous challenges for caregivers, care-workers and care regimes and infrastructures in general. Women and minorities bore much of the burden of care for family members, and often charged with managing the generalized instability and precarity caused by the COVID-19 pandemic and top-down infrastructure failures (Thomason and Macias-Alonso, 2020). In one way, class, gender and racial disparities were prevalent because of the ways in which the Covid-19 pandemic and lockdown impacted who could stay home—with only 20% and 16% of African Americans and LatinX workers respectively able to work from home (Thomason and Macias-Alonso, 2020). Further, inside the home, women assumed most of the responsibility for childcare and homecare even while continuing to work full-time. When forced to choose between one’s profession and taking care of family, it is most often women who leave their professions (Landivar et al., 2020). In September 2020 when many schools went to online learning for the year, 80% of the 1.1 million people who left the workforce were women (US Bureau of Labor Statistics, 2020). These social, spatial, and gendered conditions and the social, political, and economic structures and contexts in which they occur, illustrate how care and care work remains marginalized, undervalued, and exploitative. At the same time, the COVID-19 pandemic and lockdown have revealed and exacerbated the housing crisis and a crisis of care that exist in multiple spaces and scales. As stated above, these crises and their impacts are part of broader social and political infrastructures in the United States and other nations that produce and reinforce neoliberal regimes and care systems that isolate and marginalize people and communities. The COVID-19 crisis is an opportunity from which we can start to reimagine our society through research that identifies shadow infrastructures of care, through different kinds of housing regimes that prioritize collective care practices, and through care practices that promote mutuality and solidarity. Sun Valley case study We apply this discussion to the empirical case study of Sun Valley, a public housing community in Denver, Colorado. At the time of our study, Sun Valley Homes was a public housing community built in the 1950s and home to approximately 1300 Denver residents who made up an economically impoverished, yet diverse community that included refugees and immigrants, Black and LatinX families, single parent households, and individuals who suffer from chronic health conditions or who are permanently disabled, and where more than 30 languages are spoken. Sun Valley is one of the poorest neighborhoods in Denver, physically and symbolically separated from the city center by the football stadium to the north, I-25 to the East and South, and the Platte River to the Northeast. The surrounding neighborhood has traditionally been characterized by food deserts, poor public transportation, and lack of other public amenities (Jackson, 2018). At the same time, the community was organized around two-story condominium style housing, with a front and back door that opened out into open, shared yards and sidewalks (Figure 1). The community also had small playgrounds and a garden where vegetables are grown that are then sold at a weekly farmer’s market. These socio-spatial conditions contributed to the many care practices that were implemented during the COVID-19 lockdown.Figure 1. Original Sun Valley community during the COVID-19 pandemic. When the COVID-19 pandemic began, our team was conducting a health impact assessment of the Sun Valley housing redevelopment on its residents. In the US, public housing policy has suffered from disinvestment and shifting paradigms from government funded housing communities to an individualized and voucher-based system that works within the capitalist, private market (Goetz, 2012; US Department of Housing and Urban Development (HUD), 2021). The current program, the Choice Neighborhood Initiative (CNI), is part of urban redevelopment projects that are transforming distressed public housing communities into public-private mixed-income, mixed-use spaces around the country. The Sun Valley neighborhood is one such redevelopment. This construction, which was in its final phases when the City of Denver imposed a strict stay-at-home lockdown, had already caused significant environmental and social stress to Sun Valley residents who were still living in the old housing as construction went on around them. For many Sun Valley residents who lost their jobs or whose wages were cut during the pandemic, the lockdown exacerbated their already precarious economic and health well-being. Yet, as a public housing neighborhood and in part because of the on-going redevelopment, there was some degree of stability and support through secure housing tenure and some assistance from the on-site CNI office, tasked with facilitating Sun Valley residents with the redevelopment project. During COVID, staff from CNI and community residents were instrumental in organizing and delivering food and household staples to residents. Throughout the first few months of the pandemic, the “People’s Team”, as they called themselves, transformed their role and outreach from one that was facilitating with the redevelopment process, to instead focus exclusively on addressing the urgent needs and demands of the community. As we will discuss, this was mainly due to the leadership and ingenuity of CNI staff and community residents who organized around the immediate needs and demands of the community in the context of extreme crisis and uncertainty created by the pandemic and lockdown. Similarly, we examine how the social and spatial organization of the houses and Sun Valley community itself were utilized to facilitate opportunities for implementing collective care practices under lockdown that were spearheaded by the People’s Team, and later continued by community residents. As we will argue, these and other care practices that emerged from the Covid-19 lockdown, allowed for the Sun Valley community to become a site of emergence of collective, community-based care practices that started inside the house and home, and that increasingly moved out into the neighborhood and community spaces in the initial months of the COVID-19 pandemic and lockdown. Methods In May 2020, 17 Sun Valley households were provided with cameras and asked to capture images of the resident’s daily lives and experiences under lockdown. One of our team was already conducting research on air quality in the neighborhood and suggested we do a photo-elicitation project on residents’ experiences during the COVID-19 lockdown. We were able to recruit individuals with the help of the CNI manager. She provided us names of residents interested in participating and we then contacted them. The participants were quite diverse, ranging from young adults to senior citizens, and included both women and men. They also included refugees, young and old, single parents, and family members who were disabled. All interviewees lived with other family members except for one retiree who lived alone. Through phone calls and a final video interview, we conducted informal and semi-structured interviews with 17 households and the CNI manager between April and July 2020. Bi-weekly phone calls allowed us to create a relationship and/or rapport with participants. A final zoom interview was used to jointly review the photographs with participants. The use of photography and technology were largely based on the limitations posed by the pandemic and the inability to physically be with residents or in the neighborhood. Residents documented different aspects of their daily lives during the lockdown, first taking photos inside their homes and later, outside. Many of the documented activities revolved around family, food, planting, celebrations, and play, and highlight how Sun Valley residents engaged in care practices to manage the lockdown and the lingering uncertainty of subsequent months (see: Auerbach et al., 2021). These methods, and the data they generated, follow recent calls for geographers to turn to the community’s geographical knowledge and to value ‘epistemologies and ontologies of resilience’ (Eaves and Al-Hindi, 2020: 134). In the next sections we discuss some of the different care practices and infrastructures that were utilized to provide support, comfort, and relief to the Sun Valley residents. In some cases, these were initiatives started by CNI and the People’s Team. In others, the residents were able to use certain resources and spaces to help them get by during the first few months of the pandemic. When examined together, along with the spatial organization of the community it is possible to begin to identify the ways in which Sun Valley residents were able to benefit from and engage in care practices and infrastructures in a time of extreme uncertainty and crisis. Sun Valley, COVID-19 and the people’s team “The families have to stay home” In some ways Sun Valley was in a unique position to address COVID-19, due to the presence of the Choice Neighborhood Initiative (CNI), originally there to facilitate the relocation process for Sun Valley residents from a traditional two-story public housing community to a mixed-income apartment building across the street. Located in the community and part of a more holistic transition from traditional public housing to the Section 8 voucher system, CNI was charged with providing information and guidance to residents during the transition process, and to mitigate any issues or concerns that arose. Once the COVID-19 pandemic and lockdown started, the CNI leadership quickly changed their focus to facilitate the gathering and distributing of provisions and resources to all of the residents. We should note, these were not policy decisions made by officials and then passed on to the CNI leadership. Instead, these decisions were made by individuals who recognized their positionality and power and were able to refocus those resources to address the immediate needs of the community. In an interview with Sarah 1 , the manager of CNI, she explained,There was no framework for any of it. It was literally straight into emergency mode. These families have to stay home, and these are the most vulnerable families. And buses are shut down. I mean, it’s just these are the families that have to stay home. We can’t have an outbreak in a complex like this. And these are the families where it is the hardest for them to stay home. So [we did] anything we could do to help them with that. (In-person interview, 06/15/2020) Although employing some of the organizational resources, the CNI leadership, along with the community members hired by the program, re-established themselves as the “People’s Team,” and started to put into place necessary infrastructure that would help families get through the first few months of the lockdown. As Sarah explains, “Food was the first priority and we partnered with metro ministries to [receive food] every 3 weeks.” The People’s Team would receive a 60-pound box of food containing meat and cheese that they would deliver to the doors of the families. They also partnered with Work Force options for Women, so that every Tuesday through Friday they picked up 250 frozen meals that were also delivered, so that “…if people really needed to stay at home and wanted to, they could do that.” Food and cleaning supplies Food and food preparation were an important aspect of care during the initial months of the lockdown. In addition to the People’s Team ensuring that each family received enough food during lockdown, many of the residents discussed creating new dishes with family members and trying new foods. Refugee women documented making food from their native countries, or preparing food with their children (Auerbach et al., 2021). In one instance, a neighborhood pizza restaurant delivered ingredients to make pizza from scratch. In another, one woman shared photos of herself preparing crab legs for a Facebook competition with friends. Once food provisions and other basic resources were secured, and as the lockdown continued, the People’s Team brainstormed and asked residents about the next phase of urgent needs and demands of the community. This included putting together and distributing wellness kits, which included stress balls, cleaning products, toilet paper, bleach crystals and journals (Figure 2). Some of the residents were given material and taught to make masks, which were then included in the wellness kits. For some residents who lost their jobs during the pandemic, the making of masks, allowed them to stay busy at a very stressful and uncertain time.Figure 2. Cleaning supplies and other staples for Sun Valley residents. Bringing people together outdoors Later, the People’s Team noticed that families were staying indoors, even as the weather became warmer and there was more knowledge about how the virus could be transmitted. They purchased school supplies, materials to make slime, bubbles, and chalk to draw art on the neighborhood sidewalks that wound around the homes (Figure 3). There was a COVID-19 time-capsule activity that included some of the chalk art messages that contained words like “hope” and “stay safe,” public messages that were left for everyone to see. In addition to these materials and acts of play, the People’s Team set up a mobile field day, organizing activities around the family pods that neighbors had started. Each day the Team went to different areas of the community and set up activities and games for the children there. All of these activities helped to pull residents out of their homes and to safely gather with some of their neighbors.Figure 3. Chalking drawings. These initiatives started by the People’s Team can be understood as shadow infrastructures of care: community-based infrastructures of care that emerged to address the urgency of the COVID-19 crisis and lockdown for the residents of Sun Valley. They are also examples of the ways in which housing design and house/home were able to facilitate care practices and infrastructures inside the home, and later, when the weather changed, and people felt more comfortable, shared yards, sidewalks and doorways were used as spaces/infrastructures of community care. The People’s Team and residents were able to organize collective care that moved from the interior of the home to the exterior, at a time when people were tired of staying inside, but still somewhat wary of the risks of venturing out. Mutual care Working with community members and talking with residents about their needs and concerns, the People’s Team was able to create a system or infrastructure that worked to provide residents with resources and activities that supported both physical and mental health and wellness at different stages of the lockdown. Going to the residents, instead of waiting for residents to pick up supplies, particularly at a time of deep uncertainty and fear, helped to develop stronger care connections and infrastructure that were then continued by residents in myriad ways (Figures 4 and 5). The emergence of these infrastructures and connections helped to transform and mitigate the experience of fear and confusion of those first months of lockdown. Sarah described what she perceived were the initial feelings among Sun Valley residents, “The first word I would say is scared. Just the same fear that everyone felt, but they were trapped. Trapped is kind of a true word. Because moving to online school and a lot of families not having the wifi or bandwidth that they needed to access [everything]. And activities! What financial means do they have to purchase all of the materials?” (In person interview, 06/15/20).Figure 4. Unpacking food to be delivered to Sun Valley residents. Figure 5. Delivering snacks to residents. Interestingly, in interviews with Sarah and other members of the People’s Team, although they describe their work as providing for the community through food boxes, wellness kits and activities, many described feelings of great satisfaction and an opening up to the community in ways they had not done in the past. For example, when asked, how do you think (these initiatives) shaped the community throughout the pandemic? Sarah responded,“If it did half of what it did for me, then it made a difference. I think of some of our people who are here, who are senior, disabled or are families from other countries that don’t have other contacts or ways to connect to people. Just that knock on a door every few days or someone bringing you food, or someone thinking of you at least. I hope that they felt that. Because every day we were in here on our whiteboard, wondering what can we do, what do people need?” (In person interview, 06/15/20) Mikaela, a resident member of the People’s Team who had worked at a 7–11 before being hired by CNI discussed how, before starting her current job “I was nervous being in this area, but now it is ok, I see Sun Valley as a community, as a neighborhood.” Mikaela described her work during COVID by saying, “I love it. It is definitely something that I’ve always wanted. Me, Cole, Sarah and Emily were talking about how it just makes us feel good when we are going around. We’ve been delivering care kits and wellness kits to the residents, and everybody is so thankful for just the little bit that we can do for them, it just fills my heart. And I’m getting paid for it, so I couldn’t ask for anything better basically.” (Zoom interview, 05/23/20) Like Sarah, Mikaela described the work that she was doing for her community and the ways in which she was able to receive a deep sense of satisfaction from providing for others and contributing to a sense of community. These testimonies are representative of the ways in which alternative infrastructures of care are deeply relational and interdependent. By providing care, both Sarah and Mikaela felt they were engaging in transformative practices that contributed to community well-being as well as to their own, personal well-being. These are some of the mundane and everyday encounters and practices that perhaps offer glimpses of social transformation and resiliency that scholars suggest alternative infrastructures of care can provide. At the same time, by going to people’s houses and homes to deliver food, wellness kits, and other materials, the People’s Team helped to alleviate some of the desperation of families who were “trapped” inside. Art supplies and school materials, kits that included journals and stress balls, acknowledged the emotional toll of the lockdown by providing tools to help relieve the anxiety, fear and frustration of those first days and weeks of the pandemic. Many of these actions may not have had the impact the People’s Team had hoped. Additionally, there were a host of need and demands that the People’s Team could simply not address. These included keeping students engaged in school, stable internet access so that children could learn well, among others. Nevertheless, the leadership of the People’s Team provided an alternative infrastructure of care that started by meeting residents in their house and home and slowly moved outward into the community. As we discuss in the next section, although the People’s Team was instrumental in creating this infrastructure, residents also built on these connections and developed their own care practices and encounters, finding ways to utilize these and other practices and resources in order to maintain some sense of hope and well-being during lock-down. The material and the mundane of care infrastructures Our interviews with residents, where they discussed the photographs they took during lockdown, highlighted many different activities like cooking food, playing inside and outside, watching television, and taking walks outside as the weather improved. Many of the photos reflected the boredom and mundanity of life under lockdown and the ways in which families tried to remain resilient amidst uncertainty. This section discusses some of the activities that neighbors and family members engaged in together, particularly in the context of planting seeds and growing vegetables and flowers. As we discuss, planting and gardening became an important project for many families. When we asked how they started, some residents seemed unsure from where they had received the seeds and soil to be able to plant. Nevertheless, for many it became a vital activity shared with neighbors and family members. Gardening and growing as care Sun Valley has a community Grow Garden where many different vegetables are grown and sold at the weekly neighborhood farmer’s market. Many of the vegetables that are grown, are at the request of residents, many who come from different parts of the world. The Grow Garden became an important community site for residents to enjoy, particularly because the People’s Team and other community members used it as a space to put up colorful and positive messages for neighbors to read and enjoy when they walked past. Yet, it was also residents’ personal gardens and plots that provided a lot of pride and sustenance for residents (Figures 6 and 7). This became clear in the ways that residents discussed their role and excitement in creating spaces of growth and care.Figure 6. Picture of indoor plants to show how they have grown. Figure 7. Neighbors working together to prepare outdoor gardens. Alana, a young mother with a 12-year-old son with ADHD, and a husband who had been furloughed from his job, showed me photos of her and a neighbor’s husband putting up fencing around their small garden plots in the front of their homes. Normally quite flat in the way she spoke and discussed her life, Alana’s voice became animated when she talked about the tomato plants that they had saved, and how they had wanted to plant strawberries, but it was too late in the season.“We went and bought the fencing stuff so the squirrels and the kids would stay out of [our gardens]. The big pot is peppers. The small one was catnip, but I had to change it out so right now the tomato plant is in it. Me and Kaela (a neighbor) got a whole bunch [of tomato plants] from the Dream Center. Theirs weren’t surviving, so we tried to survive their plants. I have radishes and rhubarb in the garden now!” (Zoom interview, 06/02/20) Dalia, another resident of Sun Valley, who lives with her sisters and parents in a home that is too small for all of them, also took photos of the many plants they had accumulated during COVID. She explained, “When COVID started we started planting a lot. I think everyone in the house has their own plant. My mom is growing an aloe vera, my sister and stepdad are growing chilis, [my sister] calls her plant, Felipe! My brother is growing flowers, my other sister is growing little plants. So, we just started planting more and more.” When asked when and why they had started planting, she said, “I think it was COVID, we just did it to keep busy… And now that COVID is kind of dying down, we are planting a lot more. So, I guess that was a good thing too!” (Zoom interview, 06/27/20) Many families planted sunflower seeds and talked about how tall they grew. June and her son Luke, a tall red-headed high school freshman, who together with his mom discussed the dozens of photos they had taken for the project, talked about how they both enjoyed watching the sunflower seeds they had planted get so big. Luke exclaimed, “One of them is as tall as my mom!” I asked them if they had planted things in other years, but like all the other families, they said no, it was the first time. When we asked them why they had decided to grow things that year, Luke responded, “It’s exciting to see something new, kind of not created, but gave life to”. For residents of Sun Valley, planting and growing became an important form of care work that was used as a form of nourishment, self-care, and as Luke explained, life giving. Carabelli and Sharma (2021) have reflected on these new-found practices and relations with plants in the context of the pandemic through a focus on “how social isolation made many humans look at their houseplants anew, shaping new affective bonds and caring practices that came with the realization that plants care for us too, and helped us to navigate the uncertainties produced by the current pandemic” (np). These care acts illustrate the multiple ways and scales that care functions and flows, and echo similar sentiments felt by the People’s Team. The act of growing things was deeply profound for the residents. It was not simply watching things grow, which is what residents did at the Grow Garden, but rather, it was the agency they had in giving life that seemed to give residents a deep sense of satisfaction and well-being. As such, like the infrastructure of care and care-work that the People’s Team provided to residents and that were potentially transformative for both care giver and care receiver, these relational care practices around gardening and planting had a similar impact on residents themselves and the community, as they were shared with neighbors and family members and created bonds and a sense of agency and hope. Discussion and conclusion The COVID-19 pandemic and lockdown created extreme hardship for families and communities around the world, especially those already living in extreme poverty and precarity. Yet, the pandemic has also become a moment from which new practices, infrastructures and connections rapidly emerged, often in random and spontaneous ways. In the case of Sun Valley, a series of decisions, socio-spatial conditions, and institutional structures created opportunities for the development of infrastructures and services that were able to address the urgent and short term needs of the residents. These included the capacity and the initiative of the CNI manager to redirect manpower and resources that addressed residents’ immediate needs. The existence of the People’s Team, which included Sun Valley residents, helped to disseminate information about available resources and what people’s needs were. The spatial organization of the community, with its individual entrances that opened out to large lawns with sidewalks also facilitated the delivery of products and the ability to engage with residents or for residents to engage with each other outside or from a distance. Similarly, although residents initially remained inside their homes, they had some access to the outdoors, just by opening windows and doors. Later, when the weather got warmer and there was a better understanding of the virus and its transmission, the People’s Team used the yards, walkways, and home spaces to create gathering spaces, allowing people to remain close to home, but to also begin to feel safer to spend time outdoors with neighbors. In all of these care practices, the house/home became a central site of security and protection, with the CNI manager and the People’s Team going to people’s homes so that they could feel safe, but not isolated. At the same time, although each family remained in their homes, the proximity and the care infrastructures that emerged created a sense of collective solidarity and care across the community. Residents were able to benefit from these care infrastructures in numerous ways. Those working on the People’s Team described feelings of satisfaction, trust, and pride in their work and for other residents. Others felt similar sensations when planting seeds and watching them grow. Neighbors worked together to care for their plots outside of their front doorways. These activities moved out into the yards and the Grow Garden, where community games and activities were organized and positive messages were posted, for passersby to read and experience. In essence, the utilization of these neighborhood and communal spaces for the community as a whole, represent a much more collective notion of house/home that emerged through the spontaneous, urgent, and mundane activities of providing food, cleaning products, in-door activities, and later, out-door games and diversion to help reunite neighbors. Similar to Power et al.’s (2022) work on shadow care infrastructures, Sun Valley illustrates “care infrastructures that sustain and organize the care practices of people living in poverty…but that are not always readily visible within dominant welfare discourse, policy and research” (p. 1166). Sun Valley offers valuable lessons when thinking about the role of the house/home and the social and spatial development of care practices and infrastructures. Although we recognize that the house/home is not simply a place where care practices will automatically take place, it was and remains a central space from which individual and community care practices can begin. In this way, our conceptualization of house/home is inclusive of the neighborhood, the community, and those agencies and actors involved. As we demonstrate, the layout of the neighborhood, the connections between individuals, and the resources available all contributed to mitigating the negative impacts of the COVID-19 lockdown, as well producing care infrastructures from which other care practices, including self and mutual care also emerged. We acknowledge that without CNI and the initiative of the manager and the People’s Team, all of whom were leaders in the community, it is unclear if these same care infrastructures would have emerged in these same collective and communal ways. Nevertheless, we argue that this demonstrates the importance of stable community leadership and institutions in place that can facilitate and direct infrastructures in ways that are not simply top-down, or individualized practices but that facilitate community care through mutuality and solidarity. Finally, we should note that this particular COVID-19 moment in Sun Valley was short-lived, especially because the redevelopment eventually resumed, and the families of Sun Valley have now moved to the new building, or they have left the community altogether. Nevertheless, we argue that the alternative infrastructures of care that were developed during the first few months of the COVID-19 lockdown are an example of the potential of care infrastructures to build more collective, just, and care-full societies, and that more empirical research should be done, not just in the context of COVID-19, but in multiple contexts of crisis and everyday structures and practices. Ultimately, as our societies become increasingly fraught and complex due to political divisions, climate change, and government failures, perhaps it is those invisible practices and infrastructures happening at the microscales of the house/home, and in local communities around the world in which people are engaging in mundane, invisible, and interdependent relationships and care practices, that should be revealed, taken seriously, and further researched. Solange Muñoz is an Assistant Professor in the Department of Geography and Sustainability at the University of Tennessee, Knoxville.Through qualitative and ethnographic methods, her research addresses social dimensions of urban development in US and Latin American cities by examining how struggles for access to housing and to remain in the city are routinely lived and experienced by poor and minority urban populations. Jordin Clark is an Assistant Professor at West Chester University. Through interdisciplinary collaboration she publishes on community engagement within urban development projects. Her recent publications are on development projects during the COVID crisis in Social and Cultural Geography and Urban Geography. Jeremy Auerbach is an Ad Astra Assistant Professor of Geography at University College Dublin. He is a quantitative methodologist and participatory action researcher who works withurban community organizations who face housing and environmental justice issues. Lily Hardwig is a graduate of the University of Tennessee and currently works at Centro Hispano, a nonprofit organization serving East Tennessee’s Latino community. Lily’s research interests include immigration, care infrastructures, cultural exchange through languages, and narratives through storytelling. In her previous work, Lily has explored the “Infrastructures of Trust and Care in Latin American Migrant Communities” through qualitative and ethnographic research. ORCID iD Solange Muñoz https://orcid.org/0000-0001-5353-184X Note The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. 1. All names have been changed to maintain anonymity. ==== Refs References Alam A Houston D (2020) Rethinking care as alternate infrastructure. Cities 100 : 102662–102710. Altman I Werner C (eds), (1985) Home Environments. New York, NY: Plenum Press. Amin A (2014) Lively infrastructure. Theory, Culture and Society 31 (7-8 ): 137–161. Auerbach J Clark J Muñoz S (2021) Under quarantine in a city project: stories of fear, family, food, and community. Community and Society. Bristol, UK: Bristol University Press, 1: pp. 117–128. Berlant L (2016) The commons: infrastructures for troubling times. Environment and Planning D: Society and Space 34 (3 ): 393–419. Blunt K Dowling RM (2006). Home. Oxfordshire, UK: Routledge. Carabelli G Sharma P (2021) Image-maker in residence: care for plants [online]. The Sociological Review Magazine. Lancaster, UK: The Sociological Review Foundation, Lancaster University. DOI: 10.51428/tsr.febc1028 10.51428/tsr.febc1028 Chen CYC Byrne E Vélez T (2021) Impact of the 2020 pandemic of COVID-19 on families with school-aged children in the United States: roles of income level and race. Journal of Family Issues 43 : 719–740. Clark J Muñoz S Auerbach J (2022) When top-down infrastructures fail: spaces and practices of care and community under COVID-19. Social and Cultural Geography 24 : 542–562. Dovey K (1985) Homes and homelessness. In: Altman I Werner C (eds), Home Environments. New York, NY: Plenum Press. Easthope H Power E Rogers D , et al. (2020) Thinking relationally about housing and home. Housing Studies 35 (9 ): 1493–1500. Eaves L Al-Hindi KF (2020) Intersectional geographies and COVID-19. Dialogues in Human Geography 10 (2 ): 132–136. DOI: 10.1177/2043820620935247 10.1177/2043820620935247 Goetz EG (2012) The transformation of public housing policy, 1985–2011. Journal of the American Planning Association 78 : 452–463. Graham S Marvin S (2001) Splintering Urbanism: Networked Infrastructures, Technological Mobilities and the Urban Condition. NewYork, NY: Routledge. Graham S McFarlane C (2014) Introduction. In: Graham S McFarlane C (eds), Infrastructural Lives: Urban Infrastructure in Context. New York, NY: Routledge, pp. 1–15. Griego T Greene S (2020). Coloradans struggle with mental health in state ill-prepared to help them. Denver Post. Denver, Colorado: The Denver Post, LLC. https://www.denverpost.com/2020/12/06/covid-mental-health-struggle-colorado/ Hamilton S (2021) What Is Infrastructure? Congress, Administration Debate Definitions. Alexandria, VA: National Society of Professional Engineers. https://www.nspe.org/resources/pe-magazine/summer-2021/what-infrastructure-congress-administration-debate-definitions Handel A (2019) What’s in a home? toward a critical theory of housing/dwelling. Environment and Planning C: Politics and Space 37 (6 ): 1045–1062. Hobart HIJK Kneese T (2020) Radical care survival strategies for uncertain times. Social Text 38 (1 (142 ): 1–16. Hooks B (1990) Homeplace (A Site of Resistance) Yearning: Race, Gender and Cultural Politics. Boston, MA: South End Press, pp. 41-49. Horn MB (2021) The rapid rise of pandemic pods: will the parent response to Covid-19 lead to lasting change? Education Next 21 (1 ): 93–96. Jackson M (2018) “Sun Valley could be Denver’s next big thing” westworld. https://www.westword.com/news/sun-valley-one-of-denvers-oldest-and-poorest-areas-could-be-the-next-big-thing-9981911 (accessed 30 January 2022). Lancione M (2020) Radical housing: on the politics of dwelling as difference. International Journal of Housing Policy 20 (2 ): 273–289. Landivar LC Ruppanner L Scarborough WJ , et al. (2020) Early signs indicate that COVID-19 is exacerbating gender inequality in the labor force. Socius: Sociological Research for a Dynamic World 6 : 2378023120947997.34192138 Lawson V (2007) Geographies of care and responsibility. Annals of the Association of American Geographers 97 (1 ): 1–11. Lopez PJ Neely AH (2021) Fundamentally uncaring: the differential multi-scalar impacts of COVID-19 in the US. Social Science and Medicine (1982) 272 : 113707.33517126 Lynch K (2022) Care and Capitalism: Why Affective Equality Matters for Social Justice. Cambridge, UK: Polity Press. Massarenti M (2020) How Covid-19 led to a# rentstrike and what it can teach us about online organizing. Interface: A Journal on Social Movements 12 (1 ): 339–346. McCrary J Sanga S (2021) The impact of the coronavirus lockdown on domestic violence. American Law and Economics Review 23 (1 ): 137–163. Mendes L (2020) How can we quarantine without a home? responses of activism and urban social movements in times of COVID‐19 pandemic crisis in Lisbon. Tijdschrift Voor Economische en Sociale Geografie 111 (3 ): 318–332.32834145 Muñoz S (2021) Who they have become: peruvian migrants and memories of home, motherhood and violence. Gender, Place and Culture 29 (4 ): 547–567. Nicholson K (2020). Food insecurity on the rise as COVID-19 cases surge and economy sputters. Denver Post. Denver, Colorado: The Denver Post, LLC. https://www.denverpost.com/2020/11/30/colorado-food-insecurity-covid-cases-surge-economy/ Power ER (2019) Assembling the capacity to care: caring-with precarious housing. Transactions of the Institute of British Geographers 44 (4 ): 763–777. Power ER Bergan TL (2019) Care and resistance to neoliberal reform in social housing. Housing, Theory and Society 36 (4 ): 426–447. Power ER Mee KJ (2020) Housing: an infrastructure of care. Housing Studies 35 (3 ): 484–505. DOI: 10.1080/02673037.2019.1612038 Power ER Wiesel I Mitchell E , et al. (2022) Shadow care infrastructures: sustaining life in post-welfare cities. Progress in Human Geography 46 (5 ): 1165–1184. Root Policy Research (2020) State of Colorado: findings of the special eviction prevention task force. Root Policy Research. Special Eviction Prevention Task Force Report. https://www.coloradorpm.com/wp-content/uploads/2020/10/CO_Special_Eviction_Prevention_Task_Force_Report_.pdf Silver C Hyman DA (2020) COVID-19: A Case Study of Government Failure. Washington, DC: Cato Institute, Pandemics and Policy. Simone AMM (2004) People as Infrastructure: intersecting fragments in Johannesburg. Public Culture 16 (3 ): 407–429. https://www.muse.jhu.edu/article/173743 Svaldi A. (2021) Colorado paid a heavier price economically in its fight against COVID-19 than other states in 2020. The Denver Post. Denver, Colorado: The Denver Post, LLC. https://www.denverpost.com/2021/02/07/colorado-economy-covid-price-restaurants-businesses-unemployment/ Tai DBG Shah A Doubeni CA , et al. (2021) The disproportionate impact of COVID-19 on racial and ethnic minorities in the United States. Clinical Infectious Diseases: An Official Publication of the Infectious Diseases Society of America 72 (4 ): 703–706.32562416 Thomason B Macias‐Alonso I (2020) COVID‐19 and raising the value of care. Gender, Work and Organization 27 (5 ): 705–708.32837006 Tronto J (2013) Caring Democracy: Markets, Equality and Justice. New York, NY: New York University Press. Tronto JC (1993) Moral Boundaries: A Political Argument for an Ethics of Care. New York, NY: Routledge. US Bureau of Labor Statistics (2020) Labor Force Statistics from the Current Population Survey. Washington, DC: US Bureau of Labor Statistics. Retrieved from https://www.bls.gov/cps/cpsaat11.htm US Department of Housing and Urban Development (HUD) (2021) Choice Neighborhoods. Washington, DC: US Department of Housing and Urban Development (HUD). Vilenica A McElroy E Ferreri M RHJ Editorial Collective , et al. , (2020) Covid-19 and housing struggles: the (re) makings of austerity, disaster capitalism, and the no return to normal. Radical Housing Journal 2 (1 ): 9–28. Wardhaugh J (1999) The unaccommodated woman: home, homelessness and identity. The Sociological Review 47 (1 ): 91–109. Williams MJ (2017) Care-full justice in the city. Antipode 49 (3 ): 821–839.
PMC010xxxxxx/PMC10291220.txt	==== Front Games Cult Games Cult GAC spgac Games and Culture 1555-4120 1555-4139 SAGE Publications Sage CA: Los Angeles, CA 10.1177/15554120231182802 10.1177_15554120231182802 Article Mutants and Zombies Everywhere! Or Villains, Violence, and Selfishness: Questions of Humanity in the Post-apocalyptic (Pandemic) Video Game https://orcid.org/0000-0003-0745-4278 Wintle Phil 1 1 57411 d'Overbroeck's College , Media Studies, Oxford, United Kingdom of Great Britain and Northern Ireland Phil Wintle, d'Overbroeck's College, Media Studies, Oxford OX2 6JX, United Kingdom of Great Britain and Northern Ireland. Email: p.wintle@outlook.com 22 6 2023 22 6 2023 15554120231182802© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. The post-apocalyptic environment has been popular in video games for many decades—from nuclear fallout to alien invasions, there is a fascination with the decline of mankind. This article looks, in light of the Covid-19 outbreak, at the depiction of failing and inhuman worlds in video games about pandemics. The environments in games such as Bioshock (2007), Left 4 Dead (2008, 2009), and The Last of Us (2013), are unflinchingly cruel, often depicting humans (and the player) as selfish entities seeking only self-survival. These environments are destructively violent, and, although the player is often surrounded by ‘mutants', these worlds are also oppressively lonely. This article looks at the fears reflected in pre-Covid-19 pandemic video games and what they say about our world, offering a retrospective view now we are living through a global pandemic perhaps not so different from those found in these games. pandemic violence villains humanity selfishness social anxiety media edited-statecorrected-proof typesetterts19 ==== Body pmcMany video games explore post-apocalyptic environments, from zombie games such as The Resident Evil series (1996–present) to horrors in post-nuclear war settings such as the Metro series (2010–2019), or, perhaps more radically, with demons spreading from a portal to Hell as with Doom (1993–present). This article looks specifically at the post-apocalyptic pandemic game, a popular genre even before the outbreak of Covid-19. Somewhat naturally while living through a pandemic, one's curiousness was prompted to explore the depiction of pandemics in the media, specifically video games. Particularly, this article explores how human interaction is depicted in pandemic games before Covid-19, and what these games suggest about humanity and social anxiety in the contemporary age. Pandemic games, including the games of this study, are predominantly of the horror genre, as may be expected considering their subject. Horror is a particularly popular genre in video games as the genre often utilizes recognizable locations but offers a twist to make them fantastical, special, and engaging (Rouse, 2009, pp. 16–17). This is not necessarily unique to pandemic games, but it is a theme that is particularly noticeable in them as the world slips further from recognizable to fantastical and continues to disintegrate around the characters and players. Horror also has a long history of reflecting social anxiety, offering allegorical explorations of our worlds that may provide a thought-provoking commentary on contemporary society and fears (Wells, 2000, pp. 20–21). Looking at the aforementioned disintegrating worlds of the pandemic games, the reflected social anxiety is clear: division and social collapse. This article looks at how this fear is presented through violence and loneliness in the pandemic game, while also briefly considering whether this reflection of humanity is reasonable now we have lived through a global pandemic. The games of this study, namely Bioshock (2007), Left 4 Dead (2008) and Left 4 Dead 2 (2009), and The Last of Us (2013), each offers different gameplay styles, tones, narratives, and depictions of human behavior. These games have partly been selected because of their differences and are representative of the range in the genre. As shall be discussed, the pandemics in these games all manifest in different forms: The Last of Us sees a natural-viral infection; Left 4 Dead sees a man-made infection; and Bioshock a man-made psychological and physical disease. The causes of pandemics differ, and the tone of the gameplay also varies massively. That said, these games also present common elements found in pandemic games—for instance, each game sees victims mutated to become “Other” to the player. More importantly, each game offers reflections of contemporary social anxiety as they present a crumbling world with little hope, and each game, despite their differences, depicts a commonality with excessive violence and prevailing themes of selfishness and loneliness. Sub-chapters of this article look at the purpose of these themes, assessing why developers choose to depict a rather morbid view of humanity. It is fascinating, and slightly telling, that games sharing only the basic premise of spreading illness present similar views of humanity and society. First, however, I will present a brief overview of each game, focusing on their narratives and noting their gameplay styles. The Pandemic Game In Left 4 Dead and Left 4 Dead 2, the player takes a position in a party of four (a different group of survivors in each game) who work together to fight hordes of infected, zombie-like creatures. The infection is caused by a pathogen, simply called “Green Flu,” which causes aggression and mutation in victims. It is suggested that this mutation, as with many “zombie” video games and films, is also spread through bites—although survivors are more likely to be killed by the violence of the infected long before this. The survivors look for rescue from the American military, who are doing their utmost to stop the spread of the infection, although there are also some hints that the military has murdered asymptomatic carriers of the infection. Realistically, however, there is little narrative to these games, rather the focus of the games is to survive in differently themed environments. Indeed, each chapter starts with a film-style poster and also has end credits, suggesting a rather flippant tone. Unlike the other games discussed in this article, morality does not play a key role in the narrative, rather, as the film-style suggests, there is a simple cathartic fun to the violence akin to exploitation films. This focus on gameplay over narrative is somewhat justified by the games’ focus on multiplayer action, although the games still have plots that are perfectly playable in single-player. That said, Valve and Turtle Rock Studios (then called Valve South) have noticeably developed an oppressive and bleak environment for their games with the color scheme, score, and monster design, building a chilling ambiance. Elements of horror include “special zombies” such as “The Witch” who weeps like a child but lashes out if you approach, and the use of the AI director which senses how you are progressing and can change the layout, music, lighting, and difficulty of each chapter. This means unlike most video games you cannot play the chapter again with the benefit of pattern recognition leading to very little respite. This lack of control is not only engaging but also builds a rather oppressive tone. To defeat the special zombies you need to rely on other characters (whether human or AI controlled), this sense of teamwork is rewarding, yet, despite the cooperative nature of the game the sheer number of infected faced by the players highlights their isolation as an insignificant few against the hordes. The oppressive and bleak atmosphere is also furthered through ever-present and often morbid graffiti, with messages of “no hope” and “no cure” being common, and other messages suggesting the struggles of other survivors. Although it should be of note that other pieces of graffiti do offer some form of comic relief (most amusingly one note under a heartfelt but lengthy message of love and loss reads: “I'm hiding from zombies and reading this was the most terrible thing to happen to me today”). Oddly, these messages from past survivors oxymoronically present another layer of isolation—though there is a suggestion of other survivors in the game world, the lack of contact with friendly AI presents the graffiti as relics of people now long dead. In short, the Left 4 Dead games develop a sense of tension through environmental design that is exasperated by a sense of isolation, but there is an enjoyable release of this tension via the combat mechanics. The Last of Us is initially set in the present day at the outbreak of a deadly fungal-based infection that turns victims into monsters. The fungus warps the brain of a victim who then goes through different stages of infection. Firstly, victims become “runners” who, in hordes, attempt to bite and kill uninfected people. They then further lose their human qualities until, eventually, the carcass of the infected becomes a breeding ground for fungal spores to continue to spread the pandemic. The gameplay is more nuanced than in Left 4 Dead, involving action, stealth, and a small quantity of survival crafting. The narrative too is more detailed, following Joel (Troy Baker) who loses his daughter Sarah (Hana Hayes) who is shot by a soldier as chaos ensues among the American public and military. Twenty years later, survivors are forced into military-controlled quarantine zones; a rebel group, the Fireflies, resist the authority. Joel and the Fireflies find hope in a young, immune girl, Ellie (Ashley Johnson) whom Joel is tasked with taking to a medical facility in search of a cure. The path Joel and Ellie takes is dangerous, as many infected block their route, but, unlike in Left 4 Dead, there are also several human groups that attack Joel and Ellie. As shall be discussed, with this division of survivors, there is a deeper exploration of society, humanity, and love—indeed, this stronger and more potent narrative has been adapted into an HBO television series. In both Left 4 Dead and TheLast of Us, the player fights infected enemies, they are not the living dead (or zombies), but they are certainly zombie-like in their behavior. By “zombie-like” one means that the AI in these games acts in an uncanny, aggressive, and irrational manner. Both games see the infected appear somewhat humanoid, if mutated; they are often found wandering aimlessly but with any hint of player interaction they turn aggressive with the intent to kill. While they appear human, their actions lack awareness; in many ways, these figures act as moving obstacles (particularly in The Last of Us where there is more challenge in the game than just combat with the infected). Looking at the depiction of zombies in video games, Shawn McIntosh notes the “empowerment” that one feels when killing a monster and that zombies offer a “guilt-free” figure to kill as they are “evil,” unhuman, and their murder is usually in self-defense (McIntosh, 2008, pp. 12–13). Although technically not zombies, the infected in Left 4 Dead and The Last of Us are treated in the same vein; hordes are killed for cathartic fun in Left 4 Dead, and little sympathy is depicted toward the infected in The Last of Us. Indeed, in both games, the infected quickly lose any resemblance to humanity and become utterly “Other”—placating the guilt of murdering them. Perhaps this is further exasperated in Left 4 Dead as the player is likely playing with real people online, offering a further detachment between the real figures the player works with, and their AI “Other” enemy. There is little question of humanity or morality in attacking these infected. Bioshock, however, does not present zombie-like enemies that are wholly “Other”—indeed, the enemies of this game are not too far removed from the player. Bioshock, and also Bioshock 2 (2010), is set in the underwater dystopian metropolis of Rapture. Rapture was designed as a world of freedom for intellectuals; however, before the game starts, the utopia falls and becomes a world of greed and social division. Scientific developments seem to improve the environment, as a chemical substance known as ADAM is discovered. It can rewrite genetic material and, when refined into a plasmid, allows people special abilities such as producing electric bolts, or the power of telekinesis. Unfortunately, ADAM is highly addictive, and slowly the population of Rapture transforms into “Splicers” (mostly human creatures with physical and mental deformities that are the most common enemies in these two games). Arguably, the pandemic in Bioshock is an epidemic as it affects one area—Rapture. That said, Rapture is effectively designed as a new world with little to no contact with those above the ocean—as such watching the “new world” of Rapture succumb to illness is akin to the pandemic in The Last of Us. Both games after all raise questions of community, society, and self-preservation above all else. Unlike Left 4 Dead and The Last of Us, the pandemic in Rapture is undoubtedly man-made; science has been pushed too far and caused an uncontrollable drug addiction. Interestingly, there is little difference in tone or effect that this game's pandemic is man-made rather than viral; although there is arguably a greater resonance in the fact that the “infected” of Bioshock is a failing of humankind. Certainly, the Splicers appear more human, thus invoking further feelings of connection and guilt in the narrative. Although Splicers attack the player-protagonist, Jack (Adam Sietz), in their search for ADAM, they still retain some human qualities, such as their physical appearance. They also live in small groups and converse with one another (and even the player). The mutants of Bioshock are not the “guilt-free” zombie figure described by McIntosh. In fact, in the first Bioshock game, the player-character has found Rapture because of a plane crash—we are very much an unwelcome addition to the Splicers’ world. Interestingly, the player defeats the Splicers by using ADAM (and another chemical EVE) to use plasmids and powers as weapons. As such, while uncovering the mysteries of Rapture, the player becomes the very enemy they fight. Or, in other words, the player is also infected by the sickness of Rapture. For some, a game about addiction may not appear to invoke the term “pandemic.” The other two games presenting a viral infection (The Last of Us) and transmittable disease (Left 4 Dead) are more easily identifiable as pandemics in a biological sense. But I have included Bioshock to broaden this understanding of pandemics, which do, in fact, come in other forms. Noncommunicable diseases (NCDs—ranging from illnesses like cancers to addictions) are, in the words of The W.H.O., a growing global “burden” (The W.H.O., n.d). Indeed, NCDs are often referred to as epidemics and/or pandemics in the medical community. It is a different form of pandemic/epidemic, but no less worrying and no less troubling during and after Covid-19 (Sheldon & Wright, 2020). More importantly, if we are to explore the words and feelings invoked by the word “pandemic”—isolation, loneliness, illness, fearfulness, paranoia of others, and selfishness—these are all very present in Bioshock, as they are in Left 4 Dead or The Last of Us. As with Covid-19, or the pandemics in the other games, Bioshock presents a society in the grips of illness and crisis. It does not matter the form of illness (viral, transmittable, psychological and physical), pandemic games make little distinction in narrative or theme whatever the illness. Bioshock, The Last of Us, and Left 4 Dead show a society that cannot cope when challenged, a society that when pressured by illness will collapse. Despite the different illnesses, these games present the same concerns and invoke similar emotions (such as paranoia and loneliness). Like Left 4 Dead and The Last of Us, the world of Bioshock is akin to its prevailing sense of sickness and deterioration. Although the Splicers retain human qualities, their mental and physical states continue to deteriorate, and Rapture itself is portrayed through sickly green hues with constant rolling wind and groaning metal sounds throughout the environment—there is a sense that the world itself is unwell and slowly deteriorating. This sense of sickness is furthered as there is a recurring motif of medicine in the game—the second level of Bioshock takes place in the Medical Pavilion, Health Stations are frequent throughout the map, and there are a number of characters who are doctors. Moreover, dialogue throughout Bioshock (often from audio tapes) presents ADAM addiction as a spreading illness, referring to ADAM as “cancer” that “destroyed our bodies, [and] our minds.” Further lines explain how Rapture “went to hell,” while a rather eerie lullaby sung by a female Splicer details how both “Daddy” and “Mommy” will die, leaving their child (the Splicer then realizes the baby has already passed). Like The Last of Us and Left 4 Dead, the sickness in Rapture is still presented as spreading, prevailing, and getting worst in survivors. The pandemic in Rapture has resulted in division and a collapse of civilization, a desperate need for self-preservation, and unchecked violence. Again, these themes are apparent across all the games of this study—evidently, it is not the cause of the pandemic that is significant, but the results of pandemics on civilization which are presented in similarly bleak manners in each game. In each game, civilization falls into disrepute very quickly. In Bioshock, the world of Rapture was constructed in 1946 and began to fall into ruin by the mid- to late-1950s. This is only a few years before the game takes place in 1960; audio-diaries reveal that at this time there was also a civil war. By the time the protagonist arrives in Rapture there is little society left recognizable. Understandably, most survivors are hostile to the protagonist, although some vestibules of humanity remain—noticeably the player is only in Rapture for a manner of days but some characters such as Dr. Tenenbaum (Anne Bobby) are so desperate for help they rely on the player with little introduction—equally the player-character trusts any figure who does not openly attack him to fight against the illness and isolation of Rapture. The pandemic of The Last of Us begins in 2013, but the main narrative is set in 2033—explaining why so much distrust and division have arisen after 20 years of hardship and disorder. Because of this 20-year span, and the fact that the player controls Joel over these years, there is a greater exploration of community rebuilding, learning to trust others, and even the growth of affection. These themes are not ones explored in Left 4 Dead as the player controls characters only for a manner of days within a month of the initial infection. What The Last of Us and Left 4 Dead do show, however, are worlds that crumble even quicker than Rapture. In The Last of Us, in 2013, American soldiers are seen shooting and killing civilians, including Joel's daughter, in the opening levels Joel ignores the plight of others, and between 2013 and 2014 the Firefly militia is formed and begins to clash with the American military. Similarly, in Left 4 Dead within one month of the infection martial law is declared, and in levels such as “The Park” shootings of uninfected are noted by the protagonists. In short, all three games present an expectation that in times of crisis the idea of community and civilization will collapse with surprising ease, and in the case of Bioshock and The Last of Us, open hostility, selfishness, and violence are also inevitable almost instantaneously. Whether because of man-made or viral illnesses, in each game the world is now one of suffering. In Left 4 Dead, survivors fight endlessly with little hope; The Last of Us sees families torn apart by disease; and Bioshock sees humanity reduced by addiction. And yet in each game, the player is expected, to various extents, to further this dystopia through violence. Violence in the Pandemic Video Game Complaints and controversies about violent video games have a long history. However, in recent years, violence in video games has become more common and accepted, though there are still controversies. For instance, in Grand Theft Auto V (2013), the mission “By the Book” sees players torture a man in an interrogation sequence. Although a cut-scene goes on to criticize torture, the mission has split opinion (see: Bramwell, 2013; MacDonald, 2013; Petit, 2014). The discussion around the mission drew comparisons to the “No Russian” mission in Call of Duty: Modern Warfare 2 (2009), in which the player, controlling a CIA agent, murders Russian civilians as part of a mass shooting to infiltrate a terrorist group. Both these scenes are shocking for their violence; more specifically, they are controversial because of their realism. These are not games that see the player run over unrealistic figures in Death Race, or fight to the death graphically as with other controversial games like Mortal Kombat (1992). The violence in Grand Theft Auto V and Call of Duty: Modern Warfare 2 is controversial because it plays upon contemporary fears and reality—in recent years, the public has seen governments abuse their powers using torture, for example, during the Iraq War, and “No Russian” somewhat echoes previous terrorist incidences such as the Beslan massacre (2004) and the Columbine High School massacre (1999). There are other such games that offer a post-modern depiction of violence—Spec Ops: The Line (2012), for instance, is ostensibly a third-person war shooter, yet it forces players to question the morality of violence and war. The violence of these games is not noteworthy, but their reflection of our world is. Such games have moved away from a morally black-and-white dichotomy and are instead morally gray—a trait shared by the pandemic games of this study. This article does not look at the effects of violence upon the player, a topic explored well elsewhere, but instead explores how violence is used to create a narrative that challenges our view of humanity and offers a morally gray reflection of society and civilization. As with the games discussed above, the pandemic games are shocking not because of their violence, but because of what they say about humanity. For the vast majority of The Last of Us the player controls Joel who is protecting Ellie (who is still a young girl). There is a father–daughter-like bond between the two, which adds a layer of protectiveness as the player battles the infected and other human enemies. The infected in the game effectively build tension and fear; a fine example of this is in the chapter “Pittsburgh” where the player finds themself in the basement of a flooded hotel with no lights and various types of infected lurking. There are many online forums that offer lengthy discussions about the unease players have felt during this particular mission. The infected in this level are present as wholly “Other”, they are not presented as human or with human qualities, and it is never questioned whether killing the infected is wrong. In fact, the greater horror in the game is not the infected, but rather “normal” humankind itself. In the game's prologue, Joel's relatively tranquil family life is torn asunder as the infected attack their neighborhood, nearly killing Joel; his unharmed daughter Sarah is then intentionally shot by a soldier who is ordered to kill fleeing civilians to control the pandemic. The inhumanity of the scene, from both the infected and military, helps justify the violence of the rest of the game. Equally, in this scene, Joel refuses to stop for a family who have a child—instantly, the game is evoking a need for callous individualism. The murder of the infected is needed in the game for one's own survival—but so too is the need to attack or distrust others. The shock of the game is not in the infected, but in the ease and speed with which civilization falls to animal-like self-preservation. Ellie maintains a child-like innocence in the first half of the game. For instance, she gets disappointed when a comic she finds has a cliff-hanger and, much to Joel's disapproval, wonders how men can walk around after seeing male nudity for the first time in a pornographic magazine. These interludes are essential for building a rapport between player and character; indeed, Katherine Isbister argues that the more human a non-player character's (NPC) characteristics are the greater the player builds a connection and experiences “powerful feelings” (Isbister, 2017, pp. 20–22). This then makes the player experience a disturbing depleted feeling as Ellie's innocence is tainted by the violent world around her. Early in the game, she boasts “If I had a gun, I could help kill some of these fuckers [the infected],” but her first use of a gun in the game is not against the infected, but to kill a man who is attempting to murder Joel in the aforementioned “Pittsburgh” chapter. The shock of her actions, though saving Joel, makes her feel sick. This highlights the moral ambiguity of the game; there is a need to fight and kill other humans to survive, but it is still unnatural and innately wrong, especially when concerning children. Again, this sense of “wrongness” is driven by the player's connection with Ellie and her innocence—this connection to humanity is lost in this bleak world. It is the actions of humans, not the infected, that cause this suffering. The downfall of society is caused by the infected, but they act only as a mirror to present the equally monstrous behavior of the survivors in this pandemic world. Writing about The Last of Us Part II (2020), David Sims suggests that the game forces players to “confront their own brutality” (Sims, 2020). Sims notes that as Ellie has grown, she becomes more brutal in seeking revenge for the death of Joel and beats others to death—an action the player must control and cannot avoid. This, Sims writes, demonstrates how violence begets more violence and leads to unrelenting “bleakness.” As with the Grand Theft Auto V mission “By the Book”, there are perhaps questions of morality about forcing a player into such actions—indeed, Dougherty (2020) even suggests that the game “seemingly does everything in its power to make gamers feel bad about the act of playing it.” Rather than a criticism, however, Dougherty notes that The Last of Us Part II, like a sad film, endures in the memory as a piece of art. The violence in the game presents the protagonist as morally complex; perhaps at times painful to play, but there is a brutal realism because of their bleakness. This, if nothing else, reflects the pain caused by division and violence in society and highlights the lengths to which humanity can (and may) descend. As before, it is shocking, not because it is violent, but because we see the beginnings of division, selfishness, and violence in our world. Ultimately, we fear a breakdown of civilization and through Ellie, with whom the player has built a connection, we ask ourselves what actions are justified for self-preservation and love. This rather powerful, yet depressing, view of humanity is not shared by the Left 4 Dead franchise. Although the games portray an oppressive horror environment, it also presents violence as cathartically fun. The franchise has a noticeable focus on violence and gore; Left 4 Dead 2 was even banned for violence in Australia and Germany, though both countries have now released the game uncut. Unlike The Last of Us, there is little criticism or discussion of narrative or themes in Left 4 Dead. Despite being set in a pandemic world, the monsters the player faces in Left 4 Dead fulfill the traditional role of the horror monster in video games, in that they have little social commentary, offer challenges, build an aesthetic of horror (Stobbart, 2019, p. 144), and simply offer gratuitous catharsis. In Left 4 Dead, the infected remain humanoid only to an extent; the special zombies (such as Spitters, Boomers, and Tanks) appear alien. The main hordes of the infected do look “normal” but their behavior (either attacking en masse or standing still looking at the floor) is so uncanny that they become “Other.” Othering has long played a role in video games; as discussed, until recently video game morality has been rather black-and-white, good versus evil. John Markert notes that people “like the face of evil to be clear and unequivocal” (Markert, 2011, p. 7); this desire for clear black-and-white narratives is evident in video games and partly explains the continued popularity of figures such as monsters and Nazis in video games. Such figures, often simplified in games, offer an unequivocal enemy and a “moral justification for killing” (Hayton, 2012, p. 208). The infected in Left 4 Dead do not present complex personalities or figures to be saved but are dehumanized abstract evils that can be gleefully taken apart. Indeed, even in The Last of Us, there is little sympathy or moral objection to killing the infected—despite the fact many were recently human. Again, the focus on multiplayer gameplay somewhat furthers this—there is a greater weight given to challenge than narrative, as at the end of each campaign the player receives points and could receive awards such as “Headhunter” (most headshots), “Tank Slayer” (did the most damage to the special zombie “Tank”), and “General Defence” (killed the most infected), among others. The infected are not treated as complex figures but as tokens for points. There is little to no morality at play; in both games, players accept the infected to be morally black, or the evil “Other”. The Last of Us takes this further and asks to what extent humans actually align with this monstrous “Other.” Whereas Left 4 Dead enjoys the guilt-free ambiance of a morally black-and-white narrative, in this manner it is more traditional in its approach to world-building and violence. Not wanting to simplify too much, one should note that there are some who object to the notion of guilt-free cathartic violence against the non-human “Other.” Hartmann and Vorderer (2010), for instance, explore moral disengagement in violent video games including monitoring guilt in subjects playing Half-Life 2 (2004). This article does not seek to explore video games’ effects on the player. But regardless of whether the enemies of Left 4 Dead provoke guilt, or act as guilt-free abstract “Others,” as I argue, the “Othering” of non-human enemies still has a significant effect on the tone of the game. This is shown with the clear tonal shift between the fun “exploitation” style of Left 4 Dead, and the bleakness dog-eat-dog (or rather human-shoot-human) environment of The Last of Us. Equally, there is a clear moral distinction in tone between fighting the infected and other humans in The Last of Us. Perhaps, as with the games discussed earlier, the violence in The Last of Us is more impactful because it presents a fractured society we do not wish to contemplate. Left 4 Dead also presents chilling isolation and horror but allows one to combat this fear through over-the-top cathartic violence. Indeed, Left 4 Dead is so violent as to almost be cartoonish in tone despite the environment; as such it could be suggested that the violence in the game is used to relieve one's fears. Moreover, the aggression of enemies is key to justify the excessive violence (Hartmann and Vorderer) as hordes of infected repeatedly attack the player often and aggressively. This is also relevant to the Splicers of Bioshock. In Bioshock, violence is forced by necessity as Splicers attack on sight—but, as discussed, these characters do maintain some form of humanity. The combat gameplay is similar to many first-person shooters, though in Bioshock there are more shocking scenes where you can choose to be violent toward little children or save them instead. Although this appears in the game as a moral choice, the game mechanics reward you either way. Because of this, there is a lack of real moral choice. Jonathan Blow notes that the game's mechanics push the player to act selfishly as there is a lack of real sympathy or humanity in the supposed moral dilemmas of Bioshock; Blow furthers this by arguing that video games have conditioned players to accept such brutality and selfishness (Blow, 2007). Indeed, Bioshock and The Last of Us do present selfish depictions of humanity that are established by their violent, self-preserving, gameplay and narratives. But, I argue, the endings of these two games contradict their mechanics and present a different view of selfishness, love, and humanity. Selfishness and Loneliness Video games are one of the most personally interactive media; one takes control of the game, and so has a personal connection and personal experience within the narrative. In many ways, video games are inherently selfish or self-centered, focusing on self-preservation and self-betterment above all else. There are exceptions, in Left 4 Dead, for instance, teamwork is key as you will invariably rely on the other three survivors to help fight or rescue you (whether they are controlled by AI or other players). If one does become separated from their fellow survivors, either by accident or through gung-ho spirit, you will almost certainly be killed by the hordes of infected. Moreover, the game mechanics often require cooperation—for instance, “the Smoker” special zombie has a tongue that insnares players so that other players must release the victim. Left 4 Dead by its very nature is a cooperative game. Yet it is contrastingly no less isolating for its multiplayer focus—without your teammates you are vulnerable, but they too can die and leave you further alone. There are so many infected that you are overwhelmed, and the only form of contact with other survivors is through relics (graffiti, abandoned outposts). Although you work as part of a unit, there is a sense of fighting the inevitable and alone any player is vulnerable. The isolation of the game makes perfect strangers work together (both in the narrative and with players online), their individual hopelessness drives people together out of necessity—which, one could argue, is still within the realms of self-preservation above all else (especially when one considers that there remains a score-based system in the “credits” of the levels, still promoting a sense of individualism and one-upmanship throughout the game). Bioshock and The Last of Us differ in that it is their extreme isolation that causes the characters and players to act selfishly, though at other times they too offer an interesting contrast as both games show a human desire and need for communication and interaction. Until recently with morally gray games such as Spec Ops: The Line and The Last of Us Part II—and many others not discussed here, like Undertale (2015)—players entered a black-and-white world, under the impression that their character's views and actions were just and absolute. It is a recent trend in single-player gaming to make the player question their own character's selfish actions. The pandemic games of Bioshock and The Last of Us present scenes of selfishness in a crumbling world, forcing the player to query the morality of their actions. For instance, Joel distrusts and dislikes nearly all he meets on the road (even Ellie at first). Joel does become more human as his relationship with Ellie progresses, but even here, arguably to the end of the game, he is acting with his own interest. Again, we find that these games have an exceptional ability to create guilt and question humanity. In Bioshock the player comes across little girls around the map, known as “Little Sisters.” They are protected by huge, armored monsters call “Big Daddies.” The Little Sisters have been conditioned (by a mutated sea slug) to reclaim ADAM from corpses, as such they are considered valuable. The player is given the choice to remove the sea slug and release the children—which is encouraged by their “mother,” Tenenbaum—or kill the Little Sisters, “harvesting” them selfishly for their ADAM. There has been some criticism of Bioshock for this, though not because of the moral choice itself but because the choice is too easy as players are rewarded heavily for freeing the girls (Blow, 2007). Acting “good” results in reward and, therefore, acts as the correct approach for the player's best (selfish) interests; a more complex moral dilemma could be found if acting good led to little reward. Nonetheless, before knowing about the rewards, the decision to kill (and help yourself to the Little Sisters’ ADAM) or to save is quite harrowing. It forces the player to think about how far they will go to survive. The developers purposefully depict the Little Sisters in a way to evoke pity; they are all very young (between 5 and 10), where modest dresses, are shoeless, and cower from the protagonist—they are designed to inspire the image of innocence (with their glowing eyes presenting the uncanny corruption of the sea slug within). Their innocence is also highlighted through their mannerisms, as the girls skip, sing, and refer to their bodyguard Big Daddies affectionally (i.e., “Mr. Bubbles”). Perhaps it is fair also to say then that this design makes the choice somewhat obvious for players and in doing so further reduces the moral choice of the gameplay—in the face of their innocence it is harder and more obviously wrong to act too selfishly. Yet, there is an oxymoron in rescuing the Little Sisters—to harvest or save the girls (for either reward) the player must first kill their Big Daddies. These figures act as difficult mini-bosses and so defeating them feels rewarding, but they are humans (genetically enhanced) and the Sisters, even when freed, feel bonded to them, and even mourn their loss. Even to help in Bioshock, the player must inflict more pain and bring more loneliness to the world. The voluntary action of killing the Big Daddies and rescuing or harvesting the Little Sisters is designed to make the player uneasy about their choices and selfishness; yet you are compelled to act selfishly to survive as you are isolated and, for stretches of the game, alone with little to no contact from Splicers or other characters. Furthermore, the soundscape of Bioshock highlights one's loneliness; the constant groans and creaks of Rapture remind the player that they are trapped in this crumbling world—crushingly enclosed by the cold, dark of the sea. This oppressive sound design is furthered through the sickly green color scheme presenting a world of illness and unease. Truly, everything in Bioshock is uncanny. This is also emphasized by the Splicers who attempt to cover their mutated faces with masquerade masks, depicting a very human emotion of shame while acting with inhuman abilities and aggression. The further the player goes into the game they realize that Rapture is a place of deceit, manipulation, and insanity; late in the game, even the protagonist's free will is shown to be an illusion. While the player resists the world around them—initially by fighting Splicers—they are slowly absorbed into the world. You become violent, splice your DNA with ADAM, potentially show cruelty to the Little Sisters, and then, in the climax of the game, discover that you belong and were created in Rapture. In many ways, the sickness prevailing around Bioshock is personified by the player; you are on the surface human, but your actions are unnatural and tainted. Unlike in Left 4 Dead or The Last of Us, the figures in Bioshock do in some way resemble and act like humans—the singing from Splicers and Little Sisters resonant as an example of this in the oppressive and often quiet world of Rapture. There is an illness and a bubbling rage, but the humanity is not quite gone—as the player kills and mutilates their own body it is clear that they too are becoming less human as the game progresses. The game mechanics of using plasmids on oneself forces the player to become another sick and aggressive figure in the world of Rapture. However, though the player has participated in this oppressive world, they also have a chance to break it. Depending on how the player has acted toward the Little Sisters, the protagonist either ends the narrative by adopting the children and leading a positive life or they scare the children and release the Splicers—leading to the spread of hate and destruction to the surface-world. Evidently, either the player's selfishness or kindness is shown to have a great impact. Moreover, emotion and kindness, or lack thereof, plays a significance in the downfall of the antagonist, Ryan (Armin Shimerman). Ryan fails because of his hateful nature and the lack of understanding he portrays toward his son, Jack, the protagonist (Rose, 2015, pp. 20–21). In short, it is shown that selfishness leads to a lack of humanity and, ultimately, disaster. Throughout the game, we see that Jack, and by extension the player, is drawn toward human interaction, even if this is overwhelmed by their violence—it is noticeable, for instance, that the player is supposed to trust the voice of “Atlas” who appears to aid the player. Of course, as the game progresses, we realize that Atlas is actually a fraud, their real identity is Frank Fontaine (Greg Baldwin) who is trying to kill the player, but there is still a clear sense to look for hope in others when confronted with isolation. Equally, it is only when partnering with Dr. Tenebaum and her mob of Little Sisters that Ryan is defeated. We see in Bioshock a literal example of the power of uniting together, and it deepens how cruel and selfish the player has been to the Little Sisters in their journey of survival as to whether they get the good or evil ending. The illness in Bioshock is both a physical and psychological one; it is easy to get absorbed into this ill world, and in doing so hate spreads. But there is hope for humanity if the player questions and resists the negativity in the world around them; in other words, if the player resists the illness that befalls others in Rapture. Interestingly, there is the implication here that society and civilization have failed, but hope is found in the individual; arguably it is a bleak view of society, but not of the individual human condition. Hope is not a word that many would associate with Left 4 Dead; the “zombie” hordes and abandoned environments would suggest there is little to hope for, even if the limited narrative sees the four protagonists attempting to find some form of salvation. Although perfectly possible to play individually (with AI controlling the other four in the group), Left 4 Dead is designed as a multiplayer game, and teamwork is needed to be able to survive. Often the player will succumb to the infected and must hope for revival from their teammates. Unlike Bioshock, in Left 4 Dead, you are not truly alone. Looking at the campaign mode for Left 4 Dead and the teamwork needed, Scott Reed argues that there is the potential “to become more human” (Reed, 2011, p. 228). Players online who would otherwise have no social connection are forced to work together, much as the characters in the game's narrative are. There is a human bond in their struggle and a reliance on one another. Unlike Bioshock where the player may walk some distance without seeing a Splicer (though always fearing their presence), in Left 4 Dead the player is never far from the infected, often being surrounded by them. Although there are several human-like figures around the player, their uncanniness, discussed previously, creates a feeling of isolation. But as you are part of a small team there remains a purpose to fight. The characters in Left 4 Dead are forced together not by choice, but in hope—hope that through their joint humanity they may work together to survive. Admittedly, there is little redemption offered for the infected, who, again, are treated as the evil “Other”. But as the characters bond and work together there is an enduring sense of community. The inverse is found in The Last of Us. Despite its title, The Last of Us features several human factions, more so than the other games discussed in this article. Yet, nearly all other groups act as additional antagonists to be fought as frantically as any infected. Most noticeably, there is an oppressive military regime that is opposed by the rebellious Fireflies—other human groups included hunters, bandits, and cannibals (who all try to kill the player). Although Ellie admires the Fireflies and does make some friends in the game, there is division and questionable ethics displayed by the group. Joel's brother, Tommy (Jeffrey Pierce), offers a rare glimpse of community, forming a small commune. But even they only survive by fighting bandits, distrusting outsiders, and acting (relatively) selfishly. Distrust is prevalent among characters as a chief mechanic for survival. Yet characters are not one-dimensional as is exemplified in a cut scene as Ellie talks to David (Nolan North), a cannibal, who offers her food: David: No. No, I promise. It's… just the deer meat. Ellie: You're a fucking animal. [Ellie bends down and starts eating the food.] David: Oh. You're awfully quick to judgment. Considering you and your friend killed how many men? Ellie: They didn't give us a choice. David: And you think we have a choice? Is that it? You kill to survive… and so do we. We have to take care of our own. By any means necessary. Evidently, David (who oxymoronically seems like a kindly figure) has a sense of community and responsibility, but he is also willing to build this at the cost of others. In many ways, David is the epitome of the duality of humanity in this game. Humans have the capacity to love (we see this with Joel and Ellie's paternal relationship), but animalistic survival instinct is dominant. But against David we also question the actions of Joel and Ellie, are they truly much different? Unlike Bioshock or Left 4 Dead where the player is surrounded by the subhuman, in The Last of Us there are many other humans. And yet Joel and Ellie (and by extension the player) are forced away and distanced from others through violence, survival, and selfishness. It is interesting that it is not the infected that create this loneliness, but the self-imposed isolation from other humans. Toward the end of the game, Joel and Ellie reach Salt Lake City, where the Fireflies are finding a cure for the infection. Joel discovers this involves surgery on Ellie (which will kill her). To prevent this, he kills the surgeon and Marlene (Merle Dandridge), the leader of the Fireflies, and flees with a unconscious Ellie. The morality of Joel's actions has been widely debated since the game was released. It could be argued that Joel saving Ellie is in itself an act of selfishness (see, for instance, Tassi, 2013), but what the end of The Last of Us also shows is a bond between characters that trumps rationality. In a world of callous bleakness, love guides Joel, rightly or wrongly; that is a strong message for a game that shows the darkest side of humankind. Both Bioshock and The Last of Us show us that individuals are capable of love and compassion, we are forced to feel guilt as players if we do not save the Little Sisters, and made to feel painful anguish for the horrors faced by Joel and Ellie. Through realistic NPCs (Isbister, 2017, pp. 20–22, 41–42) and player-characters, these games make the player connect with the characters in the game, and so feel their pain, fear, and love. They reflect the complexities of humankind where love can prevail even in the bleakest of times, though it is fragile. Yet, as discussed, there is debate about the intention and supposed selfishness of Joel; and the rescuing of the Little Sisters in Bioshock is to the reward of the player. These games still present even our heroes as self-centered (or at least self-preserving) and present a society that is all too ready to turn on one another. This, I argue, reflects our own pre-Covid-19 fears as communities and societies have seemingly become more divided in recent years. Reality In September 2005, a mistake in coding led to a virtual pandemic in the massively multiplayer online role-playing game (MMORPG) World of Warcraft (2004). In one level, an “end boss”, Hakkar the Soulflayer (Chris Metzen), would dispel corrupted blood that infected and weakened the player. The attack was supposed to only have an effect in one area of the game, but unfortunately the blood also infected players’ pets and minions and, as such, spread the corrupted blood across the game's world. It took nearly a month for developers to fix the digital pandemic during which time deaths of player-characters were rampant, towns and cities in the game were deserted, and an in-game quarantine was advised by Blizzard Entertainment. At the time, there was some discussion by epidemiologists about using the “Corrupted Blood Incident” as a case study; although acknowledging the limitations of studying a virtual world, Balicer (2007) and Gary Smith (BBC, 2007) note that academic studies rarely consider the unpredictable nature of people during a pandemic—whether infected or not—which could be seen by studying gamers’ responses to the “Corrupted Blood Incident.” In an eerily reminiscent foreshadowing of Covid-19, some players helped others with healing abilities and obeyed quarantine, others simply fled, and some did not treat the pandemic seriously and even deliberately infected others (BBC, 2007). Players’ actions in the virtual pandemic are fairly compatible with real-life actions during Covid (Fenlon, 2020). Although a MMORPG and not a single-player narrative, and so more guided by player actions, the World of Warcraft incident demonstrates that we should not be too dismissive of how players act in the virtual world—it can be surprisingly reflective of reality. Although there is much negativity in the media surrounding Covid-19, what cannot be denied is the powerful response and selflessness that has been seen among communities during the pandemic. In the UK, for instance, the government aimed to recruit 250,000 volunteers to help the NHS during the early months of the pandemic—within two days 750,000 had volunteered (NHS, 2020). The public clapped weekly and painted rainbows to thank NHS staff, they donated £32.7 million to NHS charities as Captain Sir Tom Moore (a 100-year-old man) walked laps of his garden, and countless acts of everyday kindness have been prevalent during the Covid-19 pandemic. This selflessness was also evident in the virtual world with World of Warcraft. There are villains (in reality and virtually) and adversity is needed for challenge in video games, but the goodness of society and the communal nature of humans is often ignored in the pandemic worlds of video game narratives. This leads to a rather obvious question of why? In real life, humankind is communal, even in video games like World of Warcraft characters interacted and helped one another, so why is this not seen in games such as The Last of Us and Bioshock? I argue that the answer is fear. It is not uncommon for various forms of the media to reflect social anxiety; notably, works of horror often provide interesting reflections of society. For example, films like Cloverfield (2008) reflect the pain of 9/11 and the “torture-porn” genre, such as the Saw (2004–2010, 2017–present) and Hostel (2005–2011) film franchises, can be viewed as a reaction to the wars that followed. Horror, Aviva Briefel and Sam J. Miller explain, offers an environment where the “fundamental rules of our own reality no longer apply” which allows for an exploration of our societies under a veil of fiction (Briefel & Miller, 2011, p. 3). As with film, the reflection of social anxieties is evident in the pandemic and post-apocalyptic video games, which are nearly solely of the horror genre. This article has shown how games express and create feelings of guilt, self-centered behavior, and isolation. All of these aspects make the player question their morality and to a certain extent question what it is to be human (or at least humane). But these pandemic games show more than this; they present our fears. The emotional connection that arises from themes of guilt and isolation makes us also, in turn, question our own societies. There are two key fears these video games explore: the abject breakdown and division of civilized society, and the isolation and loneliness that stems from this. Previous pandemics and endemics such as HIV in the 1980s, the SARS outbreak of 2002–04, and recent Ebola outbreaks, have likely shaped our fear of viruses (Crockett & Zarracina, 2016). But I argue that the games of this study are more focused on society than illness. Currently, we live in a world that is clearly going through a significant social change in terms of technology, culture, and politics. As such, it is unsurprising that there is social anxiety about change and division. In The Last of Us there is a significant divide among groups; we have seen such fracturing in society (if to a lesser degree) over the past decade. This is apparent in recent years with political upsets dividing opinion (for instance Trump and Brexit)—but even events over the previous decade point to division in Western society (the 2011 London and UK riots, the Occupy Wall Street movement of the same year, the 2014 Ferguson unrest). If the dystopian video game plays upon anxieties and warns of their danger (Farca, 2018, p. 115), then the pandemic game, where society divides and is lost, stems from problems faced currently in society. The breakdown of society and civilization is exaggerated, but it is not a fear without foundation. Arguably, this social division is rather odd as we now have more avenues for connection than before, for instance through social media. However, some studies suggest social media use can actually lead to a sense of loneliness (Boursier et al., 2020; Hunt et al., 2018) and, more importantly, we have seen over recent years that social media highlights and even furthers the divisions in societies. We fear the removal of our technology, but technology has also caused loneliness and division—all themes and fears found and explored in pandemic games. In many ways, the pandemic game fetishizes the social anxieties of the day, the player is (more or less) alone as division and selfishness spread around them, as damaging as any illness. This manipulation of anxieties builds the tension found in the narratives of pandemic games; although these games can also offer a space to vent such frustration that is free from social constraint. This is most apparent in Left 4 Dead where violence is not only required but also encouraged. Even The Last of Us, which presents a “realistic” world, violently liberates players like no other medium would. Steven Poole compares Cormac McCarthy's post-apocalyptic The Road (book 2006 and film 2009) and The Last of Us because of their similar storylines, yet The Last of Us expects brutality, violence, and murder that would diminish the “tragic grandeur” of The Road's narrative (Poole, 2017). Evidently, as Blow also suggests (2007), video games appear to have a looser view of morality and humanity. Poole goes on to question why post-apocalyptic games do not show the rebuilding of society as would happen in reality. But this does not align with the anxieties the pandemic game narrative stems from, they reflect a fear that we are being divided, asking the player if we can even hope to fix our own society in the first place. This view is social anxiety to the nth degree and may also reflect the effects of George Gerbner's “mean world syndrome” where we have become use to pessimism and fear in the media and thus believe the world to be worse than it is (Laughey, 2007, pp. 20–21). Although the ends of Bioshock and The Last of Us suggest there is slight hope for the individual through compassion, the overall tone of these games is bleak. In effect, these games manipulate fears: fears of isolation and of a crumbling society—they depict a world that we as an audience are too ready to accept. Yet for all our fears, civilization has not fallen apart; history has shown us that societies pull together, even in digital environments like MMORPGs. Now we are living through a significant global pandemic and (for the most part) have not torn one another apart, one wonders if the post-apocalyptic pandemic video game will lose popularity. Or at least will they present a less pessimistic view of humanity and, perchance, a more sympathetic view of those infected? Our inherent social instincts appear stronger than we feared, or than video games have shown. Conclusion Although, as I have written, there have been many positive signs of humanity and caring during the Covid-19 outbreak, one should be careful not to view the situation from a narrow perspective. The Covid-19 pandemic has caused political and civil unrest across the world—the scale and severity of which differs from location to location and is dependent on other underlying tensions in each country. The world has seen events that are not unwholly like the panic, violence, and selfishness seen in the pandemic video game. Looting in South Africa in July 2021, triggered by the arrest of former President Jacob Zuma, saw the theft of several Covid-19 vaccines (Mlaba, 2021) and, earlier in the pandemic, there were reports of deadly riots in prisons in Columbia, Lebanon, and Italy (Azhari, 2020; BBC, 2020; Snuggs, 2020), triggered by prisoners panicking about unhygienic conditions, or else the removal of prisoner rights due to the then-developing spread of Covid-19. Yet, civilization continues in one form or another—the fears and anxieties explored in the pandemic video game are clearly not unfounded, but they are, perhaps, slightly misplaced. Clearly, the games discussed in this article present pandemics in the extreme, each shows a breakdown of society, little to no sympathy for those infected, and dog-eat-dog worlds. Pandemic games clearly reflect Blow's belief that over time gamers have been conditioned to expect inhumanity and to act inhumanely too. In Bioshock we are the outsider but are expected to kill and steal for our own survival; in The Last of Us there are several human groups, but the player is expected to trust, sympathize, and relate to Joel as he distrusts (and often kills) all those around him. Selfishness is expected, even from the heroes of the game; it is interesting that the pandemic game presents a view of society where social comradery is often the least protected value. Does this perhaps reflect the fears of a divided society? Or indeed reflect the concerns faced in the increasingly capitalist societies that these games were developed in. These games do not judge characters (or the player) for acting in such a way—in fact, it is expected and, in some cases, rewarded, even if one's self-preservation requires violence. Society, in a crisis of a pandemic, is expected to fail and the need for self-preservation will be asserted. These games largely prevail on the prevalence of division in society. And yet, while society's morality fails, these games still present individual human beings as capable of love and humanity. After all, while the player may fear division in society, they surely must know that they themselves are capable of good. And, indeed, kindness and a need for humanity still exist in these games: the relationship between Ellie and Joel, the teamwork of Left 4 Dead, or Tenenbaum's rescue of the Little Sisters all demonstrate this. Here, there is an odd divide as the games show that selfishness is to be expected but also express the connection, relatability, and compassion of humanity. Is this not, perhaps, again showing the mean world syndrome as we know that individually we are capable of love but expect the world to be worse than us, and worse than it is? In these games society has fallen, and enemies now surround, but the response to this new world differs. Although Left 4 Dead is a horror, the lightness of the narrative and gameplay style presents the game as cathartic fun—whereas The Last of Us and Bioshock have more morally gray plots that make you question what a monster actually is (a point emphasized in The Last of Us Part II); they show that humans have the capacity to descend beyond the infected. We connect with these games because they force the player, to an extent, to consider their choices, their morality, and humanity—humanity that there is hope for (as seen through Ellie and Joel's relationship). But these games do not offer representative depictions of their societies or of human nature, games like The Last of Us build a connection with the player because it presents what we fear. They show division and selfishness that in a pre-Covid-19 world were chief concerns. But now we have lived through a pandemic perhaps it is fair to say such games have become too perverse and will surely change in the future. Indeed, Turtle Rock Studios discussed their recently released Back 4 Blood (2021), which was in development before the Covid-19 outbreak, stating that their game is about building new safe places, noting that in other post-apocalyptic games “there's never any hope for the future. And there's never a way out of it, it's just the end of civilization” (McWhertor, 2020). It is slightly ironic that the developers of Left 4 Dead, the game that involves the most cathartic violence in this article (which also appears to continue in Back 4 Blood), are presenting a relatively hopeful depiction of an apocalyptic pandemic—but as we have seen in Left 4 Dead, and with pandemics in the digital and real worlds, when there is teamwork there is hope for humanity, civilization, and salvation—unless you are anything resembling a zombie that is. Author Biography Phil Wintle completed his PhD at the University of Leicester on the topic “The Representation of the Ku Klux Klan in American Cinema (1988–2016),” He is interested in topics of violence and racism in films and video games—though his research interests also extend to television and theatre. He is currently Head of Media and also teaches Drama at d’Overbroeck's, Oxford, having completed a PGCE as well as being an Associate Fellow of the Higher Education Academy (AFHEA). Previously he graduated with an MA with Merit in English from Loughborough University, after receiving an Upper Second Class (Hons) BA in Drama with English, at the same institution. The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author received no financial support for the research, authorship, and/or publication of this article. ORCID iD: Phil Wintle https://orcid.org/0000-0003-0745-4278 Primary Video Games: Bioshock. 2007. PC, Xbox 360 [Game]. 2 K Games: Novato, CA. Bioshock 2. 2010. PC, PlayStation 3, Xbox 360 [Game]. 2 K Games: Novato, CA. The last of us. 2013. PlayStation 3 [Game]. Sony Interactive Entertainment: San Mateo, CA. The last of us part II. 2020. PlayStation 4 [Game]. Sony Interactive Entertainment: San Mateo, CA. Left 4 dead. 2008. MacOS, PC, Xbox 360 [Game]. Valve: Bellevue, WA. Left 4 dead 2. 2009. MacOS, PC, Xbox 360 [Game]. Valve: Bellevue, WA. Secondary Video Games: Back 4 blood. 2021. PlayStation 4, PlayStation 5, PC, Xbox One, Xbox Series X [Game]. Warner Bros. Games: Burbank, CA. Call of duty: modern warfare 2. 2009. PC, PlayStation 3, Xbox 360 [Game]. Activision: Santa Monica, CA. Doom. 1993. MS-DOS [Game]. id Software: Shreveport, LA. [And subsequent series]. Grand theft auto V. 2013. PlayStation 3, Xbox 360 [Game]. Rockstar Games: New York. Half-life 2. 2004. PC [Game]. Valve: Bellevue, WA. Metro 2033. 2010. PC, Xbox 360 [Game]. THQ [defunct]: Agoura Hills, CA. [And subsequent series published by Deep Silver]. Mortal kombat. 1992. Arcade [Game]. Midway: Chicago. Resident evil. 1996. PlayStation [Game]. Capcom: Osaka. [And subsequent series]. Space invaders. 1978. Arcade [Game]. Taito: Tokyo. Spec ops: the line. 2012. PC, PlayStation 3, Xbox 360 [Game]. 2 K Games: Novato, CA. Undertale. 2015. MacOS, PC [Game]. Toby Fox: [Unknown]. World of warcraft. 2004. MacOS, PC [Game]. Blizzard Entertainment: Irvine, CA. Zombie army 4: dead war. 2020. PC, PlayStation 4, Xbox One [Game]. Rebellion Developments: Oxford. [NB: Only the original release platform is listed for the above games. Locations given are of publishers’ contemporary headquarters where possible]. Filmography: Cloverfield. 2008. [Film]. Matt Reeves, dir. USA: Bad Robot Productions. Hostel. 2005. [Film]. Eli Roth, dir. USA: Raw Nerve and Next Entertainment. [And subsequent series]. The Road. 2009. [Film]. John Hillcoat, dir. USA: 2929 Productions. Saw. 2004. [Film]. James Wan, dir. USA: Twisted Pictures. [And subsequent series]. ==== Refs References Azhari T. (2020). Riots rock overcrowded Lebanon prisons over coronavirus fears. [online] Aljazeera. Retrieved July 30, 2021, from https://www.aljazeera.com/news/2020/3/17/riots-rock-overcrowded-lebanon-prisons-over-coronavirus-fears Balicer R. D. (2007). Modeling infectious diseases dissemination through online role-playing games. Epidemiology, 18 (2 ), 260–261. 10.1097/01.ede.0000254692.80550.60 17301707 BBC (2007). Virtual game is a ‘disease model’. [online] BBC. Retrieved July 25, 2021, from http://news.bbc.co.uk/1/hi/health/6951918.stm BBC (2020). Coronavirus: At least 23 killed in Colombia prison unrest. [online] BBC. Retrieved July 30, 2021, from https://www.bbc.co.uk/news/world-latin-america-51999594 Blow J. (2007). Design reboot. Montreal International Games Summit. Montreal, 27–28 November. Boursier V. Gioia F. Musetti A. Schimmenti A. (2020). Facing loneliness and anxiety during the COVID-19 isolation: The role of excessive social media use in a sample of Italian adults. Frontiers in Psychiatry, 11 . 10.3389/fpsyt.2020.586222 Bramwell T. (2013). Is the most disturbing scene in GTA 5 justified? [online] Eurogamer. Retrieved July 24, 2021, from https://www.eurogamer.net/articles/2013–09-16-is-the-most-disturbing-scene-in-gta5-justified Briefel A. Miller S. J. (2011). Introduction. In Briefel A. Miller S. J. (Eds.), Horror after 9/11: World of fear, cinema, terror (pp. 1–12). University of Texas Press. Crockett Z. Zarracina J. (2016). How the zombie represents America’s deepest fears. [online] Vox. Retrieved July 29, 2021, from https://www.vox.com/policy-and-politics/2016/10/31/13440402/zombie-political-history Dougherty C. (2020, June 19). ‘The last of us part II’ Is a dark game for a dark time. The New York Times. [online]. Retrieved July 29, 2021, from https://www.nytimes.com/2020/06/19/business/last-of-us-2-review.html Farca G. (2018). Playing dystopia: Nightmarish worlds in video games and the player’s aesthetic response. Transcript. Fenlon W. (2020). The researchers who once studied WoW’s corrupted blood plague are now fighting the coronavirus. [online] PC Gamer. Retrieved July 25, 2021, from https://www.pcgamer.com/the-researchers-who-once-studied-wows-corrupted-blood-plague-are-now-fighting-the-coronavirus/ Hartmann T. Vorderer P. (2010). It’s okay to shoot a character: Moral disengagement in violent video games. Journal of Communication, 60 (1 ), 94–119. 10.1111/j.1460-2466.2009.01459.x Hayton J. (2012). Digital Nazis: Genre, history and the displacement of evil in first-person shooters. In Magilow D. H. Vander Lugt K. T. Bridges E. (Eds.), Naziploitation!: The Nazi image in low-brow cinema and culture (pp. 199–218). Continuum. Hunt M. Marx R. Lipson C. Young J. (2018). No more FOMO: Limiting social media decreases loneliness and depression. Journal of Social and Clinical Psychology, 37 (10 ), 751–768. 10.1521/jscp.2018.37.10.751 Isbister K. (2017). How games move us: Emotion by design. MIT Press. Laughey D. (2007). Key themes in media theory. Maidenhead Open University Press. MacDonald K. (2013). ‘Grand Theft Auto V’ review. [online] IGN. Retrieved July 23, 2021, from https://www.ign.com/articles/2013/09/16/grand-theft-auto-v-review Markert J. (2011). Post-9/11 cinema: Through a lens darkly. Scarecrow Press. McIntosh S. (2008). The evolution of the zombie: The monster that keeps coming back. In McIntosh S. Leverette M. (Eds.), Zombie culture: Autopsies of the living dead (pp. 1–18). Scarecrow Press. McWhertor M. (2020). ‘Back 4 Blood’ adds new complexity, bravado to the ‘Left 4 Dead’ formula. [online] Polygon. Retrieved July 30, 2021, from https://www.polygon.com/2020/12/17/22179273/back-4-blood-preview-turtle-rock-interview-left-4-dead Mlaba K. (2021). Civil unrest in South Africa: How the vaccine rollout has been affected. [online] Global Citizen. Retrieved July 18, 2021, from https://www.globalcitizen.org/en/content/unrest-in-south-africa-impact-on-vaccine-rollout/ NHS (2020). NHS volunteer responders: 250,000 target smashed with three quarters of a million committing to volunteer. [online] NHS. Retrieved July 26, 2021, from https://www.england.nhs.uk/2020/03/250000-nhs-volunteers/ Petit C. (2014). ‘Grand Theft Auto V’ Review. [online] GameSpot. Retrieved July 20, 2021, from https://www.gamespot.com/reviews/grand-theft-auto-v-review/1900-6414475/ Poole S. (2017). Why video games are obsessed with the apocalypse. [online] BBC. Retrieved July 19, 2021, from https://www.bbc.com/future/article/20170815-why-video-games-are-obsessed-with-the-apocalypse Reed S. (2011). Rhetoric goes boom(er): Agency, networks, and zombies at play. In Boluk S. Lenz W. (Eds.), Generation zombie: Essays on the living dead in modern culture (pp. 219–230). McFarland. Rose J. (2015). The value of art in ‘Bioshock’: Ayn Rand, emotion, and choice. In Cuddy L. (Ed.), Bioshock and philosophy (pp. 15–26). Wiley. Rouse R. III (2009). Match made in hell: The inevitable success of the horror genre in video games. In Perron B. (Ed.), Horror video games: Essays on the fusion of fear and play (pp. 15–25). McFarland. Sheldon T. A. Wright J. (2020). Twin epidemics of COVID-19 and non-communicable disease. BMJ, 369. 10.1136/bmj.m2618 Sims D. (2020). ‘The last of us part II’ tests the limits of video-game violence. [online] The Atlantic. Retrieved July 30, 2021, from https://www.theatlantic.com/culture/archive/2020/07/the-last-of-us-limits-video-game-violence/613696/ Snuggs T. (2020). Coronavirus: Six inmates killed during prison riots across Italy over measures to tackle virus. [online] Sky News. Retrieved July 30, 2021, from https://news.sky.com/story/coronavirus-six-inmates-killed-during-prison-riots-across-italy-over-measures-to-tackle-virus-11953759 (accessed July 30, 2021). Stobbart D. (2019). Videogames and horror: From amnesia to zombies, run!. University of Wales Press. Tassi P. (2013, June 17). Coming to terms with the difficult ending of ‘The last of us’. Forbes. [online]. Retrieved July 29, 2021, from https://www.forbes.com/sites/insertcoin/2013/06/17/comin g-to-terms-with-the-difficult-ending-of-the-last-of-us/?sh=528e36db5f47 The W.H.O. (n.d.). Noncommunicable diseases. [online] The W.H.O. Retrieved November 28, 2021, from https://www.who.int/health-topics/noncommunicable-diseases#tab=tab_1 Wells P. (2000). The horror genre: From beelzebub to blair witch. Columbia University Press.
PMC010xxxxxx/PMC10291223.txt	==== Front J Intensive Care Med J Intensive Care Med JIC spjic Journal of Intensive Care Medicine 0885-0666 1525-1489 SAGE Publications Sage CA: Los Angeles, CA 37350092 10.1177/08850666231182380 10.1177_08850666231182380 Original Research COVID-19 Infection Is Associated With Increased In-Hospital Mortality and Complications in Patients With Acute Heart Failure: Insight From National Inpatient Sample (2020) https://orcid.org/0000-0002-2008-7110 Hashem Anas MD 1 Khalouf Amani MD 1 Mohamed Mohamed Salah MD 1 https://orcid.org/0000-0001-9052-5537 Nayfeh Tarek MD 2 Elkhapery Ahmed MD 1 Elbahnasawy Mohammad MD 1 Rai Devesh MD 3 https://orcid.org/0000-0003-2086-1281 Deshwal Himanshu MD 4 Feitell Scott DO 3 Balla Sudarshan MD 5 1 6932 Internal Medicine Department , Rochester General Hospital, Rochester, NY, USA 2 4352 Evidence-based medicine, Mayo Clinic School of Medicine , Rochester, MN, USA 3 Department of Cardiology, Rochester General Hospital, Sands-Constellation Heart Institute, Rochester, NY, USA 4 Department of Pulmonary, Sleep and Critical Care Medicine, West Virginia University, Morgantown, WV, USA 5 Department of Cardiovascular Disease, West Virginia University – Health Sciences Campus, Morgantown, WV, USA Anas Hashem, Internal Medicine Department, Rochester General Hospital, 1425 Portland Ave, Rochester, NY 14621, USA. Email: anas.hashem@rochesterregional.org 23 6 2023 23 6 2023 0885066623118238002 2 2023 31 5 2023 © The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Introduction: Patients with acute heart failure (AHF) exacerbation are susceptible to complications in the setting of COVID-19 infection. Data regarding the clinical outcomes of COVID-19 in patients admitted with AHF is limited. Methods: We used the national inpatient sample database by utilizing ICD-10 codes to identify all hospitalizations with a diagnosis of AHF in 2020. We classified the sample into AHF with COVID-19 infection versus those without COVID-19. Primary outcome was in-hospital mortality. Secondary outcomes were acute myocardial infarction, need for pressors, mechanical cardiac support, cardiogenic shock, and cardiac arrest. Also, we evaluated for acute pulmonary embolism (PE), bacterial pneumonia, need for a ventilator, and acute kidney injury (AKI). Results: We identified a total of 694,920 of AHF hospitalizations, 660,463 (95.04%) patients without COVID-19 and 34,457 (4.96%) with COVID-19 infection. For baseline comorbidities, diabetes mellitus, chronic heart failure, ESRD, and coagulopathy were significantly higher among AHF patients with COVID-19 (P < .01). While CAD, prior MI, percutaneous coronary intervention, and coronary artery bypass graft, atrial fibrillation, chronic obstructive pulmonary disease, and peripheral vascular disease were higher among those without COVID-19. After adjustment for baseline comorbidities, in-hospital mortality (aOR 5.08 [4.81 to 5.36]), septic shock (aOR 2.54 [2.40 to 2.70]), PE (aOR 1.75 [1.57 to 1.94]), and AKI (aOR 1.33 [1.30 to 1.37]) were significantly higher among AHF with COVID-19 patients. The mean length of stay (5 vs 7 days, P < .01) and costs of hospitalization ($42,143 vs $60,251, P < .01) were higher among AHF patients with COVID-19 infection. Conclusion: COVID-19 infection in patients with AHF is associated with significantly higher in-hospital mortality, need for mechanical ventilation, septic shock, and AKI along with higher resource utilization. Predictors for mortality in AHF patients during the COVID-19 pandemic, COVID-19 infection, patients with end-stage heart failure, and atrial fibrillation. Studies on the impact of vaccination against COVID-19 in AHF patients are needed COVID-19 acute heart failure in-hospital mortality clinical outcomes national inpatient sample edited-statecorrected-proof typesetterts19 ==== Body pmcIntroduction The Coronavirus disease 2019 (COVID-19) 1 is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus 2 and its variant has led to significant morbidity and mortality due to its effect on multiple organ systems including the cardiovascular system. 3 The virus can directly cause damage to the cardiac myocytes via angiotensin-converting enzyme-2 (ACE-2) receptors, 4 and presents in the form of myocarditis and acute myocardial infarction (MI) causing major acute cardiovascular events, including heart failure, arrhythmias, and sudden cardiac arrest. 5 Indirectly, it can induce severe inflammatory immune response, cytokine storm, and immune system dysfunction resulting in additional myocardial injury and arterial thrombosis. 5 Prior studies have demonstrated a higher morbidity or mortality from COVID-19 infection in chronic comorbidities such as heart failure.2,6 However, despite the various reports about its clinical impact on human health in general, there is scarce data regarding the trends of in-hospital mortality, clinical outcomes, and prognostic factors in patients hospitalized with acute heart failure (AHF) and a confirmed diagnosis of SARS-CoV-2 infection. 3 Therefore, we aim to utilize a contemporary database of inpatient hospitalizations in the US, the National Inpatient Sample 2020, to evaluate the trends and outcomes of AHF during the COVID-19 pandemic. Methodology Data Source The National Inpatient Sample (NIS) is one of several databases managed by the Agency for Healthcare Research and Quality (AHRQ) through a Federal-State-Industry partnership called the Healthcare Cost and Utilization Project (HCUP). The NIS 2020 database contains administrative claims data from more than 7 million inpatient hospitalizations annually in 43 participating states plus the District of Columbia, representing more than 97% of the USA population. Since NIS data are compiled annually, the data can be used for the analysis of procedure trends over time using trend weights compiled by the HCUP. To calculate hospitalization cost which includes total expenses incurred to provide services, cost-to-charge ratio (CCR) files were used. The CCR files provide hospital-specific ratios to calculate hospitalization cost based on the specific hospitalization characteristics. Ethical Consideration Institutional Review Board approval and informed consent were not required for this study because NIS data are de-identified and publicly available. The study was conducted in compliance with the ethical standards of the responsible institution on human subjects as well as with the Helsinki Declaration. As per HCUP guidelines, observations with cell count <11 are reported as “<11”. Study Sample and Patient Selection We analyzed NIS data using the International Classification of Diseases, 10th Revision, and Clinical Modification (ICD-10-CM) claims codes. ICD-10-CM codes of I5021, I5023, I5031, I5033, I5041, I5043, I50811, and I5081 were used to identify patients hospitalized with AHF exacerbation. To identify COVID-19 cases, we used ICD-10 code of U071. All diagnosis and procedure fields were queried to select and categorize the study population, ICD-10 codes are reported in Table S1. Individuals under 18 years old and those who were admitted electively were excluded from the study. A detailed methods flowchart is presented in Figure S1. Variables Baseline patient characteristics included demographic variables (age, sex, and race), baseline comorbidities including hypertension, type 2 diabetes mellitus, coronary artery disease (CAD), prior MI, percutaneous coronary artery intervention (PCI), coronary artery bypass graft (CABG), valve surgery, pacemaker placement (PPM), and implantable cardioverter-defibrillator (ICD), atrial fibrillation, chronic heart failure, end-stage heart failure (ESHF), peripheral vascular disease (PVD), cerebrovascular accident (CVA), chronic obstructive pulmonary disease (COPD), pulmonary circulation disorders, chronic kidney disease (CKD), end-stage renal disease (ESRD), liver disease, coagulopathy, and obesity. Study Outcomes The primary study endpoint was in-hospital mortality. Secondary endpoints were clinical outcomes including (1) cardiovascular complications, such as acute MI, need for pressors, veno-arterial extracorporeal membrane oxygenation (VA-ECMO), impella, intra-aortic balloon pump (IABP), mechanical cardiac support, cardiogenic shock, and cardiac arrest, (2) pulmonary complications, such as acute pulmonary embolism (PE), bacterial pneumonia and need for ventilator, (3) septic shock and need for pressors, (4) acute kidney injury (AKI), (5) resource utilization (length of stay and cost of hospitalization). Associated procedures and complications were identified by using ICD-10-CM codes, Table S1. Statistical Analysis To estimate the effect of COVID-19 diagnosis on the outcomes of interest, we performed a 1:1 nearest-neighbor propensity-matching without replacement with a propensity score estimated using logistic regression. The matching accounted for confounding from all baseline variables. The matching yielded adequate balance as indicated by a caliper of 0.005 standard deviation, standardized mean differences for all the included covariates below 0.2, and low Kolmogorov–Smirnov statistics, Figure S2. The categorical variables were presented as counts and percentages and compared using Pearson's Chi-square. Continuous variables had a skewed distribution and were presented as median and interquartile range (IQR) and were compared using nonparametric Mann–Whitney U test. To determine independent predictors of in-hospital mortality, we conducted logistic regression analysis. Baseline variables associated with the outcome in a univariate analysis were included in a multivariate logistic regression model. Results of the logistic regression analysis were presented as odds ratios with 95% confidence intervals. The data were complete for all variables except for in-hospital mortality (193; 0.03%), sex (28; 0.004%), and race (14,087; 2.03%). We did not impute missing data because missing data of less than 5% were not likely to introduce bias. 7 The Joinpoint Regression software used t-tests to determine if monthly percent change (MPCs) and/or annual (APC [P-trend across the year of 2020]) were statistically significant from zero. For all analyses, a two-tailed P-value of ≤.05 was considered statistically significant. All statistical analyses were performed using R version 4.2.2 and SPSS version 26® (IBM Corp, Armonk, NY). Results Baseline Characteristics A total of 694,920 cases of AHF hospitalizations, out of which 660,463 (95.04%) were patients without COVID-19 and 34,457 (4.96%) with COVID-19 infection. Patients with AHF and COVID-19 were older (median age 74 vs 73 years, P < .001), and the majority were male (53.7% vs 52.1%, P < .001). Figure S3 shows the age distribution. Majority of patients were Caucasian (67.2%), followed by African American (18.3%) and Hispanic (7.5%). Baseline comorbidities such as diabetes mellitus (32.1% vs 26.6%, P < .001), chronic heart failure (77.1% vs 67.7%, P < .001), ESRD (11.2% vs 9.2%, P < .001), coagulopathy (15.6% vs 9.7%, P < .001), and obesity (27.8% vs 26.5%, P < .001) were significantly higher among patients with COVID-19. Meanwhile, CAD (43.9% vs 49.0%, P < .001), prior MI (11.5% vs 14.4%, P < .001), PCI (10.7% vs 12.7%, P < .001), CABG (10.0% vs 11.1, P < .001), atrial fibrillation (29.1% vs 30.3%, P < .001), COPD (31.9% vs 35.2%, P < .001), and PVD (2.1% vs 3.4%, P < .001) were lower among those with COVID-19. The detailed baseline characteristics before and after propensity score matching analysis are shown in Table 1. Table 1. Baseline Characteristics of the AHF Population in Unmatched Cohort and Propensity-Score Matched Cohort. Unmatched Cohort Propensity-Score Matched Cohort Variable n (%) AHF Without COVID-19 (660,463) AHF with COVID-19 (34,457) P-Value AHF Without COVID-19 (33,668) AHF with COVID-19 (33,668) P-Value Age (Median, [IQR]) 73 [62-82] 74 [64-83] <.001 72.9 (13.0) 72.6 (13.1) <.001 Female 316,070 (47.9) 15,947 (46.3) <.001 15,685 (46.6) 15,581 (46.3) .426 Race <.001 <.001 White 446,788 (69.0) 20,278 (60.2) 20,668 (61.4) 20,278 (60.2) Black 120,048 (18.5) 7402 (22.0) 7351 (21.8) 7402 (22.0) Hispanic 48,361 (7.5) 4053 (12.0) 3967 (11.8) 4053 (12.0) Asian or Pacific Islander 13,449 (2.1) 720 (2.1) 642 (1.9) 720 (2.1) Native American 3777 (0.6) 279 (0.8) 218 (0.6) 279 (0.8) Other 14,742 (2.3) 936 (2.8) 822 (2.4) 936 (2.8) Diabetes mellitus 175,933 (26.6) 11,057 (32.1) <.001 10,508 (31.2) 10,788 (32.0) .021 Hypertension 7750 (1.2) 410 (1.2) .782 352 (1.0) 408 (1.2) .045 CAD 323,713 (49.0) 15,143 (43.9) <.001 14,550 (43.2) 14,810 (44.0) .044 Prior MI 94,983 (14.4) 3976 (11.5) <.001 3711 (11.0) 3896 (11.6) .025 Prior PCI 83,646 (12.7) 3697 (10.7) <.001 3360 (10.0) 3613 (10.7) .001 Prior CABG 73,547 (11.1) 3451 (10.0) <.001 3160 (9.4) 3376 (10.0) .005 Prior valve surgery 30,725 (4.7) 1285 (3.7) <.001 1113 (3.3) 1256 (3.7) .003 Prior PPM 46,199 (7.0) 2338 (6.8) <.001 2276 (6.8) 2290 (6.8) .842 Prior ICD 45,060 (6.8) 2103 (6.1) <.001 1785 (5.3) 2041 (6.1) <.001 Atrial fibrillation 199,830 (30.3) 10,041 (29.1) <.001 9740 (28.9) 9804 (29.1) .593 Chronic heart failure 447,285 (67.7) 26,562 (77.1) <.001 26,170 (77.7) 25,970 (77.1) .067 HFrEF 104,659 (15.8) 6968 (20.2) 7040 (20.9) 6968 (20.7) HFpEF 173,044 (26.2) 12,899 (37.4) 12,929 (38.4) 12,899 (38.3) HF, other 169,582 (26.7) 6677 (19.4) 6201 (18.4) 6103 (18.1) End stage heart failure 5516 (0.8) 105 (0.3) <.001 102 (0.3) 103 (0.3) 1.00 PVD 22,706 (3.4) 716 (2.1) <.001 637 (1.9) 706 (2.1) .061 CVA 44,321 (6.7) 2154 (6.3) <.001 1934 (5.7) 2094 (6.2) <.001 COPD 232,694 (35.2) 10,982 (31.9) <.001 10,943 (32.5) 10,743 (31.9) .101 Pulmonary circulation disorder 115,218 (17.4) 3961 (11.5) <.001 3775 (11.2) 3869 (11.5) .259 CKD 259,292 (39.3) 13,028 (37.8) <.001 12,903 (38.3) 12,737 (37.8) .19 ESRD 60,880 (9.2) 3843 (11.2) <.001 3606 (10.7) 3752 (11.1) .073 Liver disease 58,626 (8.9) 2397 (7.0) <.001 2100 (6.2) 2326 (6.9) <.001 Coagulopathy 64,212 (9.7) 5378 (15.6) <.001 4928 (14.6) 5266 (15.6) <.001 Obesity 174,754 (26.5) 9586 (27.8) <.001 9199 (27.3) 9375 (27.8) .131 Abbreviations: CABG, coronary artery bypass graft; CAD, coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; DM, diabetes mellitus; ESRD, end-stage renal disease; HF, heart failure; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; ICD, implanted cardioverter-defibrillator; IQR, interquartile range; MI, myocardial infarction; PCI, percutaneous coronary intervention; PPM, pacemaker placement; PVD, peripheral vascular disease. As per HCUP regulations, observations with cell count <11 are reported as “<11”. Bold texts indicate statistical significance. Clinical Outcomes In-Hospital mortality In-hospital mortality was significantly higher among patients with COVID-19 (22.0% vs 5.0%, P < .001). There was an overall significant increase in in-hospital crude mortality across the year of 2020 (APC 106.7, 95% CI [11.0 to 284.8], P-trend = .02), with the first peak in COVID-19 mortality in April (MPC 1296.7, 95% CI [329 to 4451], P < .01) followed by a decline (MPC −35.3, 95% CI [−96.5 to 1088], P = .05) and then a second peak in December (MPC 31.9, 95% CI [−15.5 to 105.7], P = .54), Figure 1A. Conversely, the mortality rates overall trended down during the study period (APC −3.6, 95% CI [−5.6 to −1.6], P < .01) with a peak in March with a monthly increase by 7.32% (95% CI [−2.2 to 17.8], P = .10), then trended down reaching a nadir in June with a monthly reduction by 16.1% (95% CI [−22.9 to −8.6], P < .01), and plateaued from June to December (MPC −0.4, 95% CI [−1.4 to 0.6], P = .33), Figure 1B. Figure 1. Mortality incidence (A) and rate (B) in AHF patients with COVID-19 patients across the year of 2020. Clinical complications On propensity-matched analysis, cardiac arrest (4.2% vs 1.8%, P < .001), septic shock (11.2% vs 4.7%, P < .001), need for pressors (4.3% vs 2.5%, P < .001), acute PE (2.7% vs 1.6%, P < .001), bacterial pneumonia (9.9% vs 3.3%, P < .001), need for ventilator (11.6% vs 8.2%, P < .001), use of VA-ECMO (0.2% vs 0.1%, P = .02), and AKI (45.4% vs 38.4%, P < .001) were higher among AHF with COVID-19. Other complications such as acute MI (6.5% vs 5.4%, P < .001), use of mechanical cardiac support (0.6% vs 0.4%, P < .001), impella (0.2% vs 0.1%, P < .001), IABP (0.4% vs 0.2%), and need for blood transfusion (7.8% vs 6.5%, P < .001) were higher among AHF without COVID-19. On adjusted analysis propensity score-adjusted analysis, patients with COVID-19 and AHF were at higher odds of cardiac arrest (aOR 2.37 [2.15 to 2.61]), septic shock (aOR 2.54 [2.40 to 2.70]), need for pressors (aOR 1.79 [1.64 to 1.95]), acute PE (aOR 1.75 [1.57 to 1.94]), and need for ventilator (aOR 1.46 [1.39 to 1.54]) and AKI (aOR 1.33 [1.30 to 1.37]). Further detailed information regarding clinical outcomes before and after propensity score-based adjustment is listed in Table 2, Figure 2, and Table S2. Figure 2. Adjusted and unadjusted analysis for the outcomes including in-hospital mortality, cardiogenic shock, septic shock, MCS, need for pressors, acute PE, pneumonia, need for ventilator, and AKI in patients with AHF and COVID-19 infection. Table 2. In-Hospital Complications and Clinical Outcomes on the Unmatched, Crude Analysis, and Following Propensity-Score Matching. Crude Analysis Propensity-Score Matching analysis Variable n (%) AHF Without COVID-19 (660,463) AHF With COVID-19 (34,457) P-Value AHF Without COVID-19 (33,668) AHF With COVID-19 (33,668) P-Value Died during hospitalization 32,754 (5.0) 7589 (22.0) <.001 1768 (5.3) 7393 (22.0) <.001 Acute myocardial infarction 50,429 (7.6) 1896 (5.5) <.001 2172 (6.5) 1823 (5.4) <.001 VA-ECMO 792 (0.1) 59 (0.2) .010 33 (0.1) 56 (0.2) .02 Impella 1437 (0.2) 17 (0.0) <.001 51 (0.2) 17 (0.1) <.001 IABP 3754 (0.6) 71 (0.2) <.001 144 (0.4) 65 (0.2) <.001 Mechanical cardiac support 5429 (0.8) 135 (0.4) <.001 201 (0.6) 127 (0.4) <.001 Cardiogenic shock 20,932 (3.2) 810 (2.4) <.001 834 (2.5) 777 (2.3) .158 Cardiac arrest 11,736 (1.8) 1431 (4.2) <.001 604 (1.8) 1399 (4.2) <.001 Septic shock 29,161 (4.4) 3858 (11.2) <.001 1592 (4.7) 3775 (11.2) <.001 Need for pressors 14,789 (2.2) 1502 (4.4) <.001 830 (2.5) 1459 (4.3) <.001 Need for blood transfusion 44,570 (6.7) 2256 (6.5) .150 2612 (7.8) 2203 (6.5) <.001 Acute pulmonary embolism 11,858 (1.8) 934 (2.7) <.001 528 (1.6) 911 (2.7) <.001 Bacterial pneumonia 21,424 (3.2) 3412 (9.9) <.001 1102 (3.3) 3332 (9.9) <.001 Need for a ventilator 54,990 (8.3) 4021 (11.7) <.001 2773 (8.2) 3910 (11.6) <.001 AKI 250,433 (37.9) 15,654 (45.4) <.001 12,930 (38.4) 15,287 (45.4) <.001 Resource utilization Length of hospital stay (Median, [IQR]) 5.00 [3.00, 8.00] 7.00 [4.00, 13.00] <.001 5.00 [3.00, 8.00] 7.00 [4.00, 13.00] <.001 Median cost of hospitalization (Median, [IQR]) 45,663.00 [25,241.00, 88,825.00] 60,209.00 [31,814.00, 123,566.50] <.001 42,143.50 [22,960.00, 83,562.50] 60,251.00 [31,824.75, 123,762.50] <.001 Abbreviations: AKI, acute kidney injury; IABP, intra-aortic balloon pump; IQR, interquartile range; VA-ECMO, veno-atrial extracorporeal membrane oxygenation. As per HCUP regulations, observations with cell count <11 are reported as “<11”. Bold texts indicate statistical significance. Resources Utilization In terms of resource utilization, the length of hospitalization (LOS; median: 7 vs 5 days, P < .001) and the median costs of hospitalization ($60,209 vs $45,663, P < .001) for patients with AHF and COVID-19 infection were significantly higher, Table 2, Figure S4. Central illustration for our study is shown in Figure 3. Figure 3. Central illustration. Mortality Predictors In-hospital mortality varied across different clinical baseline characteristics, Table 3. On univariate analyses, in-hospital mortality was associated with variables such as COVID-19, age, gender, race, DM, HTN, CAD, prior MI, PCI, CABG, valve surgery, PPM, ICD, atrial fibrillation, chronic and end-stage heart failure, COPD, ESRD, liver disease, coagulopathy, and obesity. Table 3. Predictors of In-Hospital Mortality in Patients Diagnosed with AHF Exacerbation During Their Hospitalization. Variable n (%) Univariate Analysis Multivariate Analysis COVID-19 diagnosis 5.42 (5.27-5.57) 5.67 (5.50-5.84) Age (per year) 1.0195 (1.0187-1.0204) 1.026 (1.025-1.027) Female 0.90 (0.88-0.92) 0.90 (0.88-0.92) Race White (reference) 1 1 Black 0.82 (0.80-0.84) 0.85 (0.83-0.88) Hispanic 1.07 (1.03-1.11) 0.93 (0.90-0.97) Asian or Pacific Islander 1.20 (1.13-1.29) 1.01 (0.94-1.08) Native American 1.20 (1.06-1.36) 1.15 (1.01-1.31) Other 1.13 (1.06-1.21) 1.03 (0.97-1.11) Type 2 diabetes mellitus 0.88 (0.86-0.90) 0.97 (0.94-1.00) HTN 0.78 (0.70-0.86) 0.81 (0.73-0.90) CAD 0.81 (0.79-0.82) 0.85 (0.83-0.87) Prior MI 0.69 (0.67-0.71) 0.83 (0.80-0.86) Prior PCI 0.59 (0.57-0.61) 0.72 (0.69-0.75) Prior CABG 0.81 (0.79-0.84) 0.92 (0.89-0.96) Prior Valve surgery 0.75 (0.71-0.79) 0.77 (0.72-0.82) Prior PPM 0.74 (0.71-0.78) 0.67 (0.64-0.70) Prior ICD 0.67 (0.63-0.70) 0.69 (0.66-0.73) Atrial fibrillation 1.11 (1.09-1.14) 1.04 (1.02-1.07) Chronic HF 0.73 (0.72-0.75) 0.63 (0.62-0.65) End-stage HF 2.12 (1.95-2.30) 2.04 (1.87-2.23) PVD 1.01 (0.96-1.07) Not includeda CVA 0.96 (0.93-1.00) Not includeda COPD 0.93 (0.91-0.95) 1.03 (1.01-1.05) Pulmonary circulation disorder 1.02 (1.00-1.05) Not includeda CKD 1.00 (0.97-1.01) Not includeda ESRD 1.40 (1.36-1.44) 1.57 (1.52-1.62) Liver disease 2.45 (2.38-2.52) 2.41 (2.43-2.48) Coagulopathy 2.76 (2.69-2.83) 2.19 (2.13-2.25) Obesity 0.61 (0.59-0.62) 0.72 (0.70-0.74) Abbreviations: CABG, coronary artery bypass graft; CAD, coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; DM, diabetes mellitus; ESRD, end-stage renal disease; HF, heart failure; ICD, implanted cardioverter-defibrillator; IQR, interquartile range; MI, myocardial infarction; PCI, percutaneous coronary intervention; PPM, pacemaker placement; PVD, peripheral vascular disease. aThe statistically insignificant variables on univariate analysis were not included in the multivariate logistic regression model. Bold texts indicate statistical significance. On multivariate logistic regression model, a diagnosis of COVID-19 (OR 5.67 [5.50 to 5.84]), liver disease (OR 2.41 [2.43 to 2.48]), coagulopathy (OR 2.19 [2.13 to 2.25]) and End-stage HF (OR 2.04 [1.87 to 2.23]) were independent predictors of in-hospital mortality. Interestingly, being a female (OR 0.90 [0.88 to 0.92]), history of hypertension (OR 0.81 [0.73 to 0.90]), CAD (OR 0.85 [0.83 to 0.87]), chronic heart failure (OR 0.63 [0.62 to 0.65]), and obesity (OR 0.72 [0.70 to 0.74]) were associated with lower odds of in-hospital mortality. Discussion To our knowledge, this is the largest study to analyze the outcomes of AHF admissions in the setting of COVID-19 infection, using a large national database. Our study shows that AHF in the setting of COVID-19 infection is associated with higher in-hospital mortality, as well as higher rates of AKI, need for pressors, and need for mechanical ventilation. Predictors of mortality in AHF include a diagnosis of COVID-19, liver disease, coagulopathy, end-stage HF, and COPD. We report significant variation in AHF-related mortality rate through the year 2020 with a peak in April 2020 followed by a significant decline throughout the rest of 2020. Acute circulatory failure has been described in COVID-19 infection, usually septic in etiology. However, those with preexisting cardiomyopathy may be at risk of AHF with or without cardiogenic or mixed etiology shock state. Acute onset of cardiomyopathy leading to denovo AHF can also occur due to etiologies such as atrial tachyarrhythmias, stress cardiomyopathy, and probably myocarditis. A history of heart failure has been reported as a comorbidity in 3.3% to 21% of patients with COVID-19. 8 It is noteworthy that only a minority of those suffer AHF during COVID-19 infection. For example, chronic heart failure was only noted in 22% of patients with AHF during COVID-19. 3 However, few patients develop AHF in the absence of a prior diagnosis of heart failure. A cohort study of 3080 patients admitted with COVID-19 showed the incidence of AHF was 11.2% in patients with prior history of CHF, and 2.1% in patients without prior history of CHF. 3 Another study of 191 patients with COVID-19 showed the incidence of heart failure overall was 23% and was even higher in nonsurvivors (52% vs 12%). 9 Early studies from Wuhan, China, also showed this, in a study of 131 deceased patients who died from COVID-19 reported heart failure as a complication in 49%. 10 In contrast to prior studies, we report a high prevalence of chronic heart failure (77%) in those with AHF and COVID-19. COVID-19 infection can result in increased mortality and higher complications in those with heart failure through multiple mechanisms. 11 One proposed mechanism is through the binding of SARS-CoV-2 virus to ACE-2 receptors, which are highly expressed in pericytes of adult human hearts. 4 The infection can then result in direct myocyte infiltration and inflammation, causing impaired cardiac function.4,11 ACE-2 receptors are usually upregulated in patients with heart failure making them more susceptible to SAR-CoV-2 virus-induced myocardial inflammation and necrosis. 2 SARS-CoV-2 virus can also cause damage to the endocardium by causing direct endothelial injury and micro-thrombosis.11–13 Additionally, COVID-19 infection causes excessive release of proinflammatory cytokines such as monocyte chemoattractant protein-1, interleukin-1β; interleukin-6; tumor necrosis factor-α. These proinflammatory cytokines can result in necrosis and death of myocardial cells.2,11,14 All of these, along with other yet unknown mechanisms, result in an increased risk of AHF in patients with COVID-19 infection. COVID-19 infection and AHF had a five-fold increased risk of in-hospital mortality compared to those without COVID-19. However, the mortality rate of 22% in our study is lower than previously reported in the cohort of AHF with COVID-19. Rey et al reported a high mortality rate of 46.8% in those who developed AHF during COVID-19 infection. 3 This is likely due to the difference in the study period between the two studies. The study was Rey et al was performed in the months of March and April of 2020, during the early phase of the pandemic. The mortality we report is for the entire calendar year of 2020, during which there was significant variation in mortality rates. We report a high mortality rate in similar months in the US as well. However, the mortality rate declined significantly in the US after the initial peak in April 2020. Berg et al reported a mortality rate of 43.8% in the AHF group with COVID-19 in a multicenter study from the US. 15 Meanwhile, the study only included patients admitted to the ICU, which is associated with increased mortality rate in COVID-19 infection. 16 Our study showed that patients with AHF and concomitant COVID-19 infection have an increased risk of developing AKI compared to heart failure patients without COVID-19 infection (aOR 1.33 [1.30 to 1.37]). Conversely, one could also hypothesize that AKI could also have resulted in AHF as a result of cardiorenal syndrome. A direct toxic effect of SARS-CoV2 virus resulting in acute tubular injury has been proposed. 17 The pathophysiological mechanism of acute tubular injury is similar to the mechanism of myocardial injury, which involves local inflammation and immune cell infiltration, cytokine storm, endothelial injury, podocyte injury, and microthrombi formation.18,19 Other nonspecific factors can contribute to development of AKI such as acute circulatory failure due to cardiogenic/mixed etiology shock,, hypoxia, and use of drugs with potential nephrotoxicity. 17 Although SARS-COV-2 virus has a wide range of impact on different body systems, it remains a predominantly respiratory illness. 20 Our study showed that patients with heart failure and concomitant COVID-19 infection have a higher risk of developing PE (aOR 1.75 [1.57 to 1.94]), pneumonia (aOR 3.25 [3.03 to 3.48]), and need for mechanical ventilation (aOR 1.46 [1.39 to 1.54]) compared to heart failure patients without COVID-19 infection. It has been reported that patients with HF were at over three times higher risk of mechanical ventilation. 21 We report a high crude mortality rate in AHF with COVID-19 in March and April of 2020. This is likely due to the COVID-19 surge that overwhelmed healthcare across the US. Further, the period of lockdown and COVID-19 transmission risk resulted in a barrier to timely access to care in patients with preexisting heart disease. To mitigate this, telemedicine was rapidly adopted by multiple institutions during the pandemic. The use of telemedicine in the heart failure population was associated with no increase in adverse outcomes. 22 We hypothesize that the decline in mortality during the latter months of 2020 could be due to improved systems of care from lessons learned from the early phase of the pandemic. A Danish study reported improved outcomes in terms of all-cause and cardiovascular mortality in heart failure patients with influenza vaccination. 23 A similar benefit with use of vaccination against COVID-19 remains to be studied in heart failure population. Strengths and Limitations Our study provides a contemporary view of patients who were hospitalized with AHF in the US during the COVID-19 pandemic. We report hospitalization trends across the year, clinical outcomes, procedural interventions, and utilization of hospital resources from costs and lengths, which may aid in establishing or changing the current health care policies and interventions. However, we also recognize several limitations in our study including, but not limited to, the inherent shortcomings of the NIS database as it is based on administrative claims for billing purposes using the ICD codes that is subject to coding errors, although specificity likely remains high. 24 Also, NIS does not have information on laboratory, procedural or echocardiographic data. Imaging data such as indices of left ventricular systolic and diastolic function and cardiac biomarkers are not available. Conclusions COVID-19 infection in patients with AHF is associated with significantly higher in-hospital mortality, need for mechanical ventilation, septic shock, and AKI along with higher resource utilization. Predictors for mortality in AHF patients included COVID-19 infection, patients with end-stage HF, and atrial fibrillation. Further studies are needed to investigate the impact of vaccination and COVID-19 variants on outcomes. Supplemental Material sj-docx-1-jic-10.1177_08850666231182380 - Supplemental material for COVID-19 Infection Is Associated With Increased In-Hospital Mortality and Complications in Patients With Acute Heart Failure: Insight From National Inpatient Sample (2020) Click here for additional data file. Supplemental material, sj-docx-1-jic-10.1177_08850666231182380 for COVID-19 Infection Is Associated With Increased In-Hospital Mortality and Complications in Patients With Acute Heart Failure: Insight From National Inpatient Sample (2020) by Anas Hashem, Amani Khalouf, Mohamed Salah Mohamed, Tarek Nayfeh, Ahmed Elkhapery, Mohammad Elbahnasawy, Devesh Rai, Himanshu Deshwal, Scott Feitell and Sudarshan Balla in Journal of Intensive Care Medicine Acknowledgments Central illustration was created with BioRender.com. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iDs: Anas Hashem https://orcid.org/0000-0002-2008-7110 Tarek Nayfeh https://orcid.org/0000-0001-9052-5537 Himanshu Deshwal https://orcid.org/0000-0003-2086-1281 Supplemental Material: Supplemental material for this article is available online. ==== Refs References 1 Lai C-C Shih T-P Ko W-C Tang H-J Hsueh P-R . Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and coronavirus disease-2019 (COVID-19): the epidemic and the challenges. Int J Antimicrob Agents. 2020;55 (3 ):105924. doi:10.1016/j.ijantimicag.2020.105924 32081636 2 Bader F Manla Y Atallah B Starling RC . Heart failure and COVID-19. Heart Fail Rev. 2021;26 (1 ):1–10. doi:10.1007/s10741-020-10008-2 32720082 3 Rey JR Caro-Codón J Rosillo SO , et al. Heart failure in COVID-19 patients: prevalence, incidence and prognostic implications. Eur J Heart Fail. 2020;22 (12 ):2205-2215. doi:10.1002/ejhf.1990 32833283 4 Chen L Li X Chen M Feng Y Xiong C . The ACE2 expression in human heart indicates new potential mechanism of heart injury among patients infected with SARS-CoV-2. Cardiovasc Res. May 1 2020;116 (6 ):1097-1100. doi:10.1093/cvr/cvaa078 32227090 5 Tajbakhsh A Gheibi Hayat SM Taghizadeh H , et al. COVID-19 and cardiac injury: clinical manifestations, biomarkers, mechanisms, diagnosis, treatment, and follow up. Expert Rev Anti-Infect Ther. 2021;19 (3 ):345-357. doi:10.1080/14787210.2020.1822737 32921216 6 Panhwar MS Kalra A Gupta T , et al. Effect of influenza on outcomes in patients with heart failure. ACC Heart Fail. 2019;7 (2):112-117. doi:10.1016/j.jchf.2018.10.011 7 Dong Y Peng C-YJ . Principled missing data methods for researchers. SpringerPlus. 2013;2 (1 ):222. doi:10.1186/2193-1801-2-222 23853744 8 Italia L Tomasoni D Bisegna S , et al. COVID-19 and heart failure: from epidemiology during the pandemic to myocardial injury, myocarditis, and heart failure sequelae. Front Cardiovasc Med. 2021;8 :713560. doi:10.3389/fcvm.2021.713560 34447795 9 Zhou F Yu T Du R , et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet. 2020;395 (10229 ):1054-1062. doi:10.1016/S0140-6736(20)30566-3 32171076 10 Chen T Wu D Chen H , et al. Clinical characteristics of 113 deceased patients with coronavirus disease 2019: retrospective study. Br Med J. 2020;m1091 :1-12. doi:10.1136/bmj.m1091 11 Adeghate EA Eid N Singh J . Mechanisms of COVID-19-induced heart failure: a short review. Heart Fail Rev. 2021;26 (2 ):363-369. doi:10.1007/s10741-020-10037-x 33191474 12 Marti CN Gheorghiade M Kalogeropoulos AP Georgiopoulou VV Quyyumi AA Butler J . Endothelial dysfunction, arterial stiffness, and heart failure. J Am Coll Cardiol. 2012;60 (16 ):1455-1469. doi:10.1016/j.jacc.2011.11.082 22999723 13 Libby P Lüscher T . COVID-19 is, in the end, an endothelial disease. Eur Heart J. 2020;41 (32 ):3038-3044. doi:10.1093/eurheartj/ehaa623 32882706 14 Tufan A Güler AA Matucci-Cerinic M . COVID-19, immune system response, hyperinflammation and repurposing antirheumatic drugs. Turk J Med Sci. 2020;50 (9 ):620-632.32299202 15 Berg DD Alviar CL Bhatt AS , et al. Epidemiology of acute heart failure in critically ill patients with COVID-19: an analysis from the critical care cardiology trials network. J Card Fail. 2022;28 (4 ):675-681. doi:10.1016/j.cardfail.2021.12.020 35051622 16 Rosenthal N Cao Z Gundrum J Sianis J Safo S . Risk factors associated with in-hospital mortality in a US national sample of patients with COVID-19. JAMA Network Open. 2020;3 (12 ):e2029058. doi:10.1001/jamanetworkopen.2020.29058 33301018 17 Legrand M Bell S Forni L , et al. Pathophysiology of COVID-19-associated acute kidney injury. Nature Reviews Nephrology. 2021;17 (11 ):751-764. doi:10.1038/s41581-021-00452-0 34226718 18 Ferlicot S Jamme M Gaillard F , et al. The spectrum of kidney biopsies in hospitalized patients with COVID-19, acute kidney injury, and/or proteinuria. Nephrol Dial Transplant. 2021. doi:10.1093/ndt/gfab042 19 Golmai P Larsen CP DeVita MV , et al. Histopathologic and ultrastructural findings in postmortem kidney biopsy material in 12 patients with AKI and COVID-19. J Am Soc Nephrol. 2020;31 (9 ):1944-1947. doi:10.1681/asn.2020050683 32675304 20 Desai AD Lavelle M Boursiquot BC Wan EY . Long-term complications of COVID-19. Am J Physiol Cell Physiol. 2022;322 (1 ):C1-C11. doi:10.1152/ajpcell.00375.2021 34817268 21 Alvarez-Garcia J Lee S Gupta A , et al. Prognostic impact of prior heart failure in patients hospitalized with COVID-19. J Am Coll Cardiol. 2020;76 (20 ):2334-2348. doi:10.1016/j.jacc.2020.09.549 33129663 22 Sammour Y Spertus John A Austin Bethany A , et al. Outpatient management of heart failure during the COVID-19 pandemic after adoption of a telehealth model. JACC: Heart Failure. 2021;9 (12 ):916-924. doi:10.1016/j.jchf.2021.07.003 34857175 23 Modin D Jørgensen ME Gislason G , et al. Influenza vaccine in heart failure. Circulation. 2019;139 (5 ):575-586. doi:10.1161/circulationaha.118.036788 30586760 24 Cooper LB Psotka MA Sinha S , et al. Specificity of administrative coding for older adults with acute heart failure hospitalizations. Am Heart J. 2020;223 :1-2. doi:10.1016/j.ahj.2020.02.004 32112871
PMC010xxxxxx/PMC10291224.txt	==== Front spteu TEU Tourism Economics 1354-8166 2044-0375 SAGE Publications Sage UK: London, England 10.1177_13548166231185899 10.1177/13548166231185899 Empirical Article Effects of the COVID-19 tourism crisis on the Spanish economy Vayá Esther https://orcid.org/0000-0003-2087-5106 Garcia José R Suriñach Jordi Pons Ernest 16724 University of Barcelona , Spain José R Garcia, AQR Research Group-Research Institute of Applied Economics, University of Barcelona, Avda Diagonal, 690, Barcelona 08034, Spain. Email: jrgarcia@ub.edu 24 6 2023 24 6 2023 13548166231185899© The Author(s) 2023 2023 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. This study addresses the impact of the COVID-19 pandemic on the Spanish tourism sector and economy in general, at the national and regional levels, through a comparative analysis between the evolution observed in the sector and the evolution that could have happened had the pandemic not occurred. This study was conducted in two stages. First, the total tourist expenditures for 2020 and 2021 were predicted under the assumption that the pandemic had not occurred. In the second stage, the losses in terms of turnover, jobs and the contribution of the tourism sector to the gross domestic product (GDP) that would have occurred without the pandemic were estimated. We applied the input–output method and found that for every €1000 less of tourist spending due to the pandemic, €1883 less were contributed to the GDP, and for every €100,000 less in spending, 2.8 jobs were lost. tourism COVID-19 tourist spending prediction economic impact input–output tables E65 C22 C67 L83 R15 Agència de Gestió d’Ajuts Universitaris i de Recerca (AGAUR). Generalitat de Catalunya. Project 2020PANDE00060 edited-statecorrected-proof typesetterts10 ==== Body pmcIntroduction The tourism sector, which is important to the Spanish economy and has high drag effects on other economic sectors, has been one of the industries most affected by the COVID-19 pandemic. In the European Union as a whole (European Union, 2021), the tourism sector employs 22.6 million people (11% of total EU employment) and accounted for 9.5% of the gross domestic product (GDP) in 2019. Satellite Accounts of Tourism (Instituto Nacional de Estadística [INE], 2022) indicate that, for Spain as a whole, the tourism sector employs 2.67 million people (12.7% of total Spain employment) and accounted for 157,355 million € (12.6% of its GDP) in 2019. 1 In addition, in some regions (or autonomous communities, CCAA) specialising in this sector, these percentages can be more than double. 2 COVID-19 has particularly affected this sector as limiting people’s mobility has been one of the main measures taken to tackle the pandemic. The objective of this article is to study the impact of the pandemic on the tourism sector and the economy in general, at the national and regional level, through a comparative analysis between the evolution observed in the sector in 2020 and 2021 (‘COVID Scenario’) and the evolution that this sector would have experienced had the pandemic not occurred (‘No-COVID scenario’). This objective was achieved in two stages. First, monthly information was collected for the total tourist expenditure for the maximum number of periods available, and the monthly series for the years 2020 and 2021 was predicted under the assumption that the pandemic had not occurred. Based on this information, we conducted a comparative analysis of the real evolution of the tourist expenditure variable and the no-cost scenario, thus detecting potential losses to the tourism sector. In the second stage, we estimated the potential economic impact derived from the reduction in tourism spending due to the COVID-19 pandemic in global terms for the entire economy, not only in the tourism sector. To do so, we used the input–output method, which considers the impact of a demand shock, such as a decrease in tourist spending, on all economic sectors – therefore, the multiplier and drag effects (indirect and induced) and not only the direct impact on tourism. In this way, and as in the first stage, the total impact results from both actual tourism spending in the years 2020 and 2021 and tourism spending that would have been recorded had there not been a pandemic. A comparison of the two scenarios allows to estimate the impact of the pandemic on the global economy. This estimate was conducted both at the national level and for a selection of Spanish regions (those for which information is available on total tourist spending, both for residents and non-residents) and was quantified based on the main economic macromagnitudes – production, employment and Gross Value Added (GVA), as well as its impact in terms of contribution to the GDP. Literature review Interest in analysing the impact of the COVID-19 pandemic on the tourism sector and the economy as a whole is undoubtedly high. Extensive literature has addressed this subject, including both the impact already caused by the pandemic and the expected impact. 3 With regards to Spain, Gago-García et al. (2021) analysed the impact on employment in the tourism sector, both from the territorial point of view (municipalities and provinces) and gender, concluding that the most affected territories are highly specialised in the sector (sun and beach tourism), especially due to the fall in international demand, while the most resilient have been rural and mountainous (due to the preferences of national tourism). Along the same lines, Arbulú et al. (2021a) highlight the decline in tourism – especially international – in 2020 and the disparate effect in the Spanish regions (given that national tourism can cover only 10% of overnight stays in the Balearic Islands and Canary Islands but 70% in Castilla-La Mancha). Rodousakis and Soklis (2022) analysed the impact of international tourism on the Spanish and German economies, 4 and Duro et al. (2021) confirmed that the most vulnerable territories in Spain are the Balearic and Canary Islands, the provinces on the Mediterranean coast and the Community of Madrid. More generally, Minondo (2021) analysed the impact of COVID-19 on trade in goods and services for Spanish provinces (with special reference to the tourism sector). Arbulú et al. (2021b) present points in common with the present paper, having analysed the impact on GDP using input–output models, although based on simulated data. Baños-Pino et al. (2021) also analysed the impact on the variable expenses per person – in this case, for the Spanish province of Oviedo. Internationally, the literature on the impact of the COVID-19 pandemic on the tourism sector and the economy as a whole is extensive. It includes, among others, Sen and Kovaci (2021) for Turkey, 5 Mursalina et al. (2022) for 22 provinces in Indonesia, 6 Škare et al. (2021) for world tourism as a whole and, especially, the European Union’s (2021) findings for the set of European tourist regions. All of them have only partially considered the period affected by the pandemic. Thus, Sen and Kovaci (2021) and Škare et al. (2021) only analysed the effect of the first wave of the pandemic (Demir et al., 2021; Rodousakis and Soklis, 2022) or on the effects in the summer of 2020 (Baños-Pino et al., 2021). Mursalina et al. (2022) and Minondo (2021) analysed data up to June 2021 and August 2021, respectively, as did the European Union (2021), with data up to the summer of 2021. Others, such as Aronica et al. (2022) and Škare et al. (2021), have tried to predict the medium- and long-term consequences of tourist arrivals, indicating that the effects would be heterogeneous depending on the characteristics of health systems, the severity of the shock and the level of uncertainty induced by the pandemic, with emerging and developing countries being the most affected. Arbulú et al. (2021a) highlight the importance of attracting national tourists as a substitute for international tourists. The same authors (2021b) analysed the forecasts for the drop in tourists in the Balearic Islands in 2020 and 2021, highlighting the high level of uncertainty that existed at that time about their evolution. A more up-to-date study is that of the European Union (2021), which examines the expected evolution in European tourist regions and highlights different evolutions according to the type of destination in a general framework of uncertainty. In any case, the diversity of intensity is confirmed, affecting more those who require air travel and are more dependent on international tourism, while the less affected regions are those of coastal and rural tourism and with stronger domestic markets. The methods used in impact analyses have also been various depending on the objectives. Thus, in addition to strictly descriptive analyses, Jayasinghe et al. (2021) use the Autoregressive Distributed Lag (ARDL) and Autoregressive Integrated Moving Average (ARIMA) models; Škare et al. (2021) adopted the Panel Structural Vector Auto-regression model (PSVAR);Veryadi Purba et al. (2021) used the simple regression model; Baños-Pino et al. (2021) adopted several variants of regression models (‘regression adjustment’, ‘inverse probability weighting regression’ and ‘propensity score matching’); Mursalina et al. (2022) chose regression models with panel data; Arbulú et al. (2021a) used the value-at-risk (VaR); Duro et al. (2021) relied on principal component analysis and regression models; and Arbulú et al. (2021b) adopted Monte Carlo simulations and probabilistic prediction models, also using input–output models. Finally, Rodousakis and Soklis (2022) used a multi-sector, single-production model and data from input–output tables, and Demir et al. (2021) used qualitative data obtained from face-to-face interviews. It should be noted that the present study has three points of interest. First, the period analysed was that of 2020 and 2021 (the years most affected by the pandemic) in their entirety. Second, it estimated the global economic loss that occurred as a result of the pandemic by comparing the situation observed in 2020 and 2021 with that which would have been obtained had the pandemic not happened, and the trend had continued to rise in recent years in terms of tourist spending. Third, the estimate of the impact is not only quantified considering the direct effects on the sector but also the indirect and induced effects on all sectors, which is a closer approximation to the true impact of the pandemic. To achieve the proposed objectives, we present the results of the first stage of the study, where the prediction of total tourist spending was made and compared with the real monthly spending for 2020 and 2021. Subsequently, we present the results of the second stage of the study, in which the loss due to the impact on the global economy derived from the reduction in tourist spending observed in 2020 and 2021 due to the pandemic was estimated. Finally, we present the conclusions. Comparative analysis of the evolution of total tourism expenditure This section presents an analysis of the difference between the real evolution of total tourist expenditure and that which would have occurred in the absence of a pandemic, distinguishing between residents in Spain and non-residents, to uncover differences in their behaviour. This analysis was conducted for the country as a whole as well as for the regions of Andalusia, the Balearic Islands, the Canary Islands, Catalonia, the Community of Madrid and the Community of Valencia – the only regions for which disaggregated expenditure information is available between residents and non-residents. Regional analysis is relevant because tourism typologies in different regions differ significantly: sun and beach tourism, water sports and outdoor activities and cultural, gastronomic and urban tourism. The results 7 allow to quantify the total impact of COVID-19 for 2020 and 2021 in terms of billing, GVA and level of occupancy. In relation to the analysis of the differential behaviour of total spending, the method used was based on comparing the real evolution of these series, as obtained from the Resident Tourism Survey (Familitur) and Tourist Expenditure Survey (Egatur), 8 with respect to a baseline scenario (the no-cost scenario). The method used to obtain the simulated series associated with this scenario was based on the JDEMETRA + application 9 and predicting the future series (March 2020 to December 2021) based on the historical series available before the pandemic. The periodicity of the analysed series was monthly. The JDEMETRA + software determines the best ARIMA model associated with each analysed time series, considering deterministic effects, such as seasonality and the Easter effect. Table 1 shows the variation in total tourist spending in 2020 and 2021 compared with spending in 2019, distinguishing between spending by residents and non-residents. As can be seen, for the national total, a reduction of 69.9% occurred in 2020 compared to 2019, and a certain improvement in 2021 is also observed, although the annual total remained 50.0% below that of 2019. Distinguishing tourists by origin, it is observed that the reduction in spending was much greater in the case of non-residents (78.5% in 2020 compared to 45.2% for residents), a fact that is explained by the much greater mobility restrictions placed on non-resident tourists than on resident tourists and the difference in the means of transportation used by both, with a clearly higher incidence of air transport in the case of non-residents. 10 Likewise, the recovery in 2021 appears to have been much more pronounced in residents (15.3% decrease compared to 2019) than in non-residents (62% reduction).Table 1. Variation of the total expenditure made with respect to the year 2019: residents and non-residents. Total Residents Non-residents Year 2020 Year 2021 Year 2020 Year 2021 Year 2020 Year 2021 Spain −69.9% −50.0% −45.2% −15.3% −78.5% −62.0% Andalusia −65.3% −45.1% −43.2% −13.8% −76.8% −61.6% Canary Islands −68.8% −53.3% −45.9% −16.4% −71.4% −57.4% Catalonia −76.8% −64.1% −46.8% −14.2% −82.9% −74.1% Valencian Community −66.3% −41.2% −46.6% −8.2% −74.1% −54.2% Balearic Islands −84.6% −46.7% −41.7% 15.4% −87.6% −51.0% Madrid’s community −74.1% −61.6% −60.8% −31.3% −77.9% −70.2% Source: Based on data from the Tourist Expenditure Survey (Egatur) and Resident Tourism Survey (Familitur), INE. At the regional level, homogeneity was observed in 2020, with the most notable exceptions being the Community of Madrid and the Balearic Islands. Thus, in relation to resident tourists in Madrid, the largest drop in tourist spending occurred in 2020 (60.8% reduction) compared to reductions between 41.7% and 46.8% in the rest of the regions. In the case of tourist spending by non-residents, the greatest drop occurred in the Balearic Islands, with a reduction of 87.6% compared to the 71.4% in the Canary Islands. On the other hand, 2021 showed greater diversity in the impact on tourist spending by residents – from a drop of 31.3% in the Community of Madrid compared to 2019 to an increase in spending in the Balearic Islands of 15.4% (in fact, in the case of the Balearic Islands, the spending of resident tourists in 2021 was higher than in 2019). In the case of non-resident spending, Catalonia and Madrid showed the greatest reductions compared to 2019 (74.1% and 70.2%, respectively), which contrasts with the 51% drop in the Balearic Islands. Figure 1 shows the behaviour of total tourist spending at the national level and for the six regions, comparing its real value (COVID Scenario) with the value of expenditure that would have been obtained in the absence of the pandemic (the no-cost scenario taken as a baseline). The conclusions are similar to those obtained after comparing the real series for 2020 and 2021 with the 2019 series. Thus, at the national level, it can be seen that the greatest losses in tourist spending occurred in the year 2020, with an expense value that does not reach 30% of the figure that would have been obtained in the case of no pandemic. Although the situation improved in 2021, total spending did not reach 50% of what it would have been in the no-cost scenario. The loss is much more pronounced in the case of spending by non-residents (with a value in 2020 of 20% of what would have been observed in a non-pandemic case) than in the case of residents (close to 55%). In 2021, resident spending was only 15% lower than it would have been in a non-pandemic situation, whereas this figure is almost 65% lower in the case of non-residents.Figure 1. Total tourist expenditure. No-COVID Scenario = 100. Source: Based on data from the Tourist Expenditure Survey (Egatur) and Resident Tourism Survey (Familitur), INE. At the regional level, results are consistent with those obtained at the national level, although certain differences were detected. Thus, it can be seen that, in 2020, the regions that showed the greatest loss of spending compared to a non-pandemic situation were Catalonia (77% lower), the Community of Madrid (76%) and, especially, the Balearic Islands (85% lower). In the case of Madrid, this fact reflects both what happened in terms of spending by residents (in fact, it is the region with the greatest relative loss compared to the rest, with spending by residents being 60% lower than what it would have been without a pandemic) and non-resident spending (76% lower). In the cases of Catalonia and the Balearic Islands, the situation is similar, although both stand out for spending by non-residents, which is much lower than the theoretical figures without the pandemic (83% and 88%, respectively). In 2021, the gap with respect to the no-COVID scenario decreases in all regions compared to the previous year – especially in the Balearic Islands, where tourist spending was 46% lower than the theoretical value without a pandemic. This reflects the extraordinary recovery in spending by residents in this community, which led to tourism spending being 8% higher than it would have been without the pandemic. Similarly, the Balearic Islands show the smallest gap in terms of spending by non-residents (recovering almost 50% of the spending figure under the no-cost scenario). It should also be noted that Madrid is the region that showed the least recovery in 2021 – 33% lower than the theoretical figure for resident spending in the case of a non-pandemic situation (15% in Spain) and 76% in the case of non-residents. Estimate of the total economic impact derived from the reduction of tourism spending due to the pandemic Once tourist expenditure predictions for 2020 and 2021 were obtained, the second stage estimated the total economic impact associated with tourist expenditure. For this estimation, a demand approximation based on the application of the input–output method was followed (Pulido and Fontela, 1993; Miller and Blair, 2009; Xie et al., 2018), which is commonly used in this type of study (Murillo et al., 2013; CLIA Europe, 2015; CERTeT Bocconi, 2015; Vaya et al., 2018; IVIE, 2019). Brief description of the method used to estimate the impact In the input–output method, the total economic impact is defined as the aggregation of three types of effects: direct, indirect and induced. In this case, the direct effect captures all the activities generated in a territory in response to tourist demand. However, the indirect effect captures the additional activity generated in the territory to cover the demand for the goods and services required by companies to develop their own activities. Thus, companies that supply these goods and services see their billings increase and, in turn, generate new multiplier effects on other companies, from which they also request other goods and services to conduct their activities. Finally, the induced effect arises as a consequence of the purchase of goods and services made by workers who, directly and indirectly, owe their jobs to the expenditure made by travellers in their territory. As a result of these indirect and induced effects, the total impact is clearly higher than the direct impact, not only in magnitude but also in terms of sectoral impact. This economic impact has been calculated for the main macromagnitudes of an economy, such as billing, gross value-added generation and full-time equivalent jobs (FTEs). The procedure was as follows. First, we compiled the latest symmetric input–output tables (IOT) available for each analysed territory: Spain (IOTS-2016), Andalusia (IOTA-2016), Catalonia (IOTC-2011), Valencian Community (IOTCV-2000), Balearic Islands (IOTBI-2014), Canary Islands (IOTCI-2005) and the Community of Madrid (IOTM-2010). 11 Next, we obtained the matrices of technical coefficients, extended technical coefficient matrix, inverse Leontief matrix and inverse extended Leontief matrix for each case. In the second stage, we assigned the sectoral distribution of tourism spending observed in 2019 and 2020 (COVID Scenario) and 2021 (COVID Scenario), and for 2020 and 2021, the tourism spending predicted under the no-cost scenario. For this distribution, we used the Tourism Satellite Accounts of Spain for 2019 and 2020, 12 assigning expenses to the following sectors: accommodation and food and beverage services; transport services; travel agency services; tour operators; creative, artistic and entertainment services; sports, recreation and entertainment services; rental services; real estate services; and retail trade services. 13 In the third stage, once the direct impact vector was defined in terms of turnover 14 for each territory, year and scenario, the remaining direct impacts were estimated: GVA-generated and FTE jobs. 15 In the fourth stage, the estimate of the corresponding indirect impact was obtained by applying the input–output method, while the induced impact was estimated in the fifth stage. The estimated total impact was obtained for each territory, year and scenario as an aggregation of the three previously calculated impacts – direct, indirect and induced. In this way, it was possible to estimate the economic losses in each territory due to the pandemic. As the objective was to estimate the impact of COVID-19 on the tourism sector and economy in general at the national (Spain) and regional levels, we compare the impact of tourist spending actually recorded in 2020 and 2021 with respect to what would have occurred had the pandemic not existed. Impact derived from tourist spending: National total Table 2 presents the estimates obtained from the total economic impact derived from tourism spending in terms of billing, GVA, number of FTE jobs and contribution to GDP for 2020 and 2021, following the method explained above. This table shows the results of the impact associated with both the year 2019 and the 2020–2021 period observed from the real tourist expenditure of those 2 years (COVID Scenario 2020 and COVID Scenario 2021), as well as the impact associated with 2020 and 2021 obtained from the forecast of tourism spending for those 2 years under the assumption of no pandemic (no-COVID scenario 2020 and no-COVID scenario 2021).Table 2. Estimate of the impact derived from Tourism Expenditure: Spain. Tourist spending (€M) (1) Total billing (2) (€M) (1) Total jobs (2) FTE (3) Total GVA (2) (€M) (1) Contribution to GDP (%) Year 2019 127,347 462,015 3,753,987 235,434 20.3% (4) ‘COVID’ 2020 38,500 139,679 1,205,555 69,093 6.9% ‘COVID’ 2021 62,012 224,982 1,941,787 111,288 10.2% ‘No COVID’ 2020 130,631 473,930 3,850,803 241,506 21.1% (5) ‘No COVID’ 2021 133,957 485,999 3,948,868 247,656 20.6% (5) ‘COVID’ 2020–2019 −88,846 (−69.8%) −322,335 (−69.8%) −2,548,432 (−67.9%) −166,341 (−70.7%) −13.4 pp (6) ‘COVID’ 2021–2019 −65,334 (−51.3%) −237,033 (−51.3%) 1,812,199 (−48.3%) −124,146 (−52.7%) −10.1 pp (6) ‘COVID’ – ‘No COVID’ 2020 −92,131 (−70.5%) −334,251 (−70.5%) −2,645,248 (−68.7%) −172,413 (−71.4%) −14.2 pp (6) ‘COVID’ – ‘No COVID’ 2021 −71,945 (−53.7%) −261,018 (−53.7%) −2,007,080 (−50.8%) −136,368 (−55.1%) −10.4 pp (6) Source: The author’s elaboration based on expenditure data from the Tourist Expenditure Survey (Egatur) and the Resident Tourism Survey (Familitur), INE. 1All monetary figures are expressed in millions of € (€M) at 2021 prices. 2The total impacts are the result of adding the direct, indirect and induced impacts. 3The employment data correspond to full-time equivalent jobs (FTEs). 4Said estimate is higher than the 12.4% contribution of tourism to the GDP of the Spanish economy estimated by the INE in the Spanish Tourism Satellite Account for 2019 (INE). 5To obtain the contribution to GDP, the predictions of the BBVA (Regional Observatory of Spain) made in 2019 for the GDP of the years 2020 and 2021 have been used. 6Percentage point (PP). This is due, among other reasons, to the fact that, unlike the estimates presented here, said impact is estimated considering the direct and indirect effects of tourism spending (without considering the induced effects). The estimate obtained in this study when only direct and indirect effects were considered was 11.9%, which is very close to the previous value of 12.4%. If only the direct effects were considered, its contribution would be 6.02%. As can be seen, the predictions show that if the pandemic had not occurred, tourist spending in 2020 would have been €130,631 million (close to €127,347 million in 2019) – a much higher figure than that actually observed in that year, which was close to €39,000 million. Thus, in 2020, real tourist spending would have been 70.5% lower than that predicted under the no-cost scenario. In 2021, these differences, although minor, were still notable, such that predicted tourism spending without the pandemic would have reached almost €134,000 million compared to €62,000 million in actual spending (€72,000 less, 53.7% lower than predicted). If, based on the amount of tourist expenditure, the total impact is estimated (considering direct, indirect and induced effects), it is observed that in 2020, the estimated real total billing 16 would be €139,679 million compared to €473,930 million, which would have been invoiced in the absence of a pandemic. In this way, the loss in terms of total billing would have been greater than €334,000 million, or approximately €915 million per day. In addition, it should be noted that the reduction in tourist spending observed as a result of the restrictions derived from the pandemic had a significant impact, affecting not only the branches of activity directly linked to tourism but also all other economic sectors. In turn, the total theoretical jobs that would have existed in 2020 under the COVID Scenario would have been 1.2 million compared to 3.85 in the no-COVID scenario; thus, it is estimated that a maximum of 2.65 million jobs would have been ‘lost’. It must be taken into account that these estimates of jobs are obtained by applying the ratio of jobs per thousand euros of production that is derived from the symmetrical input–output table for Spain corresponding to 2016. Therefore, the estimated changes in jobs are consistent with the linear reductions in billing, assuming a labour market without friction, restrictions or interventions. For this reason, these variations do not necessarily reflect what happened in the labour market. In this regard, the total non-elasticity of variations in employment to variations in production, together with the consolidation of the figure of the Temporary Employment Regulation Files (TERF) in Spain during the pandemic, means that the estimates presented for variations in work positions derived from the decrease in tourist spending must be considered a theoretical estimate and must be considered in any case as the maximum decrease that could have occurred. Analysing the impact in terms of GVA, it is found that the contribution to GDP under the COVID Scenario would have been €69,093 million compared to €241,506 million under the no-COVID scenario – a loss of €172,413 million. If these last figures are relative to the GDP actually observed in Spain in 2020 and to that predicted for that year 17 (had the pandemic not occurred), the tourism sector would have contributed 21.1% to the national GDP in the no-COVID scenario – 14.2% points higher than what was actually observed (6.9%). If the same analysis is replicated for 2021, it can be seen that as a result of the pandemic, the lower tourist spending observed compared to that predicted under the no-COVID scenario has caused a loss of slightly more than €261,000 million in turnover, 2 million fewer theoretical jobs and almost €136.5 billion less of contribution to the GDP. Thus, if, under the COVID Scenario, the contribution of the tourism sector in that year was quantified at 10.2% of the observed GDP, this figure would have been 20.6% in the absence of a pandemic (10.4% points less). If, based on the previous results, a sectoral disaggregation of the impacts is conducted, and the impacts relative to the subgroup of tourism sectors are calculated on the one hand 18 and those relative to the subgroup that contains the rest of the sectors on the other, one obtains the information shown in Table 3. As can be seen, while direct billing losses (i.e. tourist spending) are concentrated in all cases in the tourism sector group, in the case of the rest of the estimated magnitudes, these losses affect all sectors.Table 3. Estimate of the losses derived from the reduction of tourist expenditure for the national total. Comparison between grouping of sectors. Comparative estimated impacts ‘COVID’ Scenario vs ‘No-COVID’ Scenario, Year 2020. Comparison estimated impacts scenario ‘COVID’ vs ‘No COVID’ Scenario, year 2021. Losses in direct billing (M€) (1) Tourism sectors −92,131 −71,945 Other sectors 0 0 Total −92,131 −71,945 Billing losses total (M€) (1) Tourism sectors −148,616 −116,055 Other sectors −185,635 −144,963 Total −334,251 −261,018 Total job losses (2) Tourism sectors −1,480,409 −1,109,819 Other sectors −1,164,839 −897,261 Total −2,645,248 −2,007,080 GVA losses total (M€) (1) Tourism sectors −86,126 −68,910 Other sectors −86,287 −67,458 Total −172,413 −136,368 Source: The author’s elaboration based on expenditure data from the Tourist Expenditure Survey (Egatur) and the Resident Tourism Survey (Familitur), INE. 1All monetary magnitudes are expressed in millions of € and in 2021 prices. 2The employment data correspond to full-time equivalent jobs and are estimated from the input–output tables. When the comparison focuses on the estimated impact under the real COVID Scenario compared to the predicted Non-COVID Scenario for 2020, the results show, after the €92,131 million in lower tourist spending, an estimated loss of €148,616 million of total turnover in the case of the tourism sectors (€185,635 million for the other sectors), 1.5 million less theoretical FTE jobs (1.2 million for the other sectors) and €86,126 million less GVA (€86,287 million for the other sectors). Upon repeating the same exercise but for 2021 (€71,945 million of less tourist spending), the results show estimated losses for the tourism sectors of €116,055 million of total turnover, 1.1 million jobs theoretical FTE and €68,910 million of GVA (for the other sectors, these figures would be €144,963 million less total turnover, 897,261 fewer theoretical jobs and €67,458 million less GVA). Finally, if the results of the comparison between the COVID and the no-COVID scenarios are related to the loss of tourist spending, it can be concluded that, on average global terms for 2020–2021, for every €1000 that tourists did not spend in the country due to COVID-19-related mobility restrictions, Spain stopped billing €3628 and decreased its contribution to the GDP by €1883. In turn, for every €100,000 less that these tourists spent, a maximum of 2.8 FTE jobs would have been lost (data on estimated losses in the absence of TERF and under the assumption of a labour market without restrictions, friction or interventions). Impact derived from tourism spending: Regional analysis After analysing the impact of the pandemic generated by a drastic reduction in tourist spending on the national total, we briefly analyse the six regions for which complete information on tourist spending is available for both residents and non-residents. To contextualise the different starting situations, Table 4 shows a comparison of the impact of tourism spending for 2019 for the regions analysed. Catalonia is the region with the highest tourist spending in 2019 (more than €26,000 million), followed by Andalusia and the Canary Islands (a little more than €19,000 million), while the Communities of Madrid and Valencia show the lowest expenditure in absolute terms (close to €14,000 million). When the indirect and induced billings generated are added, it is found that due to tourism spending, Catalonia managed to bill almost €60,000 million in total – a figure close to €58,000 million for the Canary Islands and to €55,000 million in Andalusia. The Communities of Madrid and Valencia are located in the lower band, with a total turnover of slightly over €28,000 million. If direct and total billings are compared, it can be concluded that the Canary Islands and Andalusia have the greatest multiplier effect; thus, for every €1000 of direct billing, an additional €2000 of billing is generated indirectly and induced in the Canary Islands and €1800 in Andalusia. In contrast, Valencia and Madrid have the smallest multiplier effect: For every 1000 direct billings, 900 and 1100 additional billings are generated, respectively.Table 4. Estimated economic impact derived from Tourism Expenditure for the year 2019: regional comparison. Andalusia Catalonia Community Valencian Balearic Islands Canary Islands Community from Madrid Tourist spending (M€) (1) 19,491.8 26,218.5 13,725.9 16,401.6 19,247.4 13,745.8 Total turnover (M€) (1) 54,913.9 59,790.0 28,381.4 36,761.8 57,756.1 28,267.1 Total jobs (FTE) (2) 528,393 509,450 308,461 341,046 623,572 247,470 Total contribution to GDP (%) 19.4% 14.6% 15.4% 66.5% 74.8% 6.5% Source: The author’s elaboration based on expenditure data from the Tourist Expenditure Survey (Egatur) and the Resident Tourism Survey (Familitur), INE. 1All monetary magnitudes are expressed in millions of € and in 2021 prices. 2The employment data correspond to full-time equivalent jobs and are estimated from the input–output tables. In terms of employment, the Canary Islands and Andalusia show the greatest impact in terms of total jobs (623,572 and 528,393, respectively). Madrid generates the fewest jobs derived from tourism spending (130,813 direct and 247,470 total). When analysing the contribution to GDP derived from tourist spending, it is clear that the Balearic and Canary Islands have a higher total contribution and are far removed from the rest. Thus, tourist spending contributes 74.8% to the GDP of the Canary Islands and 66.5% to that of the Balearic Islands. These figures contrast with the values between 14.6% and 19.4% of the total contribution to the GDP of Andalusia, Catalonia and the Community of Valencia, while, in the case of Madrid, tourist spending contributes 6.5% overall. Table 5 compares the impacts of actually observed tourism spending (COVID Scenario) with the tourism spending that would have occurred in the absence of the pandemic (no-cost scenario) in 2020 and 2021. The highest direct spending and total billing losses in absolute terms occurred in Catalonia, while the highest relative losses occurred in the Balearic Islands. In the case of jobs, the greatest job losses in absolute terms occurred again in Catalonia but above all in the Canary Islands, whereas the greatest losses in relative terms occurred in the Balearic Islands. In relation to the contribution of the tourism sector to the GDP, it can be seen that if the pandemic had not occurred, its contribution would have been close to 70% in the case of the Balearic and Canary Islands, whereas, under the COVID Scenario, it decreased to 26.4% in the case of the Canary Islands (42.8% points lower than expected in the case of no pandemic) and 13.3% in the Balearic Islands (54.1% points less).Table 5. Estimated economic impact derived from Tourism Expenditure (2020–2021). Regional comparison between the ‘No-COVID’ and ‘COVID’ Scenario. Andalusia Catalonia Com. Valencian Balearic Islands Canary Islands Madrid’s community Figures absolute Relative variation Figures absolute Relative variation Figures absolute Relative variation Figures absolute Relative variation Figures absolute Relative variation Figures absolute Relative variation SPENT TOURIST (1) ‘COVID’ 2020 6786 6095 4644 2522 5966 3575 ‘No COVID’ 2020 20,880 26,429 14,858 16,344 18,945 14,923 ‘COVID’ 2021 10,392 9182 7850 8469 8752 5155 ‘No COVID’ 2021 21,945 27,136 15,994 16,229 19,334 16,295 Var. ‘COVID’ vs ‘No COVID’ 2020 −14,095 −67.5% −20,333 −76.9% −10,214 −68.7% −13,823 −84.6% −12,979 −68.5% −11,349 −76.0% Var. ‘COVID’ vs ‘No COVID’ 2021 −11,553 −52.6% −17,954 −66.2% −8143 −50.9% −7760 −47.8% −10,582 −54.7% −11,141 −68.4% BILLING TOTAL (1) ‘COVID’ 2020 19,117 13,900 9602 5,652 17,903 7351 ‘No COVID’ 2020 58,826 60,269 30,722 36,633 56,848 30,688 ‘COVID’ 2021 29,276 20,940 16,233 18,983 26,263 10,600 ‘No COVID’ 2021 61,825 61,882 33,071 3,376 58,017 33,510 Var. ‘COVID’ vs ‘No COVID’ 2020 −39,709 −67.5% −46,369 −76.9% −21,121 −68.7% −30,981 −84.6% −38,945 −68.5% −23,337 −76.0% Var. ‘COVID’ vs ‘No COVID’ 2021 −32,549 −52.6% −40,943 −66.2% −16,838 −50.9% −17,393 −47.8% −31,754 −54.7% −22,910 −68.4% TOTAL JOBS (FTE) (2) ‘COVID’ 2020 196,024 125,202 113,090 56,263 208,198 68,038 ‘No COVID’ 2020 566,037 513,532 333,900 340,874 613,771 268,666 ‘COVID’ 2021 300,190 188,607 191,191 188,966 305,408 98,115 ‘No COVID’ 2021 594,892 527,280 359,427 338,478 626,390 293,373 Var. ‘COVID’ vs ‘No COVID’ 2020 −370,012 −65.4% −388,330 −75.6% −220,810 −66.1% −284,611 −83.5% −405,573 −66.1% −200,627 −74.7% Var. ‘COVID’ vs ‘No COVID’ 2021 −294,702 −49.5% −338,672 −64.2% −168,235 −46.8% −149,512 −44.2% −320,981 −51.2% −195,258 −66.6% CONTRIBUTION TO GDP ‘COVID’ 2020 7.6% 4.0% 5.9% 13.3% 26.4% 1.9% ‘No COVID’ 2020 21.1% 14.9% 16.9% 67.5% 69.2% 7.1% ‘COVID’ 2021 10.6% 5.7% 9.4% 39.2% 36.0% 2.5% ‘No COVID’ 2021 20.9% 15.7% 17.2% 63.6% 67.8% 7.4% Var. ‘COVID’ vs ‘No COVID’ 2020 −13.5 −64.0% −10.9 −72.9% −11.1 −65.3% −54.1 −80.2% −42.8 −61.9% −5.2 −73.6% Var. ‘COVID’ vs ‘No COVID’ 2021 −10.3 −49.3% −10.0 −63.8% −7.8 −45.2% −24.4 −38.4% −31.9 −47.0% −4.9 −66.0% Source: Own elaboration based on expenditure data from the INE Tourism Expenditure Survey (Egatur) and Resident Tourism Survey (Familitur) and predicted GDP data from BBVA. 1All monetary magnitudes are expressed in millions of € and at 2021 prices. 2The employment data corresponds to full-time equivalent jobs. In turn, if the comparison between the COVID Scenario and the no-COVID scenario is related to the loss of tourist spending in each region, it can be concluded that, on average global terms, for every €1000 that tourists stopped spending as a consequence of mobility restrictions, a total of between a minimum of €2056 euros in the Community of Madrid (€1073 of contribution to GDP) and a maximum of €3001 in the Canary Islands 19 (€1677 of contribution to GDP) would have been lost. In turn, for every €100,000 euros less spent by tourists, between a minimum of 1.8 FTE jobs in the Community of Madrid and a maximum of 3.1 jobs in the Canary Islands (if there were no TERF and under a labour market with the aforementioned characteristics of no friction, no restrictions and no interventions) would have been lost. Conclusions This study aimed to examine the impact of the pandemic on the tourism sector and economy in general through a comparative analysis between the evolution actually observed in the sector in 2020 and 2021 and the evolution that would have happened had the pandemic not occurred. The analysis was conducted for the national total and at the regional level, choosing six regions for which representative information on tourist spending by both residents and non-residents was available. The analysis shows results of interest. First, an estimate is offered on what the economic impact would be if a situation like the one experienced were to occur again and mobility restrictions such as those imposed during the pandemic were applied. Second, it was possible to approximate what the potential job losses might have been if the TERF had not been applied. Third, due to multiplier effects (indirect and induced impacts), the drastic reduction in tourism spending has had notable repercussions that are not confined to typical tourism sectors but extend to the entire economy. The loss of potential turnover in the Spanish economy could have tripled the loss of direct tourist spending. In addition, the significant impact of the pandemic has been verified, especially in regions such as the Balearic and Canary Islands, where the year before the pandemic the contribution of tourism spending to GDP had been 66.5% and 74.8%, respectively; in other regions, the impact is lower in relative terms but also considerable. Thus, as the Spanish economy depends to a large extent on tourism, it is important to take measures to guarantee its sustainability and diversify the economy to reduce vulnerability to health, economic or tourism sector crises. Finally, it should be noted that the analysis has limitations. The first one is related to the use of the input–output method, whose starting hypothesis conditions the results obtained (constant technical coefficients, constant returns to scale, absence of restrictions on productive factors, no substitution between inputs, no response of wages to variations in demand, static single impact etc.). Likewise, the estimates made only allude to the impact of the pandemic in terms of the reduction in tourist spending and its consequences for the economy but do not consider other impacts caused in other sectors where the effect was the opposite (the pandemic generated, e.g. increased activity in sectors such as health or those related to information and telecommunications). Nevertheless, estimating a general equilibrium model to overcome the aforementioned limitations is highly complex, especially as regards its estimation at the regional level. Author biographies José Ramón García (jrgarcia@ub.edu) is an Associate Professor at the Department of Econometrics, Statistics and Applied Economy of the University of Barcelona (Spain). He obtained PhD in Economics and Management from the University of Barcelona, and he is a member of the Regional Quantitative Analysis Research Group (AQR-IREA research group). He is specialised in impact studies. He has participated in different competitive projects related to impact studies, and he published various papers about this topic, specially focused on tourism economic impact. Ernest Pons Fanals (epons@ub.edu) is an Associate Professor of Applied Economics at the University of Barcelona. He has a graduate degree in Economics and obtained PhD in Applied Economics and MA in Education. He is an expert in econometrics and statistical analysis of time series. His research has focused on economic forecasting, economic indicators, measuring economic impacts and quantitative analysis of the economic situation. Jordi Suriñach (jsurinach@ub.edu) is a full-time professor of Applied Economics and Director at the Department of Econometrics, Statistics and Applied Economy at the University of Barcelona. He obtained PhD in Economics and Management Science from the same university. He is a member of the Regional Quantitative Analysis Research Group (AQR-IREA). His expertise is related with regional and urban issues, focused on quantitative techniques, economic impact, public policy evaluation, economic forecasts and tourism. His research activity has been published in more than 125 scientific articles (74 Web of Science and 69 Scopus), and he participated and coordinated several International, European and national Research projects. Esther Vaya (evaya@ub.edu) is an Associate Professor at the Department of Econometrics, Statistics and Applied Economy of the University of Barcelona (Spain). She obtained PhD in Economics and Management from the University of Barcelona, and she is a member of the Regional Quantitative Analysis Research Group (AQR-IREA research group). She is specialised in impact studies and spatial econometrics. She has participated in a lot of competitive projects related to impact studies, and she has published various papers about this topic, specially focused on tourism economic impact. ORCID iD José R Garcia https://orcid.org/0000-0003-2087-5106 Notes The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Agència de Gestió d’Ajuts Universitaris i de Recerca (AGAUR). Generalitat de Catalunya.; Project 2020PANDE00060. 1. The Tourism Satellite Accounts quantify the impact of the tourism sector by adding two effects: the direct effect, which considers only those sectors that are recipients of tourism demand, and the indirect effect, which also considers all the effects on the rest of the economic sectors that appear as a consequence of the intersectoral relations of purchase of goods and services arising within the productive process. 2. For example, in IBESTAT (2022), the Official Institute of Statistics of the Balearic Islands, in the Input-Output Framework (2014), estimated that the tourism sector accounts for 34.8% of the GDP of the Balearic Islands. 3. Utkarsh and Sigala (2021) compiled 177 articles published up to January 2021 and grouped them into four major themes, one of which is the impact of COVID-19 on tourism and hotel agents (especially on the well-being of employees and employers). 4. Quantifying a fall, in the year 2020, of 0.58% in German GDP and 4.54% in Spanish. 5. The impact is a 69% decline in tourist arrivals and a 65% decline in tourism receipts, and the foreign trade deficit coverage rate of tourism receipts in Turkey decreased by almost 80% in 2020. 6. It examines the effect of the COVID-19 pandemic on tourist arrivals and investment (domestic and foreign investment) and on the economic growth of the tourism sector in Indonesia. 7. The authors can provide the numerical results of the actual values and the predictions. 8. Both surveys come from the INE. 9. JDEMETRA + is a software application developed by the National Bank of Belgium in collaboration with the Deutsche Bundesbank and Eurostat, and it is officially recommended to members of the European Statistical System and the European System of Central Banks for seasonal and calendar adjustment in official statistics. 10. Although the results are not included in this article, the series of travelers and average spending per traveler have also been analysed. In this sense, it should be noted that the average cost per tourist is clearly higher in the case of non-residents than residents. The larger decrease in the number of non-resident tourists, together with their higher spending, would explain the greater impact in terms of total spending by non-residents than by residents. 11. See INE (2016), Institute of Statistics and Cartography of Andalusia (2016), Idescat (2011), IVIE (2000), IBESTAT (2022), ISTAC (2005) and IE (2010). 12. It was decided to use the Spanish Tourism Satellite Account to conduct the sectoral distribution of tourist spending in all the territories analysed since not all the selected regions had updated their Satellite Accounts for the years of study. 13. The item ‘other non-characteristic products’ was assigned to the retail trade sector. 14. The direct billing generated corresponds to the identified tourist spending. 15. In all cases, the corresponding price-level correction was made according to the year of disposal of each IOT used. 16. It is important to emphasise that this decrease in billing is estimated as a consequence of the reduction in tourist spending. However, this estimate does not reflect the final change in the country’s billing since, as a result of the pandemic, other non-tourism economic sectors also saw their business significantly affected, both positively and negatively. 17. The predictions of the BBVA (Regional Observatory of Spain) made in 2019 for the GDP of 2020 and 2021 were used. 18. It should be noted that the sectors defined as tourism and in which the direct impact of tourism spending is concentrated are the following: retail trade services, except for motor vehicles and motorcycles; ground transportation services; maritime and inland waterway transport services; air transport services; accommodation and food and drink services; real estate services; rental services; travel agency, tour operator and other reservation services; creative, artistic and entertainment services, library, archive, museum and other cultural services; and sports, recreation and entertainment services. 19. Andalusia would closely follow the Canary Islands, with a total turnover loss of €2817 for every €1000 of lost tourist spending. ==== Refs References Arbulú I Razumova M Rey-Maquieira J , et al. (2021a) Can domestic tourism relieve the COVID-19 tourist industry crisis? The case of Spain. Journal of Destination Marketing and Management 20 : 100568. DOI: 10.1016/j.jdmm.2021.100568 10.1016/j.jdmm.2021.100568 Arbulú I Razumova M Rey-Maquieira J , et al. (2021b) Measuring risks and vulnerability of tourism to the COVID-19 crisis in the context of extreme uncertainty: the case of the Balearic Islands. Tourism Management Perspectives 39 : 100857. DOI: 10.1016/j.tmp.2021.100857 10.1016/j.tmp.2021.100857 34580625 Aronica M Pizzuto P Sciortino C (2022) COVID-19 and tourism: what can we learn from the past? The World Economics 45 : 430–444 Special Issue: COVID 19, Trade and Trade Policy. DOI: 10.1111/twec.13157 10.1111/twec.13157 Baños-Pino JF Boto-García D Del Valle E , et al. (2021) The impact of COVID-19 on tourists’ length of stay and daily expenditures. Tourism Economics 29 (2 ): 437–459. DOI: 10.1177/13548166211053419 10.1177/13548166211053419 CERTeT Bocconi (2015) Analysis dell’impatto Socio - Economic Delle Attivita Crocieristiche of the Port of Civitavecchia, PORTI of ROMA and LAZIO. Milan, Italy: Università Commercial Luigi Bocconi. CLIA Europe (2015) The Cruise Industry. Contribution of Cruise Tourism to the Economies of Europe. 2015 Edition. Brussels: Cruise Lines International Association. Demir M Demir ŞŞ Dalgıç A , et al. (2021) Impact of COVID-19 pandemic on the tourism industry: an evaluation from the hotel managers' perspective. Journal of Tourism Theory and Research 7 (1 ): 44–57. DOI: 10.24288/jttr.857610 10.24288/jttr.857610 Duro JA Perez-Laborda A Turrion-Prats J , et al. (2021) Covid-19 and tourism vulnerability. Tourism Management Perspectives 38 : 100819. DOI: 10.1016/j.tmp.2021.100819 10.1016/j.tmp.2021.100819.34873568 European Union (2021) Regional Impacts of the COVID-19 Crisis on the Tourist Sector (CCI 2020CE16BAT074). Final Report. Directorate-General for Regional and Urban Policy. DOI: 10.2776/179573. Gago-García C González-Relaño R Serrano Cambronero M , et al. (2021) Impact of the COVID-19 crisis on employment in the tourism sector in Spain: territorial and gender perspectives. Bulletin of the Association of Spanish Geographers 91 . DOI: 10.21138/bage.3162 10.21138/bage.3162 IBESTAT (2022) Institute of statistics of the balearic Islands (IBESTAT). Input-Output Framework of the Balearic Islands. https://ibestat.caib.es/ibestat/page?&p=px_tablas&nodeId=1fd68a0c-76ad-4d7e-acaf-db32e51c7ba7&path=economia%2Fcomptes-economiques%2Fmarc-input-output&lang = is. Idescat (2011) Statistical institute of Catalonia (Idescat). Input-Output Framework of Catalonia. https://www.idescat.cat/estad/mioc?lang=es. IE (2010) Statistics institute. Input-Output Framework of the Community of Madrid. https://www.madrid.org/iestadis/fijas/estructu/economicas/contabilidad/estructumio.htm. INE (2016) National Statistics Institute. Annual National Accounts of Spain: Input-Output Tables. https://www.ine.es/dyngs/INEbase/es/operacion.htm?c=Estadistica_C&cid=1254736177058&menu=resultados&idp=1254735576581. INE (2022) National statistics institute. Tourism Satellite Accounts of Spain. https://www.ine.es/consul/serie.do?d=true&s=DCE4. Institute of Statistics and Cartography of Andalusia (2016) Input-output framework of Andalusia 2016. http://www.juntadeandalucia.es/institutodeestadisticaycartografia/mioan/mioan2016/index.htm. ISTAC (2005) Canary institute of statistics (ISTAC). Input-output framework of the canary Islands. http://www.gobiernodecanarias.org/istac/estadisticas/economiageneral/cuentasenomicas/contabilidadreginonal/C00019A.html. IVIE (2000) Institut Valencia d’Estadística. Input-Output Framework and Regional Accounting of the Valencian Community. https://pegv.gva.es/es/temas/economia/cuentasenomicas/marcoinputoutputycontabilidadregionaldelacomunitatvalenciana. IVIE (2019) La Contribución Socioeconómica de las Universidades Públicas Valencianas. http://dx.medra.org/10.12842/SUPV_2019. Jayasinghe M Selvanathan S Selvanathan EA (2021) The Asian tsunami and tourism industry: impact and recovery. In: Economic Effects of Natural Disasters. Australia: Academic Press, pp. 335–358. ISBN 9780128174654 . DOI: 10.1016/B978-0-12-817465-4.00021-2 10.1016/B978-0-12-817465-4.00021-2 Miller RE Blair PD (2009) Input–Output Analysis: Foundations and Extensions. 2nd edition. New York: Cambridge University Press. Minondo A (2021) Impact of COVID-19 on the trade of goods and services in Spain. Applied Economic Analysis 29 (85 ): 58–76. DOI: 10.1108/AEA-11-2020-0156 10.1108/AEA-11-2020-0156 Murillo J Wow E Romani J , et al. (2013) How important to a city are tourists and daytrippers? The economic impact of tourism on the city of Barcelona. Tourism Economics 19 (4 ): 897–917. DOI: 10.5367/te.2013.0225 10.5367/te.2013.0225 Mursalina M Masbar R Suriani S (2022) Impact of covid-19 pandemic on economic growth of the tourism sector in Indonesia. International Journal of Quantitative Research and Modeling 3 (1 ): 18–28. http://journal.rescollacomm.com/index.php/ijqrm/article/view/261. Pulido A Fontela E (1993) Input-Output Analysis: Models, Data and Applications. Madrid: Editorial Pyramid. Rodousakis N Soklis G (2022) The COVID-19 multiplier effects of tourism on the German and Spanish economies. Evolutionary and Institutional Economics Review 19 : 497–510. DOI: 10.1007/s40844-021-00222-4 10.1007/s40844-021-00222-4 Sen S Kovacı S (2021) The impact of the COVID-19 pandemic on the tourism economy: evidence from borsa istanbul. In: Handbook of Research on the Impacts and Implications of COVID-19 on the Tourism Industry. pp. 155–176. DOI: 10.4018/978-1-7998-8231-2.ch008 Škare M Ribeiro-Soriano D Porada-Rochoń M (2021) Impact of COVID-19 on the travel and tourism industry. Technological Forecasting and Social Change 163 : 120469. DOI: 10.1016/j.techfore.2020.120469 10.1016/j.techfore.2020.120469 35721368 Utkarsh Sigala M (2021) A bibliometric review of research on COVID-19 and tourism: reflections for moving forward. Tourism Management Perspectives 40 : 100912. DOI: 10.1016/j.tmp.2021.100912 10.1016/j.tmp.2021.100912 34804787 Vaya E Garcia JR Murillo J , et al. (2018) Economic impact of cruise activity: the case of Barcelona. Journal of Travel and Tourism Marketing 35 (4 ): 479–492. Veryadi Purba JH Fathiah R Steven (2021) The impact of COVID-19 on the tourism sector in Indonesia. Rise: junal aplikasi Ekonomi 3 (1 ): 389–401. Xie Y Ji L Zhang B , et al. (2018) Evolution of the scientific literature on input–output analysis: a bibliometric analysis of 1990–2017. Sustainability 10 (9 ): 3135. DOI: 10.3390/su10093135 10.3390/su10093135
PMC010xxxxxx/PMC10291225.txt	==== Front MIE spmie Management in Education 0892-0206 1741-9883 SAGE Publications Sage UK: London, England 10.1177/08920206231177375 10.1177_08920206231177375 Original Manuscript Leading schools during a pandemic and beyond: Insights from principals in the Philippines https://orcid.org/0000-0001-7777-5187 Adams Donnie 69870 Faculty of Education, University of Malaya , Kuala Lumpur, Malaysia Namoco Sarah O College of Science and Technology Education, 125834 University of Science and Technology of Southern Philippines , Cagayan de Oro City, Philippines Ng Ashley Yoon Mooi University of Nottingham, Ningbo, China https://orcid.org/0000-0001-9881-582X Cheah Kenny S.L 69870 Faculty of Education, University of Malaya , Kuala Lumpur, Malaysia Donnie Adams, Faculty of Education, University of Malaya, Kuala Lumpur 50603, Malaysia. Email: drdonnieadams@gmail.com 22 6 2023 22 6 2023 08920206231177375© 2023 British Educational Leadership, Management & Administration Society (BELMAS) 2023 British Educational Leadership, Management and Administration Society (BELMAS) This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. The Philippines is one of very few countries in the world where schools have continuously remained closed since the coronavirus pandemic began in 2019. There is little known about how school principals face the challenges that arise from the pandemic, and the future goals for the new normal. The purpose of this article is to explore the school principals’ management practices, leadership styles, challenges encountered, and future goals in response to the pandemic in the context of Philippine schools. This study employed a qualitative research approach using an open-ended online survey with 52 school principals. Findings rendered a contextualisation of their school management practices, leadership styles, challenges encountered during the pandemic, and future goals for the new normal. This study contributes to the knowledge base on school leadership during the pandemic by providing unique insights into the Philippines. COVID-19 pandemic leadership styles management practices leadership in turbulent times edited-statecorrected-proof typesetterts19 ==== Body pmcIntroduction The COVID-19 pandemic has devastated countries all over the world in terms of their economies and educational systems, which caused large-scale school closures because of the rising infectivity rate, and the lack of inoculations among school communities (Viner et al., 2020). A UNESCO report estimated that 1.5 billion students from over 165 countries were not able to attend school due to lockdowns, and efforts to curb the COVID-19 virus (UNESCO, 2020). Many problems resulted from the lockdowns, such as limited access to digital learning tools and the internet, and this led to students not learning effectively (Mathrani et al., 2021). Students also suffered from the psychosocial impact of COVID-19, such as acute panic, anxiety, obsessive behaviours, hoarding, paranoia and depression (Dubey et al., 2020). As medical scientists continue their mission to find a cure for the COVID-19 virus, school leadership research embarks on a different challenge – to try to understand how school principals engage with students and teachers while still maintaining effective and efficient operation in schools, and to ensure that learning continues (Adams et al., 2021). School principals find themselves leading in a crisis, and in an unprecedented time. Teachers are forced to migrate their teaching to online platforms, which has impacted students in various ways, such as access, equity and quality of education (Sahlberg, 2020). Both principals and teachers are dealing with frustration, chaos and uncertainty (Adams and Muthiah, 2020). On top of that, the lack of information on the situation compels them to act swiftly but guided by foresight, while contemplating the options and consequences of their decisions (Harris, 2020). While many countries have also implemented large-scale or countrywide school closures since March 2020 as an effort to decrease COVID-19 transmission, this pandemic has also brought with it the idea of a new normal, where school communities are challenged to adapt to uncertainty, and live day to day with unfolding events and circumstances. Whether at the national, regional or local school level, school closure is the practical response to high student infection rates in many countries (Viner et al., 2020). While other countries have taken the opportunity to resume in-person classes, the Philippines has lagged behind. It is one of the very few countries in the world where schools have continuously remained closed since the coronavirus pandemic began in 2019 (United Nations Children's Fund, 2022). This circumstance has forced principals to rely on virtual meetings to communicate with their stakeholders, and make decisions collaboratively while establishing various connections to stay informed about standard operating procedures (Gkoros and Bratitsis, 2022; Stone-Johnson and Miles Weiner, 2020). On the other hand, school principals are not trained to handle such crises, and they find themselves leading their schools into uncharted territory, having to make decisions, and deal with various levels of anxiety, frustration, and anger from the school community (Adams et al., 2021). There is little known about how school principals face challenges that arise from the pandemic, and the future goals for the new normal. Thus, the objective of this research was to explore the school management practices, leadership styles, challenges encountered and the future goals of principals in response to the pandemic in the Philippines school context. Leading schools in times of crisis Many crises strike without notice (Coomb, 2014). Some can be addressed quickly, while others can take a long time to settle (Smith and Riley, 2012). The tension lies in the lack of information and solutions to problems, or in trying to determine which tactics to employ in times of crisis (Fotheringham et al., 2021). During the pandemic, principals are confronted with tough decisions to make to provide assurance, hope and transparency to anxious parents who are very concerned about their children's education (Weiner et al., 2021). Principals must instil trust in their school communities so that they can ride through these turbulent times (Reid, 2021). Smith and Riley (2012) contended that a crisis is a major test of school leadership that requires the principal to act quickly and decisively. School principals need to constantly deal with unfolding events, the fluctuating emotions of school communities, and to minimise unwanted consequential effects (Coomb, 2014). Such dilemmas cannot be avoided, and leaders need to be resilient in order to understand and shoulder responsibilities, and make unconventional decisions to address such an unprecedented situation. They must be quick to ascertain the problem, and how it is connected to the bigger picture (Stone-Johnson and Miles Weiner, 2020). In Nigeria, the complex and sudden change resulting from the COVID-19 crisis has led to confusion, chaos and helplessness (Gyang, 2020). The pandemic also exacerbated the already existing digital divide in countries like India, Pakistan, Bangladesh, Nepal and Afghanistan (Mathrani et al., 2021). On the contrary, in developed nations like Sweden, schools remained open (as compared to neighbouring countries like Norway and Denmark), but their school principals faced tensions and anxiety from parents, staff members and students (Ahlström et al., 2020). School principals need to be more intuitive, resilient, creative, agile and authentic in their sets of leadership attributes (Weiner et al., 2021). As new scenarios unfold, they must be flexible, foresightful and courageous to take risks (Reid, 2021). For countries like Singapore, their preparation in home-based learning was their cornerstone to ride through the storm of the COVID-19 pandemic. Countries around the world engaged strong governance in combating the virus, and ensuring cases remain low throughout the outbreak (Abdullah and Kim, 2020). However, they too, are confronted with challenges during the pandemic in the aspect of digital transformation. Studies have reckoned the need for principals to think of a more sustainable approach especially in the aspect of managing the barriers associated with technology curriculum integration because of: (a) lack of resources; (b) varied perceptions on how their students are engaged through their teachers and (c) inadequate alignment between central agencies and local partnerships (Hung et al., 2020). There are studies that have highlighted that the principals are committed to fostering new relations and integration with school communities to withstand the effects of prolonged school closures (Stone-Johnson and Miles Weiner, 2020; Weiner et al., 2021). Their initiatives include expanding new networks, knowledge and resources, involving more collaborative decision making with stakeholders, generating new unconventional ideas, and taking risks that would lead towards a community-based model of leadership (Adams, Ng and Muniandy, 2020; Reyes-Guerra et al., 2021). The benefits are considered numerous, and among them: (a) the ability to influence active participation in the provision of education; (b) creating opportunities for students of low socioeconomic background to have access to education and (c) proper mobilisation of resources, and solutions from collaborative strategic and critical thinking (Gyang, 2020). In the Philippines, Guiamalon et al. (2022) reported four main types of coping mechanisms used by the public secondary school principals in the process of adjusting to the new norm: flexibility and creativity, cooperation among teachers, following instructions from the Department of Education (DepEd), and having community linkages. A study also found that school leaders have a very high degree of resilience due to the communication and emphasis on school community health matters. As such, there was constant feedback, support and monitoring to ensure the high quality of education is achieved (San Miguel and Pascual, 2021). Francisco and Nuqui (2020) found that school administrators in the Philippines were committed instructional leaders, maintaining their capacity to change, while taking on more responsibilities in managing curriculums, budgeting, and scheduling classes. While schools are reopening and operating under new norms, some principals have to muddle through the scene, stay vigilant, and be quick to respond to unwarranted situations (Weiner et al., 2021). Leadership has also shifted from a centralised manner to distributed forms of leadership because the old norm could no longer fit its present purpose (Harris and Jones, 2020; Thien and Adams, 2021). As such, principals require support and collaboration from all staff to stay engaged in crisis management. School leaders need to be realistic, and resign to the fact that the old normal may not return, and they must live in times of adaptations, and uncharted systemic reforms (Harris, 2020). They may have to evolve alongside the virus, and learn to live with the pandemic as long as it remains a clear and present danger. Methodology This study employed a qualitative research approach using an open-ended online survey. Qualitative surveys can produce rich and complex accounts of participants’ subjective experiences, positionings and discourses (Braun et al., 2021), with the primary goal being not to generalise, but rather to provide a rich, contextualised understanding of an aspect of human experience (Hennink and Kaiser, 2021; Polit and Beck, 2010). The limitations of open-ended questions in providing any causal explanations are fully acknowledged (Krosnick and Presser, 2010). Therefore, the findings of this study are indicative rather than definitive. In this study, the survey was self-administered with questions presented in a fixed and standard order to all participants. In ensuring that this research followed ethical standards, letters were sent to the DepEd. Once approval was obtained, letters were also sent to prospective principals to inform them of the objectives and procedures of the research. Subsequently, after informed consent was received, the principals were administered the online survey consisting of open-ended questions. On the cover of the online survey, the principals were given the choice to either to take part in the research, or otherwise, whereby participation was strictly voluntary, and anonymous. Wu et al. (2022) argued that sending surveys to a clearly defined and refined population positively impacts the online survey response rate. In the case of this study, the participants were school principals who have had administrative experience both before and during the COVID-19 pandemic. Fifty-two principals participated in the study, and the online survey response rate was 100%. To protect the participants’ identity, codes such as ME1, FE1, MH1 and FH1 were used (M = male, F = female, E = elementary, H = high school, 1 = participants’ number). Questionnaire items were in the English language as it is the official language of the Philippines, spoken proficiently by a large number of Filipinos even in different sectors, and is the medium of instruction in education (Cabigon, 2015). The profile of the 52 principals from Mindanao, Philippines who participated in the survey is shown in Table 1. Table 1. Socio-demographic profile of the participants (n = 52). Characteristic Frequency Percentage Age 30 years old and below 1 2 31–40 years old 8 15 41–50 years old 24 46 51–60 years old 16 31 61 years old and above 3 6 Sex Female 30 58 Male 22 42 Highest academic qualification Bachelor's degree 4 8 Master's degree 28 54 Doctoral degree 20 39 Years of experience as school administrator Below 5 years 15 29 6–10 years 22 42 11–15 years 9 17 16–20 years 5 10 More than 21 years 1 2 Type of school Elementary 26 50 High school 26 50 Location of school Rural 26 50 Urban 26 50 The data analysis utilised the conventional content analysis approach to determine emerging themes. This approach to data analysis is ‘generally used with a study design whose aim is to describe a phenomenon’, and is most appropriate to employ ‘when existing theory or research literature on a phenomenon is limited’. Hence, in this approach, coding categories were derived directly from the text data, ‘allowing the categories and names for categories to flow from the data’ (Hsieh and Shannon, 2005: 1279). This analysis is in consonance with Ryan and Bernard (2003), who stressed that themes can be identified from the data. Using ATLAS.ti 22 to analyse the data, 41 codes were identified. The basis for determining these emerging themes was the manner by which the research participants put emphasis on the issues, namely, by frequency counts, that aligned with the research objective of this study. This refers to the ‘topics that occur and reoccur’ (Bogdan and Taylor, 1975: 83), and ‘recurring regularities’ (Guba, 1978: 53). An excerpt of the data coding, and the corresponding emerging themes is presented in Table 2. Table 2. Excerpt of data codes, sub-themes and emerging themes. Code/statement Sub-themes Theme The role of a school leader during challenging time is a guide. Mentors School Management Practices Support, encouragement and cheer to the teachers in their achievements. Always look into the safety of all teaching personnel and provide necessary measure for their protection during this challenging time. Protectors A COVID shield in which the priority is to protect the teachers and learner. Must be the trailblazer of new opportunities for the teachers to explore… Trailblazers Leaders should advance their skills in the use of educational technology… Always act as a team. Distributed Leadership Leadership Styles Delegating tasks to teachers. The Word of God has always been his guide towards achieving the vision and mission of the institution. Transformational Leadership Motivate teachers to enrol in graduates’ studies. The insufficient number of learning materials to meet the need of 1:1 ratio. Implementation of Self-Learning Modules Challenges Encountered Students submit incomplete or unanswered self-learning modules. Parents that are usually arguing, asking questions and clamouring for face-to-face learning Poor Parental Support Parents are forced to tutor their children despite the difficulty of teaching because they themselves do not know what and how to teach. Teachers must be physically and mentally fit before letting them to start teaching students. Safety and Well-Being Future Goals This can be done by debriefing the students and teachers from the trauma of the COVID-19 grief. The poor internet connection, especially to rural areas is a big problem. Educational Technology Training Teachers must be trained to be technologically equipped so that they will be adoptive to change. Findings Results from the qualitative data analysis led to several themes that kept emerging throughout all the responses regardless of the probing sequence. This section of the article presents the qualitative findings according to the following themes: school management practices, leadership styles, challenges encountered and future goals. School management practices School management practices refer to the various strategies, policies, procedures and actions that school principals use to manage and operate schools effectively. These practices are designed to ensure that the school functions smoothly, efficiently and in accordance with its mission and goals (Gorton and Alston, 2018). School management practices in the Philippines during the pandemic took the form of operating schools remotely due to the need for social distancing (Adams et al., 2021). In such a circumstance, school principals acted as mentors, protectors and trailblazers. Mentors School principals as mentors refers to the practice of providing guidance, support and professional development opportunities to teachers and staff (Bressman et al., 2018). As mentors, school principals may offer advice, share expertise, provide feedback on performance and serve as role models for others (Searby et al., 2017). The data showed that as a mentor, the school principals ‘guided teachers and helped them perform their tasks during the pandemic’ (FH4). This means that the school principal needs to be ‘highly competent and knows how to rally their stakeholders’ (MH3) in order to ‘enlighten their teachers about the situation and encourage them to carry out their tasks in order to achieve the school goals’ (ME6). In this way, the school principal provided ‘support, encouragement and injected optimism to the teachers’ (FH15). This is a crucial time for the school principal to lead with compassion as FE5 pointed out, that they ‘lead with a heart, listen with a heart, and decide with a heart’. The findings showed school principals who serve as mentors played an important role in guiding teacher instructional practices and improving teacher performance. By taking a compassionate and understanding approach to their work, the school principals, in turn, made teachers feel valued and supported in their work. The key findings are presented in Figure 1. Figure 1. Key findings under principals as mentors. Protectors The concept of school principals as protectors refers to the role that principals play in creating and maintaining safe and secure learning environments for teachers, students and staff (Adams et al., 2021). As protectors, school principals are responsible for implementing policies and procedures that promote physical and emotional safety, such as developing emergency response plans, and addressing issues of bullying and harassment (Naranasamy and Adams, 2020). The data from the interviews show as a protector, the safety of all in the school was the priority of the principals as ME8 described the situation as ‘always looking into the safety of all teaching personnel and providing necessary measures for their protection during this challenging time’. As such, the school principals was ‘vigilant of their teachers’ welfare’ (ME4). MH7 and MH1 also described the principal's protector role with words such as ‘safety officer’ and ‘covid shield’. By being a protector, the school principal also gave hope to the students and teachers by ‘providing clarity and direction, and instilling hope while remaining focused on the best possible outcomes for the students and the whole school community’ (FH7). The findings highlights school principals in the Philippines played a critical role in promoting both their teachers and students well-being. The school principals acted as protectors of their teachers’ and students’ physical and emotional safety and created a safe learning environments in their schools. The key findings are presented in Figure 2. Figure 2. Key findings under principals as protectors. Trailblazers School principals as trailblazers refers to their role in leading innovation and change in learning and teaching. As trailblazers, school principals are responsible for identifying and implementing new and effective teaching strategies, technologies and policies that can improve student learning outcomes (Yuting et al., 2022). The findings show school principal acted as trailblazers by implementing new ways of learning and teaching. FE1 shared that there were ‘new opportunities for the teachers to explore educational technology so that teachers can cope and impart knowledge and skills to their students during this pandemic’. FE13 added that the school principal ‘advance skills in the use of educational technology to develop innovations and adjustment in the school’. FE8 said that the school principal ‘paved the way for the delivery of quality education in all possible ways that is safe for the teachers and students’. Despite the uncertainties during the crisis, school principals in the Philippines were innovative and able to think creatively in response to the pandemic. They provided their teachers with the necessary support and resources to implement innovative teaching practices and played a critical role in identifying new and effective teaching strategies and technologies. The key findings are presented in Figure 3. Figure 3. Key findings under principals as trailblazers. Leadership styles Leadership styles refer to the characteristic ways in which school principals approach their roles and responsibilities (Harris and Jones, 2020). There are several different leadership styles that principals may adopt, however, this is dependent on the school context and the needs of students, teachers and the other school stakeholders (Noman et al., 2018). Distributed leadership Distributed leadership refers to the extent to which leadership functions are distributed among formal and informal leadership positions in the leadership team, including the school principal, assistant principals and teacher leaders, rather than relying on the school principal a single leader to make all decisions in the school (Thien and Adams, 2021). The data showed that the leadership styles practised by the school principals was akin to distributed leadership. FE7 reported the collaborations between the school principal, and the teachers and members of the community. On the other hand, FH11 mentioned that despite numerous challenges faced at such a time, the school principal were ready to act together with the teachers to solve problems. This was enabled by the principals’ leadership style of ‘empowering teachers’ (FH11), ‘delegating tasks to teachers’ (ME5) and ‘connecting to external partners for support’ (MH5). These insights reveal school principals in the Philippines adopted a distributed leadership style during the crisis and engaged multiple leaders with different expertise, skills and perspectives that contributed to effective decision-making and shared leadership responsibilities. In doing so, the principals created a more collaborative and inclusive school environment. The key findings are presented in Figure 4. Figure 4. Key findings under principals distributed leadership. Transformational leadership Transformational leadership is a leadership approach that involves inspiring and motivating teachers to achieve a common goal, often by creating a shared vision (Bass and Avolio, 1994). This involves school principals working collaboratively with teachers and other school stakeholders to create a positive school culture and improve student outcomes (Bellibaş et al., 2021). Most participants shared that their school principals exhibited traits of transformational leadership. Participants of the study mentioned their school principals has ‘good rapport with the teacher and stakeholders’ (FE9), ‘open communication’ (FH10, ME6, FE12, ME7), and practiced ‘active listening’ skills (ME6, FE12, ME7). These findings indicated school principals in the Philippines had good rapport with their teachers, open communication and active listening. The principals also modelled positive behaviours and encouraged two-way communication with their teachers, showcasing characteristics of transformational leaders. The key findings are presented in Figure 5. Figure 5. Key findings under principals transformational leadership. Challenges encountered School principals face a variety of challenges in their roles as educational leaders (Adams and Muthiah, 2020). The primary responsibility of school principals during the pandemic is to navigate the uncertainties caused by the COVID-19 crisis. Very often they need to navigate through unprecedented situations, calling upon their ingenuinity and innovativeness as well as critical interpretation of the policies handed down by the local and national authorities. Implementation of Self-Learning Modules Like all schools in other parts of the world, the schools in the Philippines needed to be closed to stop the spread of the pandemic. To ensure learning still happened when the schools were closed, the DepEd introduced the Modular Distance Learning programme. In this programme, students could access Self-Learning Modules (SLMs) which were made available to them either online, or at different strategic locations in villages (Fideliz, 2021). With this programme, school principals faced several challenges. As the principals, teachers and students need to work, teach and learn from their own homes, the first challenge that all parties had to deal with was the poor internet connection, specifically in the rural areas (DepEd, 2020). It was either non-existent, or extremely slow, and in some cases, students had no access to learning as they had no computers, and smartphones were hardly suitable to be used for teaching and learning. This caused the delayed retrieval of SLMs. FH1, FE10 and ME7 remarked that the submission of assignments by students was often delayed. FH5 summed it up: ‘This makes the students and teachers unable to finish their tasks on time’. FH7, FE14, FE16, ME8 and MH3 shared that the distribution of copies of SLMs to certain locations was hampered by limited resources, whereby the DepEd was unable to provide sufficient copies for each student. All these led to a shortage of learning materials. In addition, teachers were neither ready nor skilled in teaching with the use of digital devices as shared by FE1 and FH11, that ‘some teachers are not computer or internet savvy’. Another challenge was the students’ need for continuous guidance from teachers. Left on their own for most of the time with the SLMs, students often submitted incomplete, or unanswered assignments. FE10 mentioned that this could be attributed to the fact that ‘students are having difficulty in comprehending lessons on their own’. Not only was learning made challenging by the absence of ‘detailed instruction from the teachers’, but also, as FH7 said, ‘due to the actual performances not being observed’. MH12 added that, for science-based subjects, online learning did not provide ‘the observation of a phenomenon in the laboratory’. Such inadequacies of online learning made learning less attractive, and consequently, students lost interest in learning. The key findings are presented in Figure 6. Figure 6. Key findings under implementation of self-learning modules. Poor parental support Parental support to schools can take many forms and is crucial to the success of students and the school as a whole. By working together with school principals and teachers, parents can help create a supportive learning environment that promotes positive student outcome (Weiner et al., 2021). Perhaps the most formidable challenge faced by school principals was the parents’ attitude towards their children's learning. According to the Philippine Statistics Authority, the poverty incidence among the population increased to 23.7% during the first half of 2021 from 21.2% in the same period in 2018. This estimates to an increase of 3.9 million Filipinos living in poverty. COVID-19 has taken a heavy toll on the people, specifically on the livelihood of rural communities. The loss of jobs, and thus incomes, and the stress of coping with the COVID-19 restrictions, with its severe economic impact, would inadvertently cause their children's learning to take a back seat. Furthermore, parents equate effective learning with face-to-face teaching as ME8 shared that ‘parents are usually arguing, asking questions and clamouring for face-to-face learning’. When parents felt that online learning did not bring any benefit, taking their children out of school to work and help put food on the table became inevitable. This means that school principals did not only face the challenge of high absenteeism, but also whether these children would be returning to the schools in the future. Besides, parents were not in a position to tutor their children at home, being uneducated themselves, as informed by FH3, that the parents ‘do not know what and how to teach’. Under such a circumstance, parents had to make a decision on whether to allow their children to continue with their schooling, or to drop out. In the Philippines, like other developing and low-income countries, the children's perceived educational ability is a key determinant of whether the parents choose to keep them in school, or put them to work instead (World Bank, 2018). Due to the aforementioned COVID-19-induced ineffective schooling environment, the parents’ belief in the power of education to transform lives dwindled. The key findings are presented in Figure 7. Figure 7. Key findings under poor parental support. Future goals Future goals refers to school principals creating a vision to ensure that their schools continue to provide the best possible learning environment for their students and setting strategies to achieve their goals (Bass and Avolio, 1994; Bellibaş et al., 2021). Safety and well-being School principals need to prioritise safety and well-being of their teachers and students to ensure that their schools provide a safe and healthy learning environment (San Miguel and Pascual, 2021). This includes implementing policies and practices that promote safety and health, providing training to teachers and staff, and collaborating with parents (Naranasamy and Adams, 2020). The data showed that the immediate need is to make the schools safe as FH7 stressed the need to ‘set up a safe school before the opening of face-to-face classes’, and that ‘teachers be physically and mentally fit before letting them to start teaching the learners’. This need goes beyond the normal sanitising standard operating procedure (SOP), and educational achievement. School principals need to look beyond academic instructions, and into the socio-emotional and mental health of students, which is showing early indications of concern. Providing online and/or hybrid teaching and learning methods takes away opportunities to learn through interactions with peers as well as building relationships with adults, and to celebrate the success and milestones achieved. In planning for the future, FE16 and ME8 emphasised the need ‘to ensure protocols are in place for safe face-to-face learning’. This needs flexibility to change and implement appropriate policies and procedures as new data and information emerge. For learning to take place, school principals need to ensure that students and teachers turn up, and re-engage with learning. The key findings are presented in Figure 8. Figure 8. Key findings under safety and well-being. Educational Technology Training Educational Technology Training refers to the process of equipping teachers and students with the skills and knowledge needed to effectively use technology in the classroom (Hung et al., 2020). This includes the effective use of computers, tablets, as well as software and applications such as educational games, learning management systems and online resources. Much discourse from the data revolves around the usage of educational technology. ME6 felt that ‘principals should look into innovative and creative ways to address the various problems faced in the country, such as providing devices that are solar-powered’. ME8 added that ‘teachers need to be trained to use the computers and teach by using the applications available’. School principals, therefore, need to look into the continuous professional development of teachers so that they are ready to carry out remote teaching for their students to continue learning if and when the next crisis strikes (Adams et al., 2021). The key findings are presented in Figure 9. Figure 9. Key findings under educational technology training. The overall themes and sub-themes are reflected in the model displayed in Figure 10. Figure 10. Model of the overall themes and sub-themes of the qualitative findings. Discussion This qualitative study aimed at exploring the school management practices, leadership styles, challenges encountered and the future goals of principals in response to the pandemic in the Philippines school context. The current study fulfils the call for studies about how school principals face these challenges that arise from the pandemic, and how they address them in local contexts (Harris, 2020; Harris and Jones, 2020). This final section links the main findings to insights from the literature. School principals’ management practices in response to the pandemic The findings show school principals played a critical role in guiding and supporting teachers, ensuring the safety of the school community, and implementing new ways of learning and teaching. Their management practices was the catalyst in ensuring that teachers are able to deliver instructions effectively. Additionally, the principals provided necessary measures for protection, were vigilant of their teachers’ welfare, and gave hope to teachers by providing clarity and direction on online instructions. The school principals acted as trailblazers by implementing new ways of learning and teaching, and made adjustments to ensure that quality education could be delivered in a safe manner. Similarly, Pollock (2020) discuss the importance of school principals prioritising the well-being of the teachers and learners, and at the same time, making sure learning is still taking place. The finding that school principals led with compassion is particularly noteworthy. As Netolicky (2020) highlighted, principals must communicate the organisation's direction while considering the well-being of its constituents with empathy and humanity. School principals in this study listened to the needs of their teachers and made decisions based on a consideration of their well-being, in addition to achieving the student learning goals. Principals maintained an instructional role, but were emotionally literate above and beyond their duties to implement DepEd mandates, such as safety measures (Francisco and Nuqui, 2020). Van der Vyver et al. (2014) stressed that the managerial role of school principals has shifted towards a more caring and supportive role. Gray (2009) argued that intelligent leaders are those who are able to discern the emotions of the people around them. The pandemic caused a paradigm shift among school principals in the Philippines from being a traditional authoritative figure in schools towards a more empathic role. School principals’ leadership styles during the crisis The results indicate that school principals adopted a distributed leadership style during the crisis worked together with teachers to solve problems, which empowered teachers to take on responsibilities and tasks that were traditionally held by the principal. Additionally, the school principal sought external support when necessary to address issues that were beyond their expertise. Cranston and Kusanovich (2015) points out that school principals should not lead alone, but needs to involve others, and be ready to accept new ways of looking at leadership roles in schools, such as involving teachers in the decision-making process. Leadership in times of crisis has shifted to distributed forms of leadership because the old norm could no longer fit its present purpose (Harris and Jones, 2020; Thien and Adams, 2021). Furthermore, school principals exhibited traits of transformational leadership by inspiring and motivating teachers to work towards a shared goal. The principals had good rapport with teachers and stakeholders, were open to communication, and practiced active listening skills. This finding echoed other similar studies, particularly Menon (2023) study that found school leaders utilised practices and behaviours associated with transformational leadership in order to manage the crisis effectively at the school. Challenges encountered during the pandemic Principals faced various challenges during the pandemic. Among the key challenges were poor internet connection, lack of teacher preparedness, skill in using digital devices and, absence of detailed instructions from teachers that resulted in students being left to navigate the online learning on their own. Moreover, the most significant challenge faced by school principals during the pandemic was the parents’ attitude towards their children's learning. Parents equated effective learning with face-to-face teaching and were not fully convinced that online learning was a viable alternative. These findings resonated with the literature in which school principals faced challenges during the pandemic in terms of parents negative perception on online learning, poor internet connectivity, and teachers lack of skills in using technology for teaching (Adams et al., 2021; Ahlström et al., 2020; Hung et al., 2020). Future goals for the new normal The pandemic has put the country's learning behind by many years, causing tremendous loss in terms of the economy, and global growth (World Bank, 2020). It has also taken a huge toll on students’ academic progress. Hence, school principals need to respond to multiple horizons to facilitate students’ recovery and beyond. Under such a bleak forecast, they need to prepare for the new normal, and get their schools pandemic-ready for the future. The qualitative results suggest that the safety of students and teachers is a top priority for school principals, particularly during the endemic phase. School principals were found to be taking proactive measures to ensure that schools were safe for face-to-face learning. This included setting up safety protocols and ensuring that teachers were physically and mentally fit to resume teaching in-person (Pollock, 2020). Teachers’ inability to conduct lessons via technology has received much criticism despite other factors possibly being the reason, such as poor internet connectivity, and the unpreparedness for the sudden onslaught of the pandemic. Nevertheless, much can be done on this front. Teachers need continuous professional development so that they are ready to carry out remote teaching for their students to continue learning if and when the next crisis strikes (Adams et al., 2021). Additionally, school principals have to generate hope for the children returning to school, and obtaining an education in the new normal (Cahapay, 2021; Guiamalon et al., 2022). Conclusion The findings of this study contribute to the scarce knowledge base on school leadership during a pandemic by providing unique insights into the Philippines. The Philippines is one of the very few countries in the world where schools have continuously remained closed since the pandemic (United Nations Children's Fund, 2022). Being the first study of this nature conducted in the Philippines, it has contextual originality. The study highlights evidence based on school management and leadership practices during a pandemic that both describes and analyses how school principals face the challenges that arise from the pandemic, and the future goals for the new normal. One thing that the COVID-19 pandemic has clearly taught school principals in the Philippines is that they do not and should not be alone in dealing with crises. Principals had to shift the traditional hierarchical top-down relationship to a horizontal school–community relationship. With the assistance and involvement of their teachers, communities and school authorities, they can work in unison to address the challenges posed by the crisis and get their schools prepared for the new normal and, pandemic-ready for the future. Additionally, school principals are now more responsive to the teachers’ socio-emotional needs, and have repositioned their schools to reflect ‘caring institutions’ (Noddings, 2015). It is important to build upon this study in other regions of the world to determine the context-specificity of empirical findings. It is time for this knowledge base to receive greater attention and focus from practitioners and local policymakers, especially in leadership preparation and development programmes for school principals ship. Author biographies Sarah O Namoco is currently the chairperson of the Department of Technical and Technology Education in the University of Science and Technology of Southern Philippines, Cagayan de Oro City. Prior to joining the university, she worked as a secondary school teacher in the Department of Education. She finished her Doctor of Education from the Universiti Sains Malaysia. Donnie Adams obtained his PhD in Educational Leadership from Universiti Malaya under the Bright Sparks scholarship and was awarded the Universiti Malaya's Excellence Award 2016: PhD Completion in Less than 3 years. He was recently awarded the Emerald Young Researcher Award 2021 and Universiti Malaya's Excellence Award in 2019. His research and development work interests are in inclusive school leadership and school-wide reformation of inclusive education agenda in Malaysia. Ashley Yoon Mooi Ng is an associate professor in Education, and the interim head of School for The School of Education and English in the University of Nottingham Ningbo, China. She sits as an advisor in the International Council of Global School Leadership, and a Senior Research Fellow at the University of Nottingham, Malaysia. Her research interests are in innovative talent development for the future, teacher professional development, school leadership development, mentoring, and gender and leadership. Kenny SL Cheah is currently the chair of CRELAMP (Center for Research in Educational Leadership, Administration, Management and Policy) based in the Faculty of Education, University Malaya. He teaches, writes, and publishes in the subjects of Educational Leadership & Management, He is also a senior certified trainer with ATLAS.ti. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iDs: Donnie Adams https://orcid.org/0000-0001-7777-5187 Kenny S.L Cheah https://orcid.org/0000-0001-9881-582X ==== Refs References Abdullah WJ Kim S (2020) Singapore’s responses to the COVID-19 outbreak: a critical assessment. American Review of Public Administration 50 (6–7 ): 770–776. Adams D Cheah KSL Thien LM , et al. (2021) Leading schools through the COVID-19 crisis in a South-East Asian country. Management in Education 08920206211037738. DOI: 10.1177/08920206211037738 Adams D Muthiah V (2020) School principals and 21st century leadership challenges: a systematic review. Journal of Nusantara Studies (JONUS) 5 (1 ): 189–210. Adams D Ng AMY Muniandy V (2020) Principal leadership preparation towards high-performing school leadership in Malaysia. Asian Education and Development Studies 9 (4 ): 425–439. Ahlström B Leo U Norqvist L , et al. (2020) School leadership as (un) usual. Insights from principals in Sweden during a pandemic. International Studies in Educational Administration (Commonwealth Council for Educational Administration & Management (CCEAM) 48 (2 ): 35–41. Bass BM Avolio BJ (1994) Improving Organizational Effectiveness through Transformational Leadership. Thousand Oaks, CA: Sage. Bellibaş MŞ Kılınç AÇ Polatcan M (2021) The moderation role of transformational leadership in the effect of instructional leadership on teacher professional learning and instructional practice: an integrated leadership perspective. Educational Administration Quarterly 57 (5 ): 776–814. Bogdan R Taylor SJ (1975) Introduction to Qualitative Research Methods: A Phenomenological Approach to the Social Sciences. New York: Wiley. Braun V Clarke V Boulton E , et al. (2021) The online survey as a qualitative research tool. International Journal of Social Research Methodology 24 (6 ): 641–654. Bressman S Winter JS Efron SE (2018) Next generation mentoring: supporting teachers beyond induction. Teaching and Teacher Education 73 : 162–170. Cabigon M (2015) State of English in the Philippines: Should we be concerned? Retrieved from British Council Philippines: https://www.britishcouncil.ph/teach/state-english-philippines-should-we-be-concerned-2. Cahapay MB (2021) The phenomenon of leading without guidebook: educational leadership practices of philippine school principals in virulent COVID-19 times. International Journal of Educational Leadership and Management: 1–3. DOI: 10.2139/ssrn.3825325. Coomb WT (2014) Ongoing Crisis Communication: Planning, Managing, and Responding. Thousand Oaks, California: Sage Publications. Cranston J Kusanovich K (2015) Learning to lead against the grain dramatizing the emotional toll of teacher leadership. Issues in Teacher Education 24 (2 ): 63–78. Department of Education (2020) Revised guidelines on alternative work arrangements in the Department of Education during the period of state of national emergecy due to COVID-19 pandemic, DepEd Order, no. 011, 2. 2020. Encyclopedia of Educational Leadership and Administration 1–16. Dubey S Biswas P Ghosh R , et al. (2020) Psychosocial impact of COVID-19. Diabetes & Metabolic Syndrome-Clinical Research & Reviews 14 (5 ): 779–788. Fideliz (2021) Modular printed learning modality: A responses to covid 19 threat on education. Retrieved from Press Reader: https://bit.ly/3o4CS5E Fotheringham P Harriott T Healy G , et al. (2021) Pressures and influences on school leaders navigating policy development during the COVID-19 pandemic. British Educational Research Journal 48: 201–227. Francisco CD Nuqui AV (2020) Emergence of a situational leadership during COVID-19 pandemic called new normal leadership. Online Submission 4 (10 ): 15–19. Gkoros D Bratitsis T (2022) E-leadership and distance education in Greece during COVID-19 pandemic. Management in Education: 1–13. DOI: 10.1177/08920206221081611. Gorton RA Alston JA (2018) School Leadership & Administration: Important Concepts, Case Studies, & Simulations (10th ed.). New York: McGraw-Hill. Gray D (2009) Emotional intelligence and school leadership. International Journal of Educational Leadership Preparation 4 (4 ): 1–3. Guba EG (1978) Toward a methodology of naturalistic inquiry in educational evaluation. Monograph 8. Los Angeles: UCLA Center for the Study of Evaluation. Guiamalon TS Lumapenet HT Katog MS , et al. (2022) Coping with COVID-19: how public secondary school principals adapt to the new normal? International Journal of Early Childhood 14(1): 2363–2367. Gyang TS (2020) Educational leadership response to the COVID-19 pandemic crisis in Nigeria. International Studies in Educational Administration (Commonwealth Council for Educational Administration & Management (CCEAM) 48 (3 ): 73–79. Harris A (2020) COVID-19–school leadership in crisis? Journal of Professional Capital and Community 5: 321–326. Harris A Jones M (2020) COVID 19 – school leadership in disruptive times. School Leadership & Management 40 (4 ): 243–247. Hennink M Kaiser BN (2021) Sample sizes for saturation in qualitative research: a systematic review of empirical tests. Social Science & Medicine: 114523.34785096 Hsieh HF Shannon SE (2005) Three approaches to qualitative content analysis. Qualitative Health Research 15 (9 ): 1277–1288.16204405 Hung D Huang JS Tan C (2020) Leadership in times of pandemics: reflections from Singapore. International Studies in Educational Administration 48 (2 ): 56–63. Krosnick JA Presser S (2010) Question and questionnaire design. In: James DW and Peter VM (Eds) Handbook of Survey Research, San Diego, CA: Elsevier, pp. 263–314. Mathrani A Sarvesh T Umer R (2021) Digital divide framework: online learning in developing countries during the COVID-19 lockdown. Globalisation Societies and Education 20: 625–640. Menon ME (2023) Transformational school leadership and the COVID-19 pandemic: perceptions of teachers in Cyprus. Educational Management Administration & Leadership: 1–18. DOI: 10.1177/17411432231166515. Naranasamy K Adams D (2020) Headmasters safety management and performance in two Malaysian National Primary Schools. International Online Journal of Educational Leadership 3 (1 ): 43–58. Netolicky DM (2020) School leadership during a pandemic: navigating tensions. Journal of Professional Capital and Community 5 (3–4 ): 391–395. Noddings N (2015) The Challenge to Care in Schools: An Alternative Approach to Education (2nd ed.). New York: Techer College Press. Noman M Awang Hashim R Shaik Abdullah S (2018) Contextual leadership practices: the case of a successful school principal in Malaysia. Educational Management Administration and Leadership 46 (3 ): 474–490. Polit DF Beck CT (2010) Generalization in quantitative and qualitative research: myths and strategies. International Journal of Nursing Studies 47 : 1451–1458.20598692 Pollock K (2020) School leaders’ work during the COVID-19 pandemic: a two-pronged approach. International Studies in Educational Administration 48 : 38–44. Reid DB (2021) Suppressing and sharing: how school principals manage stress and anxiety during COVID-19. School Leadership & Management 42: 1–17. Reyes-Guerra D Maslin-Ostrowski P Barakat MY , et al. (2021) Confronting a compound crisis: the school principal’s role during initial phase of the COVID-19 pandemic. Frontiers in Education 6 : 617875. Ryan GW Bernard HR (2003) Techniques to identify themes. Field Methods 15 (1 ): 85–109. Sahlberg P (2020) Will the pandemic change schools? Journal of Professional Capital and Community 5 (3/4 ): 359–365. San Miguel NV Pascual EA (2021) School leaders’ resilience amidst pandemic in the division of laguna, Philippines. International Journal of Research Publications 88 (1 ): 22–22. Searby L Browne-Ferrigno T Wang CH (2017) Assistant principals: their readiness as instructional leaders. Leadership and Policy in Schools 16 (3 ): 397–430. Smith L Riley D (2012) School leadership in times of crisis. School Leadership & Management 32 (1 ): 57–71. Stone-Johnson C Miles Weiner J (2020) Principal professionalism in the time of COVID-19. Journal of Professional Capital and Community 5 (3–4 ): 367–374. Thien LM Adams D (2021) Distributed leadership and teachers’ affective commitment to change in Malaysian primary schools: the contextual influence of gender and teaching experience. Educational Studies 47 (2 ): 179–199. UNESCO (2020) COVID-19 response. Available at: https://en.unesco.org/covid19. United Nations Children’s Fund (2022) Where are we on education recovery? https://www.unicef.org/media/117626/file/Where%20are%20we%20in%20Education%20Recovery?.pdf van der Vyver CP van der Westhuizen PC Meyer LW (2014) Caring school leadership: a South African study. Educational Management Administration and Leadership 42 (1 ): 61–74. Viner RM Russell SJ Croker H , et al. (2020) School closure and management practices during coronavirus outbreaks including COVID-19: a rapid systematic review. Lancet Child & Adolescent Health 4 (5 ): 397–404.32272089 Weiner J Francois C Stone-Johnson C , et al. (2021) Keep safe, keep learning: principals’ role in creating psychological safety and organizational learning during the COVID-19 pandemic. Frontiers in Education 5 : 618483. World Bank (2018) World Development Report 2018. Learning to Realize Education’s Promise. Washington, DC: The World Bank. World Bank (2020) The COVID-19 Pandemic: Shocks to Education and Policy Responses. Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/33696. Wu MJ Zhao K Fils-Aime F (2022) Response rates of online surveys in published research: a meta-analysis. Computers in Human Behavior Reports 7 : 100206. Yuting Z Adams D Lee KCS (2022) The relationship between technology leadership and teacher ICT competency in higher education. Education and Information Technologies 27 (7 ): 10285–10307.
PMC010xxxxxx/PMC10291297.txt	==== Front Sports Health Sports Health SPH spsph Sports Health 1941-7381 1941-0921 SAGE Publications Sage CA: Los Angeles, CA 36154544 10.1177/19417381221120639 10.1177_19417381221120639 Current Research Effects of Moderate-Intensity Training Under Cyclic Hypoxia on Cardiorespiratory Fitness and Hematological Parameters in People Recovered From COVID-19: The AEROBICOVID Study Dellavechia de Carvalho Carlos MD * Bertucci Danilo Rodrigues PhD Ribeiro Felipe Alves MD Costa Gabriel Peinado MD Toro Diana Mota PhD Camacho-Cardenosa Marta PhD Brazo-Sayavera Javier PhD Sorgi Carlos Arterio PhD Papoti Marcelo PhD Trapé Átila Alexandre PhD * Carlos Dellavechia de Carvalho, MD, Ribeirao Preto Medical School, University of Sao Paulo, Avenida Bandeirantes 3900, Vila Monte Alegre, Ribeirão Preto, SP, Brazil (email: carlos_dellavechia@hotmail.com) 25 9 2022 Jul-Aug 2023 25 9 2022 15 4 558570 © 2022 The Author(s) 2022 American Orthopaedic Society for Sports Medicine This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background: Recent studies have indicated that people who live at altitude have a lower incidence of coronavirus disease (COVID-19) and lesser severity in infection cases. Hypothesis: Hypoxia exposure could lead to health benefits, and it could be used in the recovery process as an additional stimulus to physical training to improve cardiorespiratory fitness (CRF). Study Design: Randomized controlled clinical trial. Level of Evidence: Level 2. Methods: The 43 participants, aged 30 to 69 years, were divided into control group (CG, n = 18) and 2 training groups: normoxia (NG, n = 9) and hypoxia (HG, n = 16). Before and after the intervention were evaluated the lactate threshold 2 (L2), peak oxygen uptake (VO2peak), and a blood sample was collected at rest to evaluate hematological adaptation. Both groups performed an 8-week moderate-intensity physical training on a bike. The HG were trained under normobaric hypoxic conditions (fractional inspired oxygen [FiO2] = 13.5%). Results: The 8-week intervention promoted a similar improvement in CRF of people recovered from COVID-19 in the HG (L2 = 34.6%; VO2peak = 16.3%; VO2peak intensity = 24.6%) and NG (L2 = 42.6%; VO2peak = 16.7%; VO2peak intensity = 36.9%). Only the HG presented differences in hematological variables (erythropoietin = 191.7%; reticulocytes = -32.4%; off-score = 28.2%) in comparison with the baseline. Conclusion: The results of the present study provide evidence that moderate-intensity training in normoxia or hypoxia promoted similar benefits in CRF of people recovered from COVID-19. Furthermore, the hypoxia offered an additional stimulus to training promoting erythropoietin increase and hematological stimulation. Clinical Relevance: The present exercise protocol can be used for the rehabilitation of people recovered from COVID-19, with persistent low CRF. In addition, this is the first study demonstrating that physical training combined with hypoxia, as well as improving CRF, promotes greater hematological stimulation in people recovered from COVID-19. altitude erythropoietin aerobic power aerobic capacity coronavirus universidade de são paulo https://doi.org/10.13039/501100005639 Vida Project - 3518Â?2020 universidade de são paulo https://doi.org/10.13039/501100005639 Integrated Research Projects in Strategic Areas from the Dean of Research - 2021.1.10424.1.9 typesetterts1 cover-dateJuly/August 2023 ==== Body pmcIn March 2020, the World Health Organization classified the coronavirus disease (COVID-19) caused by the severe acute respiratory coronavirus 2 (SARS-CoV-2) as a pandemic. 52 News related to COVID-19 is commented on in daily media and has great concern and impact on Global Public Health. 23 The COVID-19 pandemic is an unprecedented health emergency of our era, causing mortality and disease worldwide. The clinical condition is diverse, ranging from asymptomatic infection to acute respiratory syndrome and damage to various body systems, increasing inflammatory markers, cardiovascular disorders, lung injuries, and kidney damage. 23 However, a new demand arises in the post-COVID context because some symptoms can last and limit people recovered from COVID-19. After recovery, it is possible to identify people with alterations in cardiovascular and pulmonary systems, in addition to the hematological parameters.5,23,28 Pulmonary injuries and cardiovascular disorders that impair cardiorespiratory fitness (CRF) have been described predominantly for hospitalized people with COVID-19 but also in asymptomatically infected individuals. 2 The persistent symptoms related to this context are associated with a measurable functional deficit in physical fitness, highlighting the reduced CRF.14,27 Clavario et al 12 determined the functional capacity of COVID-19 survivors in addition to the safety and tolerability of using cardiopulmonary exercise testing (CPET) in 225 patients with confirmation of COVID-19 3 months after hospital discharge. It was verified that 88% of the patients had peak oxygen consumption (VO2peak) below the predicted. The authors highlighted that 80% of patients experienced at least 1 disabling symptom, an unrelated decrease of VO2peak and functional capacity. They concluded that approximately 33% of COVID-19 survivors have functional limitations 3 months after discharge, associated with muscle impairment. Recent studies indicate that people who live at high altitudes (above 3000 m) have a lower incidence of COVID-19 and less severity in cases of infection.1,3,11,30,32,48 In addition, Brazilian cities with high altitude and low relative humidity have a reduced relative incidence and mortality rate of COVID-19. 21 The factors involved in this lower susceptibility to COVID-19 are related to physiological and anatomical adaptations in the lungs, improving perfusion and capacity. Furthermore, the increase in erythropoietin (EPO) concentrations generates a cytoprotective effect with a broad function that reduces inflammatory conditions and microvascular lesions. 1 In addition, recent studies have demonstrated the possibility of using EPO as an auxiliary approach in treating COVID-19.16,35,46,53 On the other hand, moderate-intensity interval exercise itself can reduce chronic inflammation and strengthen the immune system,8,42,51,59 reducing the severity and mortality of viral diseases. 34 In addition, higher levels of CRF can produce short-term improvements in the immune and respiratory systems, 40 both affected by COVID-19. 61 Training methods using hypoxia as an ergogenic resource (cyclic hypoxia) have existed since the 1960s.17,18,25,44,50 The most recent studies have determined that this type of intervention can present favorable results in health parameters and that the physical training associated with the normobaric hypoxia condition is safe and can be performed with different populations, for example, in the reduction of the fat mass with a concomitant increase of lean mass, 10 and increased CRF. 9 Thus, we speculated that an interval training program of moderate-intensity performed in cyclic normobaric hypoxia could be an efficient proposal in rehabilitating people recovered from COVID-19 to improve their damaged CRF and increase the hematological stimulation. For this, we aimed to study the effects of 8 weeks of moderate-intensity cyclic hypoxic training on the cardiorespiratory capacity and hematological responses of people recovered from COVID-19. Methods Ethical Review This study was approved by an institutional review board. All participants gave written informed consent. Participants Sixty-nine participants were recruited, of which 43 completed all assessments and were therefore included in the study according to the following inclusion criteria: men and women aged between 30 and 69 years, approximately 30 days since the recovery of clinical signs or medical discharge (in case of hospitalization); and having previous experience in aerobic exercise. The exclusion criteria were as follows: exposure to an altitude higher than 1500 m in the last 3 months; significant physical limitations to carry out the evaluations and intervention; acute or chronic clinical illnesses without medical supervision; anemias; use of immunosuppressive drugs; pregnant women; hormone replacement; smokers; and excessive use of alcohol or drugs. In addition, an evaluation of the health status was carried out. The participants who did not present limitations or discomfort that could prevent the performance of the evaluations or the intervention were enrolled in the proposed intervention. Experimental Design The study design is a randomized controlled clinical trial composed of 3 groups: the control group (CG, n = 18), participants who were not available to join the intervention and accepted to carry out a follow-up through the evaluations; and the physical training groups, which were randomly divided according to the association of training with hypoxia (HG, n = 16) or normoxia (NG, n = 9). The experimental protocol of the AEROBICOVID study (Figure 1) consisted of (1) familiarization and carrying out the initial evaluation (baseline [BSLN]) in the 3 sessions of week 0, with CPET on a bike and blood collection, (2) an 8-week intervention with a partial evaluation developing a CPET to adjust the training load between weeks 4 and 5 (half of the intervention), (3) reevaluation at week 9 with the same evaluations of week 0, following the end of the intervention (post), all biological concerns were described elsewhere. 56 Figure 1. Experimental design illustration. Blood drop, blood collection; CG, control group; HG, hypoxia group; HR, heart rate; NG, normoxia group; RPE, rate of perceived exertion; VO2, oxygen uptake; W, week. Assessments Anthropometric Variables Body mass and height were measured by a 200 kg capacity electronic weighing scale (Mic-Pp, Micheletti), enabling calculation of the body mass index (BMI) using the equation body mass/height2. Questionnaires As control variables, the International Physical Activity Questionnaire (IPAQ) short version, validated in Brazil 37 measured the usual level of physical activity. In addition, the Food Consumption Markers Form of the Ministry of Health 39 was used to assess the frequency of food consumption. Participants were instructed to maintain similar physical activity and eating habits during the study. Hematological Parameters Blood collection was performed by peripheral venous access after 8 hours overnight fasting, carried out by a trained and specialized professional. The hemogram parameters, such as total red blood cell count, hematocrit, and hemoglobin concentration, were evaluated at the Clinical Analysis Laboratory, Faculty of Pharmaceutical Sciences of Ribeirão Preto according to the technical service’s standard routine and methodology. The hematologic stimulation index (off-score) was obtained from hemoglobin (Hb in g/L) and reticulocytes (Ret %): off-score = Hb (in g/L) 60√Ret(%).26,47 The EPO plasma concentration was determined by immunoassay according to the fabricant (EPO ELISA Kit R&D Systems). Peak Aerobic Power (VO2peak) A CPET was used to estimate lactate thresholds 1 (L1) and 2 (L2), VO2peak and iVO2peak. The CPET was performed in a pendular cycle ergometer with mechanical braking (Ergométrica, Monark). First, the participants started a 5 minute warm-up without any additional load; after that, the intensity was increased by 15 watts every 2 min until the participant did not maintain the 60 rpm cadence or volitional exhaustio Oxygen uptake was measured breath by breath by the gas analyzer (K4b2, COSMED), calibrated according to the manufacturer’s specifications. The VO2peak was defined as the highest VO2 average in the last 60 seconds in the test. The iVO2peak was the lowest intensity at which the individual reached VO2peak during the test. If the participant did not complete the last stage of the incremental test, the iVO2peak is estimated by the equation proposed by Kuipers et al. 33 Blood samples (25 μL) were collected from the earlobe at the end of each stage, using previously calibrated heparinized capillaries. Blood samples were immediately dispensed and homogenized in microtubes containing 1% sodium fluoride for blood lactate concentration [La] analysis using the YSI 2300 STAT analyzer (YSI, Yellow Springs, OH). The bisegmented method was used, in which the points obtained between [La] and the intensities were subjected to 3 linear adjustments so that the 2 intersections obtained will be assumed as L1 and L2 intensities. 41 Concomitantly, heart rate (HR) and rate of perceived exertion (RPE) were monitored at the end of each stage. Intervention Hypoxia Instrumentation A unidirectional mask (Air Safety) was connected to a 3 m long flexible hose (IVPU, vacuum air PU 1.1/2 cm), and the opposite end was connected to a tent (2 m wide, 3m long, and 2 m high; Colorado Altitude Training Tent), with 12,000 l of air capacity, connected to a hypoxia generator (CAT 430, Altitude Control Technologies). The system was turned on throughout the experiment because the air outlet in the hypoxia generator is approximately 50 L/min. Therefore, the participants’ average ventilation after each effort could, in some cases, be greater than 120 L/min. Thus, it was possible to guarantee air safety with a low inspired fraction of oxygen (FiO2 = 13.5%) for the entire training session. The oxygen concentration monitoring inside the tent was made through an oxygen sensor (Oxygen Sensor R-17MED, Teledyne Analytical Instruments). The detailed strategy for hypoxia instrumentation is available in the published protocol of this study. 56 For the NG, the participants performed the same procedures; however, the opposite hose end received ambient air (FiO2 = 20.9%) without the participant’s knowledge (blind procedure). The altitude where the experiment occurred was 540 m above sea level. Hypoxic Dose Calculation and SpO2 by FiO2 index The hypoxic dose was initially defined in km∙h = (m/1000)∙h, 22 considering that participants may show different peripheral oxygen saturation (SpO2) reductions for the same FiO2, leading to lower oxygen availability in hypoxia conditions concomitantly with an oxygen demand increase resulting from exercise. Then, each participant’s hypoxic dose was estimated using 2 distinct methods assumed from the product between exposure time in hours (t) by the SpO2 average reduction in the training session, considering a resting SpO2 value corresponding to 98%. Thus, hypoxic dose = (98 SpO2 average) ∙ t, adapted from Millet et al. 38 It was also calculated by the product between SpO2 by FiO2, the index (SF) proposed by Soo et al. 54 Training Program The training sessions were performed thrice a week, with a total duration of up to 50.5 min. The initial part (warm up) and the final stage (cool down) lasted for 5 and 3 min, respectively, and were performed in low intensity, corresponding to “easy” by the RPE. In the main part, the intensity was based on L2 values obtained in the initial and partial evaluation (90-100% L2 from first to the fourth week, and 100-110% L2 from fifth to the eighth week). Each training set consisted of efforts lasting 5 min in the intensity of L2 with a break of 2.5 min between efforts. According to the week of training, the number of sets variated from 3 to 6 sets (Figure 1). HR and RPE were used to control the intensity during all training periods. Furthermore, blood oxygen saturation was monitored to evaluate the hypoxia and training responses using a pulse oximeter (D300C1, Dellamed). The measurements were recorded in 3 moments: rest, end of each effort, and end of the break. Statistical Analyses The data normality was checked using the Shapiro-Wilk test. After the normality confirmation, data were expressed in means (SD). Then data were analyzed using a 2 × 3 (time [BSLN × 8-W] × group [CG × HG × NG]) repeated-measures analysis of variance (ANOVA) to verify the possible variations in statistics. Additionally, the effect size (η2) was calculated for the comparisons between groups and interpreted according to Cohen. 13 Statistical difference was defined as P < 0.05, and the Bonferroni post hoc test was used if necessary. The statistical tests were performed with the software JASP (Version 0.13.1.0). Responsiveness to intervention was computed by comparing typical error (TE) and the smallest worthwhile change (SWC). 29 TE was calculated by dividing the SD of the trial-to-trial difference score by √2. 29 The SWC was derived from between-subject SD multiplied by either 0.2,6,29 representing the typical small effect. The option to present the “thresholds” (SWC and 2 × TE) is due to their relevance in clinical application, ie, changes greater than SWC and especially 2 × TE, regardless of statistical significance, may indicate clinical significance. It was considered responsive for those who, for each variable, present values higher than 2 times TE.6,29,57 Those analyses were performed on Microsoft Office Excel spreadsheets. Results Control Variables No differences were observed between groups in food records between baseline and 8 weeks (Appendix Tables A1 and A2, available in the online version of this article). In addition, no alterations were observed in walking and sitting times. Still, an increase in physical activity levels (moderate and vigorous) related to participation in the intervention program was observed (Appendix Table A3, available online). Anthropometric Variables No statistics differences were observed between groups in age (years), CG 47.3 (10.5), NG 50.1 (10.5) and HG 49.0 (9.7), P = 0.78, and effect size (h2 = 0.012). The participant’s body mass and BMI are listed in Table 1, and no statistical alterations were observed; therefore, the differences in body mass and BMI were lower than the smallest worthwhile change. Table 1. Anthropometric variables for the 3 experimental groups Time Group Time × Group BSLN 8W Diff P Value η2 P Value η2 P Value η2 Body mass (kg) CG 85.28 (14.47) 85.86 (15.06) 0.58 NG 79.78 (16.04) 80.17 (16.56) 0.39 0.94 9.214-7 0.36 0.050 0.13 6.399-4 HG 88.68 (10.08) 87.79 (10.87) -0.89 BMI (kg/m2) CG 30.22 (4.33) 30.41 (4.45) 0.19 NG 29.04 (3.13) 29.16 (2.99) 0.12 0.91 3.009-6 0.58 0.027 0.12 0.001 HG 31.00 (3.88) 30.65 (3.84) -0.35 BMI, body mass index; BSLN, baseline; CG, control group; Diff, difference; 8W, 8 weeks; η2, effect size; HG, hypoxia group; NG, normoxia group; SWC, smallest worthwhile chance. Internal Training Load and Hypoxic Dose Internal training loads, obtained from exercise time and RPE average, were similar between groups (NG = 3071 ± 742 a.u., and HG = 2954 ± 1032 a.u.). The hypoxic dose, calculated during the training sessions from FiO2, was 9.97 for NG and 64.6 km/h for HG. At the same time, the hypoxic dose calculated by peripherical oxygen saturation was 160.5 (93.2) for NG and 1471.3 (354.6) %∙h for HG. Thus, in each session, the SF index means calculated from FiO2 and SpO2 were 464.0 (3.1) and 657.2 (17.5) SpO2∙FiO2-1 for the NG and HG, respectively. Maximum Parameters of CPET Table 2 describes the results obtained at the maximum intensity reached by the participants in the CPET at baseline and 8 weeks. Statistical improvements were observed in VO2peak (L/min) and iVO2peak in both trained groups (NG and HG) compared with the baseline. However, no difference was observed between them. No differences were found in the CG. Regarding the relation to body mass VO2peak (mL/kg/min), HRpeak, and [La]peak no changes were detected. Statistical changes in RPE were observed only in HG after 8 weeks training (P = 0.01). Table 2. Effects of moderate-intensity cyclic hypoxic training on cardiorespiratory capacity and monitoring variables of the incremental test of patients recovered from COVID-19 Time Group Time × Group BSLN 8W Diff P Value η2 P Value η2 P Value η2 VO2peak (mL/min) CG 1801.8 (414.7) 1843.6 (479.2) 41.8 NG 1763.3 (634.2) 2058.6 (713.5)* 295.3 <0.01 0.039 0.85 0.007 0.01 0.015 HG 1764.1 (428.4) 2051.1 (478.2) 287.0 VO2peak (mL/kg/min ) CG 21.6 (5.7) 21.9 (6.4) 0.3 NG 22.1 (6.5) 26.0 (8.3) 3.9 <0.01 0.481 0.61 0.024 <0.01 0.301 HG 20.1 (5.2) 23.6 (8.3) 3.5 iVO2peak (watts) CG 106.3 (41.2) 113.5 (42.9) 7.2 NG 95.7 (52.4) 131.0 (57.7) 35.3 <0.01 0.062 0.79 0.010 <0.01 0.017 HG 106.7 (36.2) 133.0 (37.6)** 26.3 [La]peak (mM) CG 6.8 (1.7) 5.7 (1.6) -1.1 NG 6.2 (2.5) 6.9 (1.9) 0.6 0.94 3.556-5 0.55 0.021 0.02 0.049 HG 6.6 (2.2) 7.1 (1.7) 0.6 HRpeak (bpm) CG 151.5 (18.1) 149.3 (17.4) -2.2 NG 160.2 (17.6) 158.4 (13.2) -1.8 0.78 3.372-4 0.21 0.062 0.60 0.004 HG 156.6 (13.7) 158.8 (13.2) 2.2 RPEpeak CG 8.7 (1.4) 9.2 (1.5) 0.5 NG 8.6 (1.6) 7.2 (2.3) -1.4 <0.01 0.044 0.16 0.057 <0.01 0.071 HG 9.6 (0.9) 8.2 (2.4)* -1.4 BSLN, baseline; CG, control group; COVID-19, coronavirus disease 2019; Diff, difference; 8W, 8 weeks; η2, effect size; HG, hypoxia group; HR, heart rate; iVO2, VO2 intensity; [La], lactate concentration; NG, normoxia group; P, significance level; RPE, rate of perceived exertion; VO2, oxygen uptake. Data presented as mean (SD). * P < 0.05, *P < 0.01 between BSLN and 8W in the same group. Submaximum Parameters of the CPET Tables 3 and 4 describe the values associated with the submaximum parameters of the incremental test, L1 and L2. For L1 analysis, the value of VO2 (mL/min and mL/kg/min) and power were statistically higher after the 8 weeks of training in the NG. In addition, the HG improved the power associated with L1. For L2, similar improvements were observed in VO2 (mL/min and mL/kg/min) and power in NG and HG after the intervention compared with baseline. Table 3. Effects of moderate-intensity cyclic hypoxic training on lactate threshold 1 (L1) reached in the incremental test and monitoring variables of people recovered from COVID-19 Time Group Time × Group BSLN 8W Diff P Value η2 P Value η2 P Value η2 VO2L1 (mL/min) CG 1191.1 (197.5) 1263.8 (301.3) 72.7 NG 1229.9 (215.7) 1603.8 (397.1) 373.9 <0.01 0.134 0.27 0.060 0.08 0.039 HG 1198.5 (195.0) 1411.4 (278.8) 212.9 VO2L1 (mL/kg/min) CG 13.7 (2.5) 14.5 (3.5) 0.8 NG 15.0 (1.8) 19.5 (4.7)* 4.5 <0.01 0.128 0.05 0.126 0.07 0.040 HG 13.1 (2.2) 15.6 (3.2) 2.5 IntensityL1 (watts) CG 37.5 (17.3) 49.4 (24.6) 11.9 NG 34.3 (21.1) 63.8 (33.5)* 29.5 <0.01 0.191 0.75 0.014 0.06 0.024 HG 28.0 (13.5) 54.1 (26.5)** 26.1 [La]L1 (mM) CG 1.67 (0.5) 1.33 (0.4) -0.3 NG 1.87 (0.7) 1.65 (0.6) -0.2 0.09 0.043 0.24 0.055 0.79 0.007 HG 1.82 (0.5) 1.68 (0.5) -0.1 HRL1 (bpm) CG 109.4 (15.3) 108.7 (8.8) -0.7 NG 121.4 (16.5) 123.4 (8.2) 2.0 0.42 0.008 0.04 0.148 0.56 0.014 HG 109.6 (6.9) 115.2 (12.4) 5.6 RPEL1 CG 2.3 (0.5) 2.6 (1.02) 0.3 NG 2.6 (1.3) 2.1 (0.7) -0.5 0.22 0.020 0.39 0.040 0.17 0.020 HG 3.1 (1.2) 2.5 (0.51) -0.6 BSLN, baseline; CG, control group; COVID-19, coronavirus disease 2019; Diff, difference; 8W, 8 weeks; η2, effect size; HG, hypoxia group; HR, heart rate; iVO2, VO2 intensity; [La], lactate concentration; NG, normoxia group; RPE, rate of perceived exertion; VO2, oxygen uptake. Data presented as mean (SD). * P < 0.05, *P < 0.01 between BSLN and 8W in the same group. Table 4. Effects of moderate-intensity cyclic hypoxic training on lactate threshold 2 (L2) reached in the incremental test and monitoring variables of people recovered from COVID-19 Time Group Time × Group BSLN 8W Diff P Value η2 P Value η2 P Value η2 VO2L2 (mL/min ) CG 1559.3 (321.9) 1644.7 (468.4) 85.4 NG 1508.8 (509.3) 1828.9 (577.4) 320.1 <0.01 0.060 0.92 0.004 0.10 0.01 HG 1511.5 (327.5) 1751.3 (399.0)* 239.8 VO2L2 (mL/kg/min ) CG 18.63 (4.1) 19.54 (5.6) 0.9 NG 19.00 (5.8) 23.15 (6.7)* 4.1 <0.01 0.065 0.41 0.037 0.05 0.02 HG 17.11 (3.6) 19.98 (5.6)* 2.9 IntensityL2 (watts) CG 73.8 (36.9) 87.8 (42.6) 14.0 NG 65.8 (42.1) 93.8 (43.6)* 28.0 <0.01 0.078 0.90 0.005 0.16 0.01 HG 73.1 (26.9) 98.4 (38.3)** 25.3 [La]L2 (mM) CG 3.5 (1.1) 3.2 (0.9) -0.3 NG 3.5 (1.3) 3.8 (1.3) 0.4 0.67 0.002 0.28 0.034 0.50 0.02 HG 3.7 (1.0) 4.0 (1.3) 0.3 HRL2 (bpm) CG 136.0 (14.4) 132.3 (15.4) -3.7 NG 143.4 (18.5) 144.5 (11.0) 1.1 0.77 4.376−4 0.20 0.063 0.60 0.01 HG 138.8 (11.8) 139.5 (17.0) 0.7 RPEL2 CG 4.9 (1.1) 5.2 (2.0) 0.3 NG 5.1 (2.2) 4.0 (1.3) -1.1 0.07 0.029 0.52 0.020 0.08 0.04 HG 5.5 (1.3) 4.6 (1.4) -0.9 BSLN, baseline; CG, control group; COVID-19, coronavirus disease 2019; Diff, difference; 8W, 8 weeks; η2, effect size; HG, hypoxia group; HR, heart rate; iVO2, VO2 intensity; [La], lactate concentration; NG, normoxia group; RPE, rate of perceived exertion; VO2, oxygen uptake. Data presented as mean (SD). * P < 0.05, *P < 0.01 between BSLN and 8W in the same group. Figure 2 shows, from the individual changes related to SWC and 2 × TE, that both NG and HG improved the aerobic capacity and power variables compared with CG. Figure 2. Mean and SD of the difference after 8 weeks of intervention compared to the smallest worthwhile change (SWC), and 2 × typical error (TE) for variables absolute VO2peak (a), VO2L1 (b), VO2L2 (c), relative VO2peak (d), VO2L1 (e), VO2L2 (f), iVO2peak (g), intensityL1 (h) and intensityL1 (i). CG, control group; HG, hypoxia group; NG, normoxia group; L1, lactate threshold 1; L2, lactate threshold 2; VO2, oxygen uptake; gray circles, CG; red circles, NG; blue circles, HG. Significant differences were not observed for erythrocyte counts, hemoglobin concentration, and hematocrit percentage, comparing the results after 8 weeks training and baseline (Table 5). Conversely, EPO levels and off-score were found to be higher in all 8-week experimental groups than at baseline (P < 0.01), but nondifference between training groups protocols were demonstrated (CG versus NG or HG). On the other hand, a decreased level of reticulocytes was observed for all 8-week groups, which was also statistically different in comparison to baseline (P < 0.01), but also nondifference between experimental groups (Table 5). In addition, HG showed an improvement in EPO and off-score and a decrease in reticulocytes after the 8-week intervention. Table 5. Effects of moderate-intensity cyclic hypoxic training on hematological parameters of people recovered from COVID-19 Time Group Time × Group BSLN 8W Diff P Value η2 P Value η2 P Value η2 EPO (mIU/mL) CG 8.7 (9.4) 14.6 (12.6) 5.9 NG 3.4 (2.7) 13.4 (8.8) 10.0 <0.01 0.170 0.40 0.033 0.04 0.024 HG 7.2 (7.4) 21.0 (13.8) 13.8 Erythrocytes (m/μL) CG 4.6 (0.3) 4.7 (0.4) 0.1 NG 4.5 (0.2) 4.5 (0.2) -0.1 0.90 3.235-5 0.64 0.020 0.28 0.005 HG 4.7 (0.4) 4.7 (0.4) -0.0 Hemoglobin (g/dL) CG 14.1 (1.1) 14.2 (1.2) 0.2 NG 13.9 (0.8) 13.7 (0.7) -0.1 0.73 1.673-4 0.75 0.013 0.38 0.003 HG 14.1 (1.6) 14.1 (1.6) 0.1 Hematocrit (%) CG 41.8 (2.7) 42.4 (3.1) 0.6 NG 41.6 (2.1) 40.8 (1.6) -0.8 0.94 1.388-5 0.36 0.044 0.46 0.044 HG 42.5 (4.8) 42.6 (4.3) 0.1 Reticulocytes (m/mm3) CG 86.0 (24.1) 78.2 (19.0) -7.8 NG 88.2 (26.5) 70.4 (24.8) -17.8 <0.01 0.088 0.95 0.002 0.01 0.030 HG 99.0 (46.1) 66.9 (30.4)* -32.1 Off-Score CG 60.3 (18.4) 68.5(19.7) 8.2 NG 56.5 (18.8) 63.1(18.2) 6.6 <0.01 0.056 0.82 0.008 0.18 0.010 HG 56.4 (25.6) 72.3(20.2)* 15.9 BSLN, baseline; CG, control group; COVID-19, coronavirus disease 2019; Diff, difference; 8W, 8 weeks; EPO, erythropoietin; η2, effect size; HG, hypoxia group; NG, normoxia group. * Considered to P < 0.05 between 8W and BSLN in the same group. Figure 3 shows, from the individual changes above SWC and 2 × TE, that both NG and HG improved EPO and off-score, and decreased reticulocytes compared to CG, but not for erythrocytes, hemoglobin, and hematocrit. Figure 3. Mean and SD of the difference after 8 weeks of intervention compared to the smallest worthwhile change (SWC), and 2 × typical error (TE) (gray area) for blood variables absolute. (a) EPO (erythropoietin); (b) erythrocytes; (c) hemoglobin; (d) hematocrit; (e) reticulocytes; (f) off-score. CG, control group; HG, hypoxia group; NG, normoxia group; gray circles, CG; red circles, NG; blue circles, HG. Discussion The main findings of the present study were that 8 weeks of moderate-intensity training improves the CRF of people recovered from COVID-19. In addition, hypoxia promoted advances similar to training in normoxia, with a greater hematological stimulation. Other studies had systematically reported the relationships between maximal exercise capacity 7 and CRF (ie, VO2peak) 45 and survival rate for patients with various pathologies. In the case of symptomatic COVID-19 individuals, as well as other respiratory diseases, a reduction in VO2max has also been observed. Barbagelata et al, 4 in a cross-sectional study, showed that patients with the post-COVID-19 syndrome had significantly lower (~10%) VO2peak (25.8 ± 8.1 mL/kg/min) compared with asymptomatic individuals (28.8 ± 9.6 mL/kg/min). Debeaumont et al 15 evaluated physical fitness and its relationship with functional dyspnea by performing CPET in COVID-19 survivors 6 months after hospital discharge, those admitted to the general ward had a relatively preserved VO2peak (87%), while those requiring the intensive care unit had a moderately reduced VO2peak to 77%. These authors concluded that persistent dyspnea was associated with reduced physical fitness at 6 months. In the present study, participants who underwent the experimental training program (NG and HG) showed significant improvements in VO2peak of 3.9 mL/kg/min and 3.5 mL/kg/min, respectively, besides the fact that a clinic change was observed for these groups, because most individuals showed responsiveness (changes greater than SWC and 2 × TE) to the intervention, while the CG does not significantly modify their values. These results suggest that the proposed training model effectively improved the aerobic power of people recovered from COVID-19. Compared with NG, HG also showed similar improvements in all physiological responses (L1, L2, and VO2) and in their respective workloads. We understand that this is a significant result from a practical point of view, especially for the HG participants, who achieved aerobic and functional gains similar to the NG, with a likely reduction in the external workload during exercise in hypoxia compared with the same training model performed in normoxia. This assumption is made because the evidence50,60 supports that hypoxic training impairs the capacity to maintain the same intensity as under normoxia conditions. Yano and Asano 60 found 12% and 39% reductions in the lactate threshold (LT) workload determined at 2000 m [(596 mm Hg) (759 ± 81 kpm/min)] and 4000 m [(462 mm Hg) (591 ± 60 kpm/min)] hypobaric hypoxia compared with LT obtained at sea level (861 ± 45 kpm/min). Sharma et al 50 found reductions of 6% and 4% in the intensity of LT and VO2peak, respectively, of mid distance runners at 2100 m of normobaric hypoxia. Furthermore, these authors concluded that, in general, altitude training at the same intensity seems to correspond to an increase in the difficulty of approximately 30%. In another study, these authors 50 found 5.5% reductions in velocity at VO2max (20.1 ± 1.3 vs 19.0 ± 1.0 km/h) in highly trained runners at 2100 m. The highest proportion of participants with relevant (higher than SWC) changes in VO2peak was observed in participants who trained in hypoxia, even though both groups showed similar substantial increases in intensity corresponding to VO2peak. NG and HG participants having undergone the same training protocol, the intensity of the stimuli was prescribed based on the internal training load variables (ie, HR and RPE), corresponding to L2. In fact, in the present investigation, the internal training loads were similar, and corroborated our findings, Liu et al 36 reported similar gains from training in hypoxia (FiO2 = 15.3%) compared with the same program performed in normoxia with exercise intensities based on HR corresponding to 80% of VO2peak, despite the load, the average external work of HG was 25% lower than NG. Collectively, these studies seem to demonstrate that during training performed at altitude, the absolute workload in LT and VO2peak is substantially reduced, and this presents a significant advantage, especially in the health area, because it is possible to achieve the same internal load with less mechanical stress. A study pointed out that the first adaptations usually observed after sufficient exposure to hypoxia are hematologic, with an increase in the number of erythrocytes and hematocrit leading to more oxygen transport. 49 However, training performed in hypoxia can also affect other genetic factors controlled by HIF-1α that are associated with performance adaptations and muscle adaptations without necessarily increasing oxygen carrying.20,55 Additionally, our results showed that participants belonging to the HG increased the EPO levels compared to CG and NG. Although EPO and blood parameters were performed before and after 8 weeks, it is still possible to observe positive clinical effects of hypoxia on the reduction of reticulocytes with a concomitant increase in off-score. These last results might be clinically relevant because EPO stimulates the production of erythrocytes and, consequently, red blood cells, which facilitate oxygen transport to the target tissues. 31 There is still no consensus in the literature regarding the minimum dose needed to produce EPO. Wojan et al. 58 recently investigated the effects of hypoxia exposure itself on EPO production. They found that eight 4 minute passive cycles of intermittent hypoxia, with a target SpO2 of 80%, represent the shortest protocol to increase serum EPO levels in healthy individuals. Therefore, despite being slightly lower than recommended during passive exposure, the hypoxia doses used in the present investigation were sufficient to stimulate EPO increases, probably due to the combination with moderate-intensity training. As clinical changes in hematological variables were observed for all groups, this phenomenon could mean a natural process of EPO increase and reticulocyte levels decrease after a long time post-COVID-19 recovery than an effect of training. The increase in EPO, predominantly observed for HG, may represent particular importance for people recovered from COVID-19. Pramsohler et al 43 recently investigated the relationship between COVID-19 and EPO levels in 59 COVID-19 patients hospitalized in the intensive care unit divided according to disease severity into mild, severe, and critical. Reduced hemoglobin levels were found in the critically ill group and the group of deaths. In addition, the coefficient of variation of the red blood cell distribution width and the ferritin values were significantly higher in the intubated and deceased groups. Finally, it was found that the EPO levels of patients who died were substantially lower than the CG and the group of surviving patients. The other aspect refers to a greater hematological stimulation promoted by hypoxia, which can be seen by the decrease of 32.5% in reticulocytes, concomitantly with the substantial increase of 28% of the off-score for HG. Faulhaber et al, 19 using a “single-blind” model, compared the effects of exposure to hypoxia (continuous and cyclic) on “key markers” of hematologic adaptation, stress, and cardiac damage in elderly people. Both hypoxia protocols lasted approximately 70 minutes, and SpO2 severity was 85%. Red cell content increased only on day 5 of exposure to hypoxia, compared with baseline values (+7.7%, P < 0.01), whereas hematocrit and off-score increased only at the end of the experiment. The authors concluded that there are differences in responses arising from continuous and cyclic hypoxia protocols when the objective is to stimulate hematological alterations. Despite discussions about the risks and benefits of using EPO,16,24,46 clinical and randomized studies are still needed to demonstrate its effectiveness in people recovering from COVID-19. In the present study, the EPO increment was succeeded from the combination of exposure to hypoxia and moderate-intensity aerobic training, which is, therefore, a possible nonpharmacological strategy to improve EPO levels and hematological parameters in individuals recovered from COVID-19. This study had significant limitations, including the external load during the training sessions that was not controlled, limiting the conclusions regarding the exercise dose performed for each experimental group. In addition, the number of participants allocated to each group is reduced, especially in NG, due to participants’ dropout during the project. Therefore, the gravity during the disease and the impairment level after COVID infection were not the same for all participants. Also, age and physical fitness distributions were heterogeneous because diverse populations were enrolled to provide a more generalizable clinical approach. In conclusion, based on the findings reported in the present study, 8 weeks of moderate-intensity training in normoxia or hypoxia promoted similar benefits in CRF of people recovered from COVID-19. Finally, the hypoxia exposure was an additional stimulus to training, which increased EPO levels and promoted hematological stimulation. Therefore, this type of intervention is suggested as an alternative nonpharmacological treatment for individuals recovering from COVID-19. Supplemental Material sj-docx-1-sph-10.1177_19417381221120639 – Supplemental material for Effects of Moderate-Intensity Training Under Cyclic Hypoxia on Cardiorespiratory Fitness and Hematological Parameters in People Recovered From COVID-19: The AEROBICOVID Study Click here for additional data file. Supplemental material, sj-docx-1-sph-10.1177_19417381221120639 for Effects of Moderate-Intensity Training Under Cyclic Hypoxia on Cardiorespiratory Fitness and Hematological Parameters in People Recovered From COVID-19: The AEROBICOVID Study by Carlos Dellavechia de Carvalho, Danilo Rodrigues Bertucci, Felipe Alves Ribeiro, Gabriel Peinado Costa, Diana Mota Toro, Marta Camacho-Cardenosa, Javier Brazo-Sayavera, Carlos Arterio Sorgi, Marcelo Papoti and Átila Alexandre Trapé in Sports Health: A Multidisciplinary Approach The authors acknowledge and thank all the participants of the AEROBICOVID study and University of Sao Paulo (USP) employees working in the pandemics and helping develop this project. It also used in this work the hypoxia equipment acquired from the Foundation for Research Support of the State of São Paulo financial support (Process No. 2016/12781-5). The authors report no potential conflicts of interest in the development and publication of this article. This work was supported by the USP Vida Project (3518–2020) and the call for Integrated Research Projects in Strategic Areas from the Dean of Research, USP (2021.1.10424.1.9). The funders do not have a role in the study design and collection, analysis, and interpretation of data and manuscript writing. ==== Refs References 1 Accinelli RA Leon-Abarca JA . En la altura la COVID-19 es menos frecuente: la experiencia del Perú. Arch Bronconeumol. 2020;56 :760-761.32782091 2 Anaya JM Rojas M Salinas ML , et al . Post-COVID syndrome. A case series and comprehensive review. Autoimmun Rev. 2021;20(11):102947. 3 Arias-Reyes C Zubieta-DeUrioste N Poma-Machicao L , et al . Does the pathogenesis of SARS-CoV-2 virus decrease at high-altitude? Respir Physiol Neurobiol. 2020;277 :103443.32333993 4 Barbagelata L Masson W Iglesias D , et al . Cardiopulmonary exercise testing in patients with post-COVID-19 syndrome. Med Clin (Barc). 2022;159 (1 ):6-11. 5 Bhatt T Kumar V Pande S Malik R Khamparia A Gupta D . A Review on COVID-19. Stud Comput Intell. 2021;924 :25-42. 6 Bonafiglia JT Nelms MW Preobrazenski N , et al . Moving beyond threshold-based dichotomous classification to improve the accuracy in classifying non-responders. Physiol Rep. 2018;6 :1-9. 7 Brawner CA Al-Mallah MH Ehrman JK Qureshi WT Blaha MJ Keteyian SJ . Change in maximal exercise capacity is associated with survival in men and women. Mayo Clin Proc. 2017;92 :383-390.28185659 8 Burtscher J Burtscher M Millet GP . (Indoor) isolation, stress, and physical inactivity: Vicious circles accelerated by COVID-19? Scand J Med Sci Sport. 2020;30 :1544-1545. 9 Camacho-Cardenosa A Camacho-Cardenosa M Brazo-Sayavera J Timón R González-Custodio A Olcina G . Repeated sprint in hypoxia as a time-metabolic efficient strategy to improve physical fitness of obese women. Eur J Appl Physiol. 2020;120 :1051-1061.32185477 10 Camacho-Cardenosa A Camacho-Cardenosa M Burtscher M , et al . High-intensity interval training in normobaric hypoxia leads to greater body fat loss in overweight/obese women than high-intensity interval training in normoxia. Front Physiol. 2018;9 :1-13.29377031 11 Choquenaira-Quispe C Saldaña-Bobadilla V Ramirez JK . Factors involved in low susceptibility to COVID-19: an adaptation of high altitude inhabitants. Med Hypotheses. 2020;143 :110068.32634733 12 Clavario P De Marzo V Lotti R , et al . Cardiopulmonary exercise testing in COVID-19 patients at 3 months follow-up. Int J Cardiol. 2021;340 :113-118.34311011 13 Cohen J . Statistical Power Analysis for the Behavioral Sciences. Revised. (Press A, ed.).; Routledge, Abingdon, England; 2013. 14 Crameri GAG Bielecki M Züst R Buehrer TW Stanga Z Deuel JW . Reduced maximal aerobic capacity after COVID-19 in young adult recruits, Switzerland, May 2020. Eurosurveillance. 2020;25 :4-7. 15 Debeaumont D Boujibar F Ferrand-Devouge E , et al . Cardiopulmonary exercise testing to assess persistent symptoms at 6 months in people with COVID-19 who survived hospitalization – a pilot study running. Phys Ther. 2021;101 (6 ):pzab099. 16 Ehrenreich H Weissenborn K Begemann M Busch M Vieta E Miskowiak KW . Erythropoietin as candidate for supportive treatment of severe COVID-19. Mol Med. 2020;26 :58.32546125 17 Faiss R Girard O Millet GP . Advancing hypoxic training in team sports: From intermittent hypoxic training to repeated sprint training in hypoxia. Br J Sports Med. 2013;47 :1-8.23367519 18 Faiss R Léger B Vesin JM , et al . Significant molecular and systemic adaptations after repeated sprint training in hypoxia. PLoS One. 2013;8 :1-13. 19 Faulhaber M Gröbner K Rausch L Gatterer H Menz V . Effects of acute hypoxia on lactate thresholds and high-intensity endurance performance-a pilot study.Int J Environ Res Public Health. 2021;18 :7573.34300024 20 Favier FB Britto FA Freyssenet DG Bigard XA Benoit H . HIF-1-driven skeletal muscle adaptations to chronic hypoxia: molecular insights into muscle physiology. Cell Mol Life Sci. 2015;72 :4681-4696.26298291 21 Fernandes JSC da Silva RS Silva AC Villela DC Mendonça VA Lacerda ACR . Altitude conditions seem to determine the evolution of COVID-19 in Brazil. Sci Rep. 2021;11 :4402.33623105 22 Garvican-Lewis LA Sharpe K Gore CJ . Time for a new metric for hypoxic dose? J Appl Physiol. 2016;121 :352-355.26917695 23 Ge H Wang X Yuan X , et al . The epidemiology and clinical information about COVID-19. Eur J Clin Microbiol Infect Dis. 2020;39 :1011-1019.32291542 24 Geier MR Geier DA . Respiratory conditions in coronavirus disease 2019 (COVID-19): important considerations regarding novel treatment strategies to reduce mortality. Med Hypotheses. 2020;140 :109760.32344310 25 Girard O Brocherie F Millet GP . Effects of altitude/hypoxia on single- and multiple-sprint performance: a comprehensive review. Sport Med. 2017;47 :1931-1949. 26 Gore CJ Hahn AG Aughey RJ , et al . Live high:train low increases muscle buffer capacity and submaximal cycling efficiency. Acta Physiol Scand. 2001;173 :275-286.11736690 27 Guzik TJ Mohiddin SA Dimarco A , et al . COVID-19 and the cardiovascular system: implications for risk assessment, diagnosis, and treatment options. Cardiovasc Res. 2020;116 :1666-1687.32352535 28 Harapan H Itoh N Yufika A , et al . Coronavirus disease 2019 (COVID-19): a literature review. J Infect Public Health. 2020;13 :667-673.32340833 29 Hopkins WG . Measures of reliability in sports medicine and science. Sport Med. 2000;30 :1-15. 30 Intimayta-Escalante C Rojas-Bolivar D Hancco I . Letter to the editor: influence of altitude on the prevalence and case fatality rate of COVID-19 in Peru.High Alt Med Biol. 2020;21 :426-427.32803989 31 Jelkmann W . Erythropoietin. Front Horm Res. 2016;47 :115-127.27348128 32 Joyce KE Weaver SR Lucas SJE . Geographic components of SARS-CoV-2 expansion: a hypothesis. J Appl Physiol. 2020;129 :257-262.32702272 33 Kuipers H Verstappen FTJ Keizer HA Geurten P van Kranenburg G . Variability of aerobic performance in the laboratory and its physiologic correlates. Int J Sports Med. 1985;6 :197-201.4044103 34 Laddu DR Lavie CJ Phillips SA Arena R . Physical activity for immunity protection: inoculating populations with healthy living medicine in preparation for the next pandemic. Prog Cardiovasc Dis. 2021;64 :102-104.32278694 35 Leventhal J Angeletti A Cravedi P . EPO in patients with COVID-19: more than an erythropoietic hormone. Am J Kidney Dis. 2020;76 :441. 36 Liu Y Steinacker JM Dehnert C , et al . Effect of “living high-training low” on the cardiac functions at sea level. Int J Sports Med. 1998;19 :380-384.9774204 37 Matsudo S Araújo T Matsudo V , et al . Questionário Internacional De Atividade Física (IPAQ): estupo de Validade e reprodutibilidade no Brasil. Rev Bras Atividade Física Saúde. 2012;6 :5-18. 38 Millet GP Brocherie F Girard O . New metric for the hypoxic stimulus, not for the response. J Appl Physiol. 2016;121 :356.27451276 39 Ministério da Saúde . Protocolos Do Sistema de Vigilância Alimentar e Nutricional - SISVAN Na Assistência à Saúde. 1st ed. (Ministério da Saúde, ed.).; Brasília, Brazil; 2008. 40 Mohamed AA Alawna M . Role of increasing the aerobic capacity on improving the function of immune and respiratory systems in patients with coronavirus (COVID-19): a review. Diabetes Metab Syndr Clin Res Rev. 2020;14 :489-496. 41 Papoti M Vitório R Araújo GG , et al . Determination of force coresponding to maximal lactate steady state in tethered swimming. Int J Exerc Sci. 2009;2 :269-279.27182321 42 Pedersen BK Saltin B . Exercise as medicine - evidence for prescribing exercise as therapy in 26 different chronic diseases. Scand J Med Sci Sport. 2015;25 :1-72. 43 Pramsohler S Burtscher M Faulhaber M , et al . Endurance training in normobaric hypoxia imposes less physical stress for geriatric rehabilitation. Front Physiol. 2017;8 :514.28785224 44 Rodríguez FA Truijens MJ Townsend NE Stray-Gundersen J Gore CJ Levine BD . Performance of runners and swimmers after four weeks of intermittent hypobaric hypoxic exposure plus sea level training. J Appl Physiol. 2007;103 :1523-1535.17690191 45 Ross R Blair SN Arena R , et al . Importance of assessing cardiorespiratory fitness in clinical practice: a case for fitness as a clinical vital sign: a scientific statement from the American Heart Association. Circulation. 2016;134 (24 ):e653-e699. 46 Sahebnasagh A Mojtahedzadeh M Najmeddin F , et al . A perspective on erythropoietin as a potential adjuvant therapy for acute lung injury/acute respiratory distress syndrome in patients with COVID-19. Arch Med Res. 2020;51 :631-635.32863034 47 Sanchis-Gomar F Martinez-Bello VE Domenech E , et al . Effect of intermittent hypoxia on hematological parameters after recombinant human erythropoietin administration. Eur J Appl Physiol. 2009;107 :429-436.19669157 48 Segovia-Juarez J Castagnetto JM Gonzales GF . High altitude reduces infection rate of COVID-19 but not case-fatality rate. Respir Physiol Neurobiol. 2020;281 :103494.32679369 49 Serebrovskaya TV . Intermittent hypoxia research in the former Soviet Union and the commonwealth of independent states: history and review of the concept and selected applications. High Alt Med Biol. 2002;3 :205-221.12162864 50 Sharma AP Saunders PU Garvican-Lewis LA , et al . Normobaric hypoxia reduces VO2 at different intensities in highly trained runners. Med Sci Sports Exerc. 2019;51 :174-182.30095742 51 Simpson RJ Campbell JP Gleeson M , et al . Can exercise affect immune function to increase susceptibility to infection? Exerc Immunol Rev. 2020;26 :8-22.32139352 52 Sohrabi C Alsafi Z O’Neill N , et al . World Health Organization declares global emergency: a review of the 2019 novel coronavirus (COVID-19). Int J Surg. 2020;76 :71-76.32112977 53 Soliz J Schneider-Gasser EM Arias-Reyes C , et al . Coping with hypoxemia: could erythropoietin (EPO) be an adjuvant treatment of COVID-19? Respir Physiol Neurobiol. 2020;279 :103476.32522574 54 Soo J Girard O Ihsan M Fairchild T . The Use of the SpO2 to FiO2 ratio to individualize the hypoxic dose in sport science, exercise, and health settings. Front Physiol. 2020;11 :1-5.32038307 55 Trapé ÁA Camacho-Cardenosa M Camacho-Cardenosa A , et al . Effects of moderate-intensity intermittent hypoxic training on health outcomes of patients recovered from COVID-19: the AEROBICOVID study protocol for a randomized controlled trial. Trials. 2021;22 :534.34384461 56 Van Thienen R Masschelein E Hulst GD Thomis M Kirby TJ Millet GP . Twin resemblance in muscle HIF-1α responses to hypoxia and exercise. Front Physiol. 2017;7 :1-11. 57 Williamson PJ Atkinson G Batterham AM . Inter-individual responses of maximal oxygen uptake to exercise training: a critical review. Sport Med. 2017;47 :1501-1513. 58 Wojan F Stray-Gundersen S Nagel MJ Lalande S . Short exposure to intermittent hypoxia increases erythropoietin levels in healthy individuals. J Appl Physiol (1985). 2021;130 :1955-1960.33955265 59 Yan Z Spaulding HR . Extracellular superoxide dismutase, a molecular transducer of health benefits of exercise. Redox Biol. 2020;32 :101508.32220789 60 Yano T Asano K . Oxygen deficit and anaerobic threshold in the incremental exercise in normoxia and hypoxia. Jpn J Physiol. 1984;34 :1129-1133.6535865 61 Yuki K Fujiogi M Koutsogiannaki S . COVID-19 pathophysiology: a review. Clin Immunol. 2020;215 :108427.32325252
PMC010xxxxxx/PMC10291885.txt	==== Front Anal Chim Acta Anal Chim Acta Analytica Chimica Acta 0003-2670 1873-4324 Published by Elsevier B.V. S0003-2670(23)00786-9 10.1016/j.aca.2023.341565 341565 Article Next-generation diagnostic test for dengue virus detection using an ultrafast plasmonic colorimetric RT-PCR strategy Jiang Kunlun a1 Lee Jung-Hoon b∗∗ Fung To Sing c1 Wu Jingrui d1 Liu Congnuan c Mi Hua a Rajapakse R.P.V. Jayanthe e Balasuriya Udeni B.R. fg Peng Yung-Kang a∗∗∗ Go Yun Young c∗ a Department of Chemistry, College of Science, City University of Hong Kong, Kowloon, 999077, Hong Kong, China b Department of Chemistry, Soonchunhyang University, Asan, 31538, South Korea c Department of Infectious Diseases and Public Health, Jockey Club College of Veterinary Medicine and Life Sciences, City University of Hong Kong, Kowloon, 999077, Hong Kong, China d Department of Chemistry, Seoul National University, Seoul, 08826, South Korea e Department of Veterinary Pathobiology, Faculty of Veterinary Medicine and Animal Science, University of Peradeniya, Peradeniya, 20400, Sri Lanka f Louisiana Animal Disease Diagnostic Laboratory, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA, 70803, USA g Department of Pathobiological Sciences, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA, 70803, USA ∗ Corresponding author. ∗∗ Corresponding author. ∗∗∗ Corresponding author. 1 These authors contributed equally to this work. 26 6 2023 26 6 2023 34156513 4 2023 19 6 2023 25 6 2023 © 2023 Published by Elsevier B.V. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The current global COVID-19 pandemic once again highlighted the urgent need for a simple, cost-effective, and sensitive diagnostic platform that can be rapidly developed for distribution and easy access in resource-limited areas. Here, we present a simple and low-cost plasmonic photothermal (PPT)-reverse transcription-colorimetric polymerase chain reaction (RTcPCR) for molecular diagnosis of dengue virus (DENV) infection. The assay can be completed within 54 min with an estimated detection limit of 1.6 copies/μL of viral nucleic acid. The analytical sensitivity and specificity of PPT-RTcPCR were comparable to that of the reference RT-qPCR assay. Moreover, the clinical performance of PPT-RTcPCR was evaluated and validated using 158 plasma samples collected from patients suspected of dengue infection. The results showed a diagnostic agreement of 97.5% compared to the reference RT-qPCR and demonstrated a clinical sensitivity and specificity of 97.0% and 100%, respectively. The simplicity and reliability of our PPT-RTcPCR strategy suggest it can provide a foundation for developing a field-deployable diagnostic assay for dengue and other infectious diseases. Graphical abstract Image 1 Keywords Biosensor Plasmonic photothermal PCR Photothermal effect Colorimetric Dengue diagnosis Handling Editor: Dr Jing-Juan Xu ==== Body pmc1 Introduction Global attention has been placed on the pandemic of the coronavirus disease 2019 (COVID-19) for the past three years. Most health resources, such as human resources, hospital care, laboratory testing, and public health services, were dedicated to COVID-19, posing a significant threat to the health system of all countries worldwide. In contrast, other illnesses, like dengue fever, suffered a substantial delay in diagnosis and hospital admissions, especially in low-income countries with a more fragile health system [1,2]. Dengue fever is the world's fastest-growing mosquito-borne disease caused by the dengue virus (DENV), an RNA virus belonging to the family Flaviviridae [3]. In the past few decades, dengue infection has spread rapidly within tropical and subtropical countries, placing nearly a third of the global human population at risk [[4], [5], [6], [7], [8], [9]]. Nowadays, it is one of the most significant public health problems in many resource-limited countries, menacing human life in different aspects and causing immense pressure on the medical management system [[10], [11], [12]]. More importantly, the combined impact of the COVID-19 pandemic and dengue epidemics can potentially cause devastating consequences for patients in hyper-endemic regions with multiple serotypes of DENV [9,12]. Without effective vaccines and specific treatment, diagnosing DENV infection in its early stages is critical to improve clinical outcomes and prevent disease transmission [13]. A simple, cost-effective, and sensitive diagnostic test for use in the field is essential for disease control, especially in remote areas with limited access to healthcare providers and facilities. However, diagnostic tools with such capabilities have yet to be readily available. Biosensors are versatile analytical devices that convert biological responses into processable and measurable signals. Since their invention in 1962, biosensors have been extensively researched and developed for their potential applications in medical diagnosis [14]. Traditional serological and virological methods have been integrated into biosensing platforms to improve diagnostic efficiency and accuracy [15]. Immunosensors are well-established biosensors that utilize antibodies as the recognition element and detection of specific antibody-antigen interaction [16]. Although immunoassay techniques, such as enzyme-linked immunosorbent assay (ELISA), are simple and convenient, detecting viral antigens at the early stage of infection is still challenging because of the low level of viral proteins in serum [[17], [18], [19]]. Moreover, interpreting serological results is still challenging due to a broad antigenic cross-reactivity to different DENV serotypes and other flaviviruses [[20], [21], [22]]. Alternatively, nucleic acid amplification technology (NAAT) has been explored as a molecular diagnostic tool for numerous viruses and contributed significantly to healthcare [23,24]. Compared to traditional virus isolation, these methods have achieved more sensitive, accurate, and rapid diagnosis [25]. In particular, NAAT-based reverse transcription polymerase chain reaction (RT-PCR) remains the gold standard for detecting many RNA viruses [[26], [27], [28]]. Owing to its ability to amplify a specific sequence exponentially, RT-PCR is currently the most sensitive and specific detection method of viral RNA [29]. However, most RT-PCR assays must be performed in special laboratories or facilities. Moreover, despite their established efficiencies, developing RT-PCR-based detection methods to point-of-care (POC) applications is hindered by its lengthy detection time, sophisticated instrumentation, and high energy consumption [30,31]. Furthermore, although isothermal nucleic acid amplification technologies can be applied in resource-limited settings, they require complex primer design and may have low sensitivity [32,33]. In recent years, photothermal metallic nanoparticles and films that provide low-energy light-based platforms for temperature control have shown great potential for developing a new generation PCR [[34], [35], [36], [37]]. The light-to-heat conversion by plasmonic nanomaterials, known as the plasmonic photothermal (PPT) effect, can be utilized as an efficient heat source for rapid nucleic acid thermocycling [36,38]. Therefore, portable PPT-based PCR devices have been fabricated by combining the photothermal effect of plasmonic nanostructures to circumvent the critical limitations of conventional PCR, such as the need for sophisticated instrumentation, high energy consumption, and lengthy detection time [33,[39], [40], [41]]. However, these devices still rely on gel electrophoresis or fluorescent signal detection to analyze the PCR amplification products, requiring additional post-processing or expensive equipment [42,43]. As a result, it overshadows the advantages of plasmonic thermocycling and impedes its further development and application in remote locations. Previously, we developed a novel PPT-based PCR coupled with colorimetric assay (PPT-cPCR) for simple, fast amplification and sensitive detection of nucleic acids [44]. Based on the photocatalytic activity of double-stranded DNA-SYBR Green I dye (dsDNA-SGI) complex, 3,3′,5,5′-tetramethylbenzidine (TMB) can be oxidized upon blue light emitting diode (LED) illumination after PPT-based amplification, resulting in the color change of the mixture from colorless to blue color that can be visually distinguished with the naked eye. Based on this platform, a PPT-RTcPCR was developed for DENV detection, and its performance was evaluated. The analytical sensitivity, specificity, and clinical performance of the assay were assessed using RNA extracted from tissue culture fluid (TCF) and human clinical samples (e.g., plasma), respectively, and compared with the reference real-time RT-qPCR recommended by the Centers for Disease Control and Prevention [28]. The results showed that the DENV-specific PPT-RTcPCR provided a simple and sensitive nucleic acid detection platform with a rapid readout at a low cost, which can be significant in resource-limited areas. 2 Materials and methods A detailed description of materials, instruments, synthetic methods, PPT-RTcPCR optical device setup, cells and viruses, clinical specimens, and statistical analysis can be found in the Supplementary Material. 2.1 In vitro transcribed RNA preparation The analytical performance of the DENV-specific PPT-RTcPCR assay was first assessed using in vitro-transcribed (IVT) RNA. Briefly, the partial sequence of the E gene of the DENV-2 NGC strain (nt 1453 to 1550; GenBank accession AF038403) was synthesized and cloned into the pGEM-3Z vector downstream of the T7 promoter sequence. The resulting plasmid pGEM-3Z-DENV was linearized by SalI restriction digestion and subsequently used as the template for in vitro transcription using an mMESSAGE mMACHINE™ T7 Transcription Kit (Invitrogen™, USA). Residual RNA was removed using TURBO Dnase (Invitrogen™, USA). Single-use IVT RNA aliquots were placed at −80 °C. Ten-fold serial dilutions of the IVT RNA (1 to 106 copies/μL) were used to determine the analytical sensitivity of the PPT-RTcPCR. The concentration of the IVT RNA was calculated as described previously [45]. 2.2 Nucleic acid extraction For analytical sensitivity analysis, viral RNA was extracted from TCF containing 1.15 × 106 focus forming unit (FFU)/mL of DENV2 and subjected to 10-fold serial dilutions (101 to 108) in nuclease-free water. For analytical specificity analysis, viral RNA was extracted from TCF samples containing DENV1, DENV4, ZIKV, CVB3, or FIPV. The archived human plasma samples were used for nucleic acid extraction without further dilution. All samples were extracted using a MagMAX™-96 Viral RNA isolation kit (Applied Biosystems, USA) by KingFisher Flex System (Thermo Fisher, USA). After the extraction steps, nucleic acids were eluted with 90 μL nuclease-free water. All nucleic acids were stored at −80 °C for further use. 2.3 DENV-specific PPT-RTcPCR description and optimization The DENV-specific PPT-RTcPCR was developed as described previously with some modifications [28,45]. Target DENV2 viral RNA was first amplified through a PPT-based RT-PCR in the presence of plasmonic photothermal nanoparticles (PMNs). The reaction mixture contained 5 μL of 2 × reaction mix, 0.2 μL of SuperScript™ III RT/Platinum™ Taq Mix, 0.1 μL of 50 μM forward (5′-CAG GCT ATG GCA CYG TCA CGA T-3′) and reverse (5′-CCA TYT GCA GCA RCA CCA TCT C-3′) primers, 2 μL of PMNs with optical density (OD) of 80, 1 μL of viral nucleic acid, 1.6 μL of nuclease-free water, and 20 μL of mineral oil. The thermocycling protocol began at 50 °C for 5 min, followed by 40 cycles between 95 °C (0 s) and 60 °C (8 s). After the PPT-based RT-PCR, 10 μL of the colorimetric solution containing 2-(N-morpholino) ethanesulfonic acid (MES) buffer (8 μL, 0.1 M), 40 mM TMB (1 μL) and 80 × SGI (1 μL), was added to the reaction solution. Using the magnetic properties of PMNs, the particles were collected with a magnet before blue light LED (300 mA, 14 V) irradiation for 2 min. With the photocatalytic activity of the dsDNA-SGI complex, TMB was oxidized under the excitation of blue light LED by adjusting the pH value of the solution to around 5 [44,46]. The absorbance of the resulting mixture was measured by NanoDrop™ OneC Microvolume UV–Vis Spectrophotometer (Thermo Scientific™). The reference real-time DENV2 RT-qPCR was performed as described previously with modifications [28]. Briefly, the reaction mixture contained 5 μL of 2 × reaction mix, 0.2 μL of SuperScript™ III RT/Platinum™ Taq Mix, 0.1 μL of 50 μM forward and reverse primers, 0.18 μL of 10 μM probe (5′- FAM-CTC YCC RAG AAC GGG CCT CGA CTT CAA-BHQ1-3′), 1 μL of viral nucleic acid, and 3.42 μL of nuclease-free water. The RT-qPCR was performed using CFX Opus 96 Real-Time PCR system (Bio-Rad) with the following protocol: 50 °C for 30 min, 95 °C for 2 min, then 40 cycles of 95 °C for 15 s, and 60 °C for 1 min. For the reference real-time RT-PCR, the samples with Ct values larger than 37 would be considered negative since the amplification results were difficult to ascertain. 3 Results and discussions 3.1 Overall workflow from sample to signal The overall workflow of DENV nucleic acid detection using the PPT-RTcPCR platform, from sample collection to target detection, is depicted in Fig. 1 . Firstly, the viral RNA extracted from plasma samples of patients suspected of dengue infection was subjected to DENV-specific PPT-RTcPCR reaction. In this system, PMNs were employed as nano-sized heaters for homogeneous heating. When an infrared (IR)-LED is illuminated, the reaction temperature can be easily and precisely controlled by tuning the light intensity. The heating protocol begins with an isothermal process for the reverse transcription of RNA into complementary DNA (cDNA), followed by two-step thermocycling for cDNA denaturation and specific target extension. Next, amplicons are visually detected using a TMB oxidation-based colorimetric strategy by adding the colorimetric solution containing MES buffer, SGI, and TMB. The amplified dsDNA is the reactant for forming the photocatalyst, triggered only when the SGI is intercalated in the dsDNA. Upon blue light LED irradiation, singlet oxygen is generated by energy transfer from SGI to dissolved oxygen in the solution, which is responsible for the subsequent TMB oxidation. The resulting solution changes from colorless to blue color, making amplicon generation visible. Without viral RNA, there would be no specific RT-PCR amplification, no formation of dsDNA-SGI complex, and thereby no oxidation of TMB so that the solution would remain colorless. Although this is a simple strategy without the requirement of complex instrumentation, there are only a few reports using a combination of the TMB oxidation method with PCR detection, much less applying it for clinical assays. Such a simple design enables our device to be developed into a POC diagnostic system.Fig. 1 Workflow of the PPT-RTcPCR platform. (A) Workflow with a timeline for PPT-RTcPCR platform combining PPT-based RT-PCR with colorimetric detection. (B) TMB-based colorimetric readout. In a negative sample, no specific amplification occurs. Without the formation of the dsDNA-SGI complex, TMB cannot be oxidized, resulting in a minor color change. In a positive sample, amplicons increase exponentially. TMB can be further oxidized by the dsDNA-SGI complex, resulting in a noticeable color change from colorless to blue. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 1 3.2 Characterization of the PPT-RTcPCR platform The PMNs were used as nano-sized heaters since they provide efficient PPT-based thermocycling and convenient magnetic separation. The particle had a magnetic iron oxide core enclosed by a plasmonic Au shell (Fig. 2 A). Iron oxide cores prepared by solvothermal reaction were first functionalized with an amine-terminated surface by organosilane. Then small AuNPs were attached to the core surface via electrostatic assembly, followed by the growth of an Au shell. To increase the stability of PMNs and minimize the nonspecific interaction during amplification, PMNs were finally modified with methoxy-polyethylene glycol (mPEG)-thiol. The elemental mapping images indicated the core was entirely encapsulated by the shell (Fig. 2B). In addition, PMNs showed a strong and broad plasmon wavelength, matching the peak wavelength of IR-LED used in our device (Fig. 2C). Magnetic measurements also confirmed that PMNs had good magnetic properties after shell coating with a magnetic moment of 4.75 emu/g (Fig. 2D).Fig. 2 Characterization and photothermal performance of PMNs. (A) TEM images of PMNs with a core-shell structure. (B) High-angle annular dark-field (HAADF) image and elemental mapping of PMNs (C) Ultraviolet–visible (UV–Vis) spectrum of PMNs. (D) Magnetization curve of PMNs. (E) The temperature profile of the solution with or without PMNs upon illumination. (F) Effect of PMN concentration on heating and cooling ramp rate. The optimal ramp rate was observed at 16 OD PMNs. (G) Representative temperature profile for PPT-based RT-PCR. The thermocycling began at 50 °C (5 min), followed by 40 cycles of 90 °C (0 s) and 60 °C (8 s). (H) Enlarged temperature profile for isothermal reverse transcription. (I) Enlarged temperature profile for PCR amplification. Fig. 2 Using PMNs in the PPT-based optical device (Fig. S1), the temperature of the particle-containing solution increased dramatically under IR-LED illumination, while there was no noticeable temperature change in the solution without PMNs (Fig. 2E). By adjusting the PMN concentration, optimal thermocycling efficiency was achieved with 16 OD PMNs, corresponding to the heating and cooling rates of 8.22 ± 0.24 °C/s and 5.27 ± 0.14 °C/s, respectively (Fig. 2F). In this optimal condition, the temperature profile, including an isothermal process (5 min at 50 °C) and 40 cycles between 90 °C (0 s) and 60 °C (8 s), was completed within 20 min (Fig. 2G, H, and I). Testing with DENV2 target RNA, we confirmed that the influence of PMNs on amplification was negligible by commercial PCR equipment. Our thermocycling protocol produced sufficient amplicons even at low target concentrations in our device (Fig. S2). To increase the signal-to-noise ratio for the subsequent colorimetric detection, the related parameters, such as the pH value, the concentration of SGI and TMB, the LED power, and the irradiation time, were further optimized (Fig. S3). As a result, the colorimetric detection was completed within 4 min, including the time for PMNs collection (<1 min), blue light irradiation (2 min), and signal measurement (<1 min). Taking advantage of ultrafast PPT-based thermocycling and simple colorimetric strategy, the overall assay time was less than 54 min (30 min for commercial RNA extraction, <20 min for thermocycling, and <4 min for signal detection). 3.3 Evaluation of analytical sensitivity and specificity of DENV-specific PPT-RTcPCR The limit of detection (LoD) for PPT-RTcPCR was investigated using serial dilutions of DENV2 IVT RNA and compared to the reference real-time RT-qPCR. The mean OD650 of nine negative samples was 0.39 ± 0.046, resulting in a threshold of OD650 0.53 (three-fold standard deviation of the blank sample) for positivity. As a result, the detection limit was estimated to be approximately 1.9 copies/μL for PPT-RTcPCR, comparable to the reference real-time RT-qPCR (2.1 copies/μL) (Fig. 3 A). The amplification signals generated by PPT-RTcPCR increased as the target concentration increased and reached the plateau when the concentration of RNA was higher than 104 copies/μL, which may be limited by the concentration of SGI and TMB (Fig. 3A). In addition, samples containing 1 copy/μL DENV IVT RNA resulted negative since the amplification signals were lower than the threshold for PPT-RTcPCR and reference RT-qPCR. Consequently, a good linear relationship was obtained when the target concentration ranged from 1 copy/μL to 104 copies/μL with a correlation equation of A = 0.47 + 0.21 log C (where A is the absorbance at 650 nm and C is the target concentration, R2 = 0.99).Fig. 3 Evaluation of analytical sensitivity and specificity of the PPT-RTcPCR platform. Comparison of the quantitative curves between PPT-RTcPCR and RT-qPCR using (A) IVT DENV RNA and (B) RNA extracted from DENV2 spiked serum samples. (C) The absorbance for DENV2 and other nonspecific targets. The results of real-time RT-qPCR are presented with Ct values, which were inversely proportional to the logarithm of copy numbers/μL of DENV2 RNA. Fig. 3 Ten-fold serial dilutions of the viral RNA extracted from normal human serum samples spiked with DENV2 were used to assess the analytical sensitivity in a more clinically relevant setting. As shown in Table S1 and Fig. 3B, PPT-RTcPCR exhibited analytical sensitivity comparable to the reference real-time RT-qPCR. The PPT-RTcPCR detected all samples containing 11.5 FFU/mL of DENV particles as positive, whereas none of the samples containing 1.15 FFU/mL were detected as positive. Thus, a good linearity (R2 = 0.99) was obtained between 1.15 × 104 FFU/mL and 11.5 FFU/mL with the A = 0.44 + 0.19 log C correlation equation, indicating PPT-RTcPCR have similar analytical performance compared to the reference RT-qPCR. The calculated LoD based on DENV2 titers was approximately 3.42 FFU/mL. The analytical specificity of PPT-RTcPCR was evaluated by testing viral RNA extracted from TCF containing DENV1, DENV4, ZIKV (a flavivirus), CVB3 (a picornavirus), and FIPV (a coronavirus). All reactions were negatives with absorbance signals under the threshold, indicating a high analytical specificity of PPT-RTcPCR (Table S2 and Fig. 3C). 3.4 Validation of the clinical performance of DENV-specific PPT-RTcPCR To determine the diagnostic performance of DENV-specific PPT-RTcPCR, a panel of 158 human plasma samples collected from clinically suspected dengue patients was tested and compared with those of the reference real-time RT-qPCR assay that was run side-by-side. Using the predefined absorbance threshold of OD650 0.53, PPT-RTcPCR detected DENV2 positive in 125 out of 131 positively confirmed samples (Table S3), whereas the reference RT-qPCR detected all 131 samples positive. PPT-RTcPCR and the reference RT-qPCR assays determined all 27 negative samples as true negatives (Table S4, Fig. 4 A, and B). Thus, the clinical sensitivity and specificity of PPT-RTcPCR were 95.4% (95% CI, 90.4%–97.9%) and 100% (95% CI, 87.5%–100%), respectively. To rule out the possibility that the threshold value calculated based on the 3σ principle makes for the differences in sensitivity observed between PPT-RTcPCR and the reference RT-qPCR, the optimal threshold value was recalculated using receiver operating characteristic (ROC) analysis. As shown in Fig. 4C, the optimal threshold was 0.51 with an area under the curve (AUC) of 0.99 [47,48]. Note that the ROC optimal threshold is highly correlated with the sample set, and in this study, decreasing the threshold from 0.53 to 0.51 resulted in higher sensitivity. With the ROC optimal threshold, the analytical and clinical performance of PPT-RTcPCR were re-evaluated. The calculated LoD for viral nucleic acid and DENV-spiked human sera were approximately 1.6 copies/μL and 2.69 FFU/mL, respectively. Given these results, the PPT-RTcPCR identified 127 of 131 positive samples as true positives and determined all 27 negatives as true negatives showing the clinical sensitivity and specificity of 97.0% (95% CI, 92.4%–98.8%) and 100% (95% CI, 87.5%–100%), respectively, compared to the reference RT-qPCR (Table 1 ). There were four samples tested as false negatives by PPT-RTcPCR, which can be potentially attributed to the extremely low concentration of target RNA with the colorimetric signal being below the detection limits of the spectrophotometer. The overall agreement value between the PPT-RTcPCR and reference RT-qPCR was 97.5% (95% CI, 83.4%–99.7%; κ = 0.92), indicating that PPT-RTcPCR possessed excellent diagnostic accuracy.Fig. 4 Clinical validation of PPT-RTcPCR platform. (A) Heat map depicting the result of PPT-RTcPCR on 131 positive samples and 27 negative samples. (B) The absorbance of the clinical testing results. The predefined threshold of 0.53 and a ROC optimal threshold of 0.51 were used to classify the samples. (C) ROC curve analysis of the clinical results. The AUC was 0.99. Fig. 4 Table 1 Comparison of the diagnostic performance between DENV2 PPT-RTcPCR and RT-qPCR assays. Table 1 DENV2 RT-qPCR Positive Negative Total Sensitivity (%, 95% CI) Specificity (%, 95% CI) Agreement (%, 95% CI) DENV2 PPT-RTcPCR Positive 127 0 125 97.0 (92.4–98.8) 100 (87.5–100) 97.5 (83.4–99.7) Negative 4 27 33 Total 131 27 158 4 Conclusion In the fight against emerging infectious diseases, developing a cost-effective platform and validating its use in clinical settings is urgently demanded. In this study, we developed a PPT-RTcPCR to detect DENV in clinical samples based on our previous system and validated its diagnostic performance. Utilizing the ultrafast PPT-based thermocycling combined with the colorimetric assay, the PPT-RTcPCR platform can complete the detection of DENV2 RNA within 54 min exhibiting high sensitivity with the LoD of 1.6 copies/μL for viral nucleic acid and 2.69 FFU/mL for DENV-spiked in human serum, comparable with the reference RT-qPCR. Validating the DENV-specific PPT-RTcPCR assay using 158 clinical plasma samples demonstrated excellent diagnostic performance with clinical sensitivity and specificity of 97.0% and 100%, respectively, and 97.5% agreement with the reference RT-qPCR. Compared to most reported diagnostic platforms (Table S5), PPT-RTcPCR is more straightforward and cost-effective, eliminating the need for complicated design or sophisticated equipment unsuitable for POC diagnostic assay while maintaining ultrahigh sensitivity comparable to real-time RT-qPCR. Although current PPT-RTcPCR involves several operations, such as adding the colorimetric solution after amplification, which makes the workflow relatively complicated, this issue could be addressed using microfluidic chips and automated systems. Moreover, taking advantage of the PPT and magnetic properties, PMNs have the potential to be applied for virus lysis and nucleic acid extraction and purification. Thus, integrating nucleic acid extraction into PPT-cPCR or PPT-RTcPCR can significantly facilitate on-site detection. Taken together, the results obtained from this study will serve basis for the development of a field-deployable diagnostic assay along with an all-in-one integrated device containing extraction, amplification, and detection capabilities for sensitive on-site nucleic acid detection not only for DENV but also for other important infectious diseases such as influenza viruses and high-pathogenic COVID-19. CRediT authorship contribution statement Kunlun Jiang: Methodology, Investigation, Validation, Formal analysis, Writing – original draft. Jung-Hoon Lee: Conceptualization, Resources, Writing – review & editing, Supervision, Project administration, Funding acquisition. To Sing Fung: Investigation, Methodology, Writing – review & editing. Jingrui Wu: Software. Congnuan Liu: Investigation. Hua Mi: Investigation. R.P.V. Jayanthe Rajapakse: Resources. Udeni B.R. Balasuriya: Resources. Yung-Kang Peng: Resources, Supervision. Yun Young Go: Resources, Supervision, Project administration, Writing – review & editing, Funding acquisition. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Supplementary data The following is the Supplementary data to this article:Multimedia component 1 Multimedia component 1 Data availability Data will be made available on request. Acknowledgments This study was funded by a grant from the Health and Medical Research Fund (10.13039/501100005847 HMRF 20190862) of the Hong Kong Government Food and Health Bureau, Hong Kong SAR, China. This work was also supported by the Soonchunhyang University Research Fund and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2022R1F1A1070162) and 10.13039/501100003716 Korea Basic Science Institute (National Research Facilities and Equipment Center Grant funded by the Ministry of Education) (No. 2022R1A6C101B794). Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.aca.2023.341565. ==== Refs References 1 Kiss P. Carcel C. Hockham C. Peters S.A. The impact of the COVID-19 pandemic on the care and management of patients with acute cardiovascular disease: a systematic review Eur. Heart J. Qual. Care Clin. Outcomes 7 2021 18 27 33151274 2 Lazarus J.V. Romero D. Kopka C.J. Karim S.A. Abu-Raddad L.J. Almeida G. A multinational Delphi consensus to end the COVID-19 public health threat Nature 2022 1 14 3 Henchal E.A. Putnak J.R. The dengue viruses Clin. Microbiol. Rev. 3 1990 376 396 2224837 4 WHO Global Strategy for Dengue Prevention and Control 2012-2020 2012 5 Bhatt S. Gething P.W. Brady O.J. Messina J.P. Farlow A.W. Moyes C.L. The global distribution and burden of dengue Nature 496 2013 504 507 23563266 6 Guzman M.G. Harris E. Dengue, Lancet 385 2015 453 465 25230594 7 Zhou Y. Wan Z. Yang S. Li Y. Li M. Wang B. A mismatch-tolerant reverse transcription loop-mediated isothermal amplification method and its application on simultaneous detection of all four serotype of dengue viruses Front. Microbiol. 10 2019 1056 31139171 8 Halstead S.B. Dengue, Lancet 370 2007 1644 1652 17993365 9 WHO Dengue and Severe Dengue 2021 10 Duong V. Lambrechts L. Paul R.E. Ly S. Lay R.S. Long K.C. Asymptomatic humans transmit dengue virus to mosquitoes Proc. Natl. Acad. Sci. USA 112 2015 14688 14693 26553981 11 Wilder-Smith A. Ooi E.-E. Horstick O. Wills B. Dengue, Lancet 393 2019 350 363 30696575 12 Eivazzadeh-Keihan R. Pashazadeh-Panahi P. Mahmoudi T. Chenab K.K. Baradaran B. Hashemzaei M. Dengue virus: a review on advances in detection and trends–from conventional methods to novel biosensors Microchim. Acta 186 2019 1 24 13 Teixeira M.G. Siqueira J.B. Jr. Ferreira G.L. Bricks L. Joint G. Epidemiological trends of dengue disease in Brazil (2000–2010): a systematic literature search and analysis PLoS Neglected Trop. Dis. 7 2013 e2520 14 Clark L.C. Jr. Lyons C. Electrode systems for continuous monitoring in cardiovascular surgery Ann. N. Y. Acad. Sci. 102 1962 29 45 14021529 15 Pashchenko O. Shelby T. Banerjee T. Santra S. A comparison of optical, electrochemical, magnetic, and colorimetric point-of-care biosensors for infectious disease diagnosis ACS Infect. Dis. 4 2018 1162 1178 29860830 16 Aydin E.B. Aydin M. Sezgintürk M.K. Advances in electrochemical immunosensors Adv. Clin. Chem. 92 2019 1 57 31472751 17 Yap G. Sil B.K. Ng L.-C. Use of saliva for early dengue diagnosis PLoS Neglected Trop. Dis. 5 2011 e1046 18 Guzman M.G. Alvarez A. Vazquez S. Alvarez M. Rosario D. Pelaez O. Epidemiological studies on dengue virus type 3 in Playa municipality, Havana, Cuba, 2001–2002 Int. J. Infect. Dis. 16 2012 e198 e203 22277259 19 Lee Y. Choi J. Han H.-K. Park S. Park S.Y. Park C. Fabrication of ultrasensitive electrochemical biosensor for dengue fever viral RNA Based on CRISPR/Cpf1 reaction Sensor. Actuator. B Chem. 326 2021 128677 20 Mackay I.M. Real-time PCR in the microbiology laboratory Clin. Microbiol. Infect. 10 2004 190 212 15008940 21 Stettler K. Beltramello M. Espinosa D.A. Graham V. Cassotta A. Bianchi S. Specificity, cross-reactivity, and function of antibodies elicited by Zika virus infection Science 353 2016 823 826 27417494 22 Rathore A.P. St John A.L. Cross-reactive immunity among flaviviruses Front. Immunol. 11 2020 334 32174923 23 Mullis K.B. Faloona F.A. [21] Specific synthesis of DNA in vitro via a polymerase-catalyzed chain reaction Methods Enzymol. 155 1987 335 350 3431465 24 Sano T. Smith C.L. Cantor C.R. Immuno-PCR: very sensitive antigen detection by means of specific antibody-DNA conjugates Science 258 1992 120 122 1439758 25 Obande G.A. Singh K.K.B. Current and future perspectives on isothermal nucleic acid amplification technologies for diagnosing infections Infect. Drug Resist. 13 2020 455 32104017 26 Huggett J. Dheda K. Bustin S. Zumla A. Real-time RT-PCR normalisation; strategies and considerations Gene Immun. 6 2005 279 284 27 Nolan T. Hands R.E. Bustin S.A. Quantification of mRNA using real-time RT-PCR Nat. Protoc. 1 2006 1559 1582 17406449 28 Santiago G.A. Vergne E. Quiles Y. Cosme J. Vazquez J. Medina J.F. Analytical and clinical performance of the CDC real time RT-PCR assay for detection and typing of dengue virus PLoS Neglected Trop. Dis. 7 2013 e2311 29 Qiu G. Gai Z. Tao Y. Schmitt J. Kullak-Ublick G.A. Wang J. Dual-functional plasmonic photothermal biosensors for highly accurate severe acute respiratory syndrome coronavirus 2 detection ACS Nano 14 2020 5268 5277 32281785 30 Yang S. Rothman R.E. PCR-based diagnostics for infectious diseases: uses, limitations, and future applications in acute-care settings Lancet Infect. Dis. 4 2004 337 348 15172342 31 Petralia S. Conoci S. PCR technologies for point of care testing: progress and perspectives ACS Sens. 2 2017 876 891 28750519 32 De Paz H.D. Brotons P. Muñoz-Almagro C. Molecular isothermal techniques for combating infectious diseases: towards low-cost point-of-care diagnostics Expert Rev. Mol. Diagn. 14 2014 827 843 25052202 33 Cheong J. Yu H. Lee C.Y. Lee J.-U. Choi H.-J. Lee J.-H. Fast detection of SARS-CoV-2 RNA via the integration of plasmonic thermocycling and fluorescence detection in a portable device Nat. Biomed. Eng. 2020 1 9 31937945 34 Kim M. Lee J.H. Nam J.M. Plasmonic photothermal nanoparticles for biomedical applications Adv. Sci. 6 2019 1900471 35 Roche P.J. Beitel L.K. Khan R. Lumbroso R. Najih M. Cheung M.C.-K. Demonstration of a plasmonic thermocycler for the amplification of human androgen receptor DNA Analyst 137 2012 4475 4481 22871797 36 Son J.H. Cho B. Hong S. Lee S.H. Hoxha O. Haack A.J. Ultrafast photonic PCR Light Sci. Appl. 4 2015 e280 e 37 Lee S.H. Park S.-m. Kim B.N. Kwon O.S. Rho W.-Y. Jun B.-H. Emerging ultrafast nucleic acid amplification technologies for next-generation molecular diagnostics Biosens. Bioelectron. 141 2019 111448 38 You M. Li Z. Feng S. Gao B. Yao C. Hu J. Ultrafast photonic PCR based on photothermal nanomaterials Trends Biotechnol. 38 2020 637 649 31918858 39 Lee J.-H. Cheglakov Z. Yi J. Cronin T.M. Gibson K.J. Tian B. Plasmonic photothermal gold bipyramid nanoreactors for ultrafast real-time bioassays J. Am. Chem. Soc. 139 2017 8054 8057 28457135 40 Almassian D.R. Cockrell L.M. Nelson W.M. Portable nucleic acid thermocyclers Chem. Soc. Rev. 42 2013 8769 8798 24030680 41 Brongersma M.L. Halas N.J. Nordlander P. Plasmon-induced hot carrier science and technology Nat. Nanotechnol. 10 2015 25 34 25559968 42 Lee Y. Kang B.-H. Kang M. Chung D.R. Yi G.-S. Lee L.P. Nanoplasmonic on-chip PCR for rapid precision molecular diagnostics ACS Appl. Mater. Interfaces 12 2020 12533 12540 32101396 43 Wu J. Jiang K. Mi H. Qiu Y. Son J. Park H.J. A rapid and sensitive fluorescence biosensor based on plasmonic PCR Nanoscale 13 2021 7348 7354 33889912 44 Jiang K. Wu J. Qiu Y. Go Y.Y. Ban K. Park H.J. Plasmonic colorimetric PCR for Rapid molecular diagnostic assays Sensor. Actuator. B Chem. 337 2021 129762 45 Go Y.Y. Rajapakse R.P.V.J. Kularatne S.A.M. Lee P.Y.A. Ku K.B. Nam S. A pan-dengue virus reverse transcription-insulated isothermal PCR assay intended for point-of-need diagnosis of dengue virus infection by use of the POCKIT nucleic acid analyzer J. Clin. Microbiol. 54 2016 1528 1535 27030492 46 Zhang X. Huang C. Xu S. Chen J. Zeng Y. Wu P. Photocatalytic oxidation of TMB with the double stranded DNA–SYBR Green I complex for label-free and universal colorimetric bioassay Chem. Commun. 51 2015 14465 14468 47 Zweig M.H. Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine Clin. Chem. 39 1993 561 577 8472349 48 Bewick V. Cheek L. Ball J. Statistics review 13: receiver operating characteristic curves Crit. Care 8 2004 1 5 14975035
PMC010xxxxxx/PMC10291891.txt	==== Front J Allergy Clin Immunol Pract J Allergy Clin Immunol Pract The Journal of Allergy and Clinical Immunology. in Practice 2213-2198 2213-2201 Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology S2213-2198(23)00696-7 10.1016/j.jaip.2023.06.031 Original Article Endotyping of IgE-mediated polyethylene glycol and/or polysorbate 80 allergy. Ieven Toon MD 12 Coorevits Lieve MLT 12 Vandebotermet Martijn MD 23 Tuyls Sebastiaan MD 24 Vanneste Helene MD 25 Santy Lisa MD 26 Wets Dries MD 2 Proost Paul PhD 7 Frans Glynis PhD 8 Devolder David PharmD 9 Breynaert Christine MD, PhD 12 Bullens Dominique M.A. MD, PhD 110 Schrijvers Rik MD, PhD 12∗ 1 KU Leuven Department of Microbiology, Immunology and Transplantation, Allergy and Clinical Immunology Research Group, KU Leuven, Leuven, Belgium 2 Department of General Internal Medicine, Division of Allergy and Clinical Immunology, University Hospitals Leuven, Leuven, Belgium 3 Department of Pulmonology, AZ Groeninge Hospital, Kortrijk, Belgium 4 Department of Pulmonology, GZA St-Augustinus Hospital, Wilrijk, Belgium 5 Department of Pulmonology, AZ Vesalius, Tongeren, Belgium 6 Department of Internal Medicine, Division of Pulmonology, St-Jozefskliniek, Izegem, Belgium 7 KU Leuven Department of Microbiology, Immunology and Transplantation, Laboratory of Molecular Immunology, Rega Institute for Medical Research, KU Leuven, Leuven, Belgium 8 Clinical Department of Laboratory Medicine, University Hospitals Leuven, Leuven, Belgium 9 Pharmacy Department, University Hospitals Leuven, Leuven, Belgium 10 Department of Pediatrics, University Hospitals Leuven, Leuven, Belgium ∗ Corresponding author: Name: Rik Schrijvers, MD, PhD Adress: Prof. Dr. Rik Schrijvers Department of General Internal Medicine University Hospitals Leuven Herestraat 49, B-3000 Leuven Belgium Telephone: +32 16 34 38 05 26 6 2023 26 6 2023 17 6 2022 7 6 2023 13 6 2023 © 2023 Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology. 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Polyethylene glycol (PEG) and polysorbate 80 (PS80) allergy preclude from SARS-CoV-2 vaccination. The mechanism(s) governing cross-reactivity and PEG molecular weight-dependency remain unclear. Objectives To evaluate PEGylated lipid nanoparticle (LNP) vaccine (BNT162b2) tolerance, and explore the mechanism of reactivity in PEG and/or PS80 allergic patients. Methods PEG/PS80 dual- (n=3), PEG mono- (n=7) and PS80 mono-allergic patients (n=2) were included. Tolerability of graded vaccine challenges was assessed. Basophil activation testing on whole blood (wb-BAT) or passively sensitized donor basophils (allo-BAT) was performed using PEG, PS80, BNT162b2, and PEGylated lipids (ALC-0159). Serum PEG-specific IgE was measured in patients (n=10) and controls (n=15). Results Graded BNT162b2 challenge in dual- and PEG mono-allergic patients (n=3/group) was well-tolerated and induced anti-S IgG seroconversion. PS80 mono-allergic patients (n=2/2) tolerated single-dose BNT162b2 vaccination. Wb-BAT reactivity to PEG-containing antigens was observed in dual- (n=3/3) and PEG mono- (n=2/3), but absent in PS80 mono-allergic patients (n=0/2). BNT162b2 elicited the highest in vitro reactivity. BNT162b2 reactivity was IgE-mediated, complement-independent, and inhibited in allo-BAT by preincubation with short PEG motifs, or detergent-induced LNP degradation. PEG-specific IgE was only detectable in dual-allergic (n=3/3) and PEG mono-allergic (n=1/6) serum. Conclusion PEG and PS80 cross-reactivity is determined by IgE recognizing short PEG motifs, whilst PS80 mono-allergy is PEG-independent. PS80 skin test positivity in PEG allergics was associated with a severe and persistent phenotype, higher serum PEG-specific IgE levels and enhanced BAT reactivity. Spherical PEG-exposure via LNP enhances BAT sensitivity through increased avidity. All PEG and/or PS80 excipient allergic patients can safely receive SARS-CoV-2 vaccines. Graphical abstract Keywords basophil activation test polyethylene glycol polysorbate 80 BNT162b2 ALC-0159 SARS-CoV-2 vaccine allergy IgE cross-reactivity ==== Body pmcFunding sources: DB and RS are supported by the Fonds Wetenschappelijk Onderzoek - Vlaanderen National Fund for Scientific Research (FWO) senior clinical investigator fellowship (1805518N, 1805523N; and 1801019N). The study was supported by UZ Leuven KOOR funding (S60734). The funder(s) had no role in the design and conduct of the study, collection, management, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Conflict of interest statement: All authors declare that they have no conflicts of interest to disclose. Abbreviations: AC: anaphylaxis control; BAT: basophil activation testing; CARPA: complement activation-related pseudoallergy; cd-BAT: complement-deprived basophil activation test; CDC: Centers for Disease Control and Prevention; DA: dual-allergy; DEG: diethylene glycol; EG: ethylene glycol; GVC: graded vaccine challenge; HC: healthy control; HDM: house dust mite; HMW: high molecular weight; LMW: low molecular weight; LNP: lipid nanoparticle; MW: molecular weight; PEG: polyethylene glycol; PS80: polysorbate 80; SEM: standard error of mean; sIgE: specific IgE; ST: skin test; tIgE: total IgE; wb-BAT: whole blood basophil activation test
PMC010xxxxxx/PMC10292026.txt	==== Front Geoforum Geoforum Geoforum; Journal of Physical, Human, and Regional Geosciences 0016-7185 1872-9398 Elsevier Ltd. S0016-7185(23)00142-2 10.1016/j.geoforum.2023.103816 103816 Forum Using wastewater to overcome health disparities among rural residents Holm Rochelle H. a Pocock Gina b Severson Marie A. c Huber Victor C. c Smith Ted a McFadden Lisa M. c⁎ a Christina Lee Brown Envirome Institute, School of Medicine, University of Louisville, 302 E. Muhammad Ali Blvd., Louisville, KY 40202, United States b Waterlab, 23B De Havilland Crescent, 0020 Persequor Technopark, South Africa c Division of Basic Biomedical Sciences, University of South Dakota, 414 E. Clark St., Vermillion, SD 57069, United States ⁎ Corresponding author at: University of South Dakota, 414 E. Clark St., Vermillion, SD 57069, United States. 26 6 2023 8 2023 26 6 2023 144 103816103816 8 2 2023 7 6 2023 16 6 2023 © 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The SARS-CoV-2 pandemic highlighted the need for novel tools to promote health equity. There has been a historical legacy around the location and allocation of public facilities (such as health care) focused on efficiency, which is not attainable in rural, low-density, United States areas. Differences in the spread of the disease and outcomes of infections have been observed between urban and rural populations throughout the COVID-19 pandemic. The purpose of this article was to review rural health disparities related to the SARS-CoV-2 pandemic while using evidence to support wastewater surveillance as a potentially innovative tool to address these disparities more widely. The successful implementation of wastewater surveillance in resource-limited settings in South Africa demonstrates the ability to monitor disease in underserved areas. A better surveillance model of disease detection among rural residents will overcome issues around the interactions of a disease and social determinants of health. Wastewater surveillance can be used to promote health equity, particularly in rural and resource-limited areas, and has the potential to identify future global outbreaks of endemic and pandemic viruses. Keywords Africa Disease surveillance Health disparities SARS-CoV-2 Vulnerable communities Wastewater surveillance ==== Body pmc1 Introduction Health disparities among rural United States residents have been well documented (Summers-Gabr, 2020, Ashburn et al., 2021, Cross et al., 2021, Jackson et al., 2021, Dani et al., 2022). Many factors contribute to this, including poorer social determinants of health metrics such as poverty, lack of education, and life expectancy (Table 1 ; Rural Health Information Hub, 2022). Moreover, rural populations have critical barriers to healthcare; there are approximately three M.D.’s in metropolitan areas for every one in non-metropolitan locations on a per capita basis. There is also a historical legacy around the location and allocation of public facilities (such as health care), with a focus on spatial extent and coverage of human services to promote efficiency (DeVerteuil, 2000), a model which leaves behind rural, low-density areas. Further, rural areas are more likely to be served by critical access hospitals, which account for over 50% of hospitals in predominately rural states such as South Dakota. These hospitals, by federal regulations: 1) must be in a rural county at least 35 miles from another acute care facility; 2) may at most have 25 beds; 3) average an annual inpatient length of stay less than 96 h; and 4) have an emergency room or department with 24-hour availability (Centers for Medicare and Medicaid Services, 2021). These hospitals are designed to provide primary care to patients for common health emergencies. However, they often transfer patients to larger facilities for more complex or serious cases. This is important for public health surveillance activities, as federal regulations limit the role of these hospitals in treating large outbreaks of diseases in rural locations, given the limited number of beds, hospital stay capacity, and reduced medical resources. Although these hospitals provide a critical lifeline for basic care for many rural residents, frontier residents must still travel long distances, some of whom must travel over 80 miles each way to receive care (South Dakota Department of Health, 2020). In Kentucky, even in rural areas with geographical access to care, this travel could also include the mountainous Appalachian region, which is prone to flooding (Kentucky Cabinet for Health and Family Services, 2022). As access to equitable healthcare remains a constant struggle in rural and frontier locations, innovation in public health surveillance approaches is needed to maintain the health of residents and capacity at healthcare facilities (Fig. 1 ).Table 1 Rural health disparities in select United States locations. Disparity Measure Location Metro Non-Metro Access to Healthcare M.D.'s per 10,000 People SD 36 11 KY 32 11 National 32 11 Social Determinates of Health Poverty SD 8.6% 14.4% KY 12.1% 19.1% National 11.5% 14.4% Population Without a High School Diploma SD 6.4% 9.1% KY 9.8% 17.2% National 11.2% 12.9% Outcomes Life Expectancy at Birth (years) SD 80.3 78.9 KY 77.3 75 National 79.3 77.4 Fig. 1 Framework for adapting rural wastewater surveillance to overcome health disparities. 2 SARS-CoV-2 Rural Health Disparities 2.1 Cases, Hospitalizations, and Deaths Health disparities have been devastating for rural residents during this pandemic, with access to healthcare being one of the social determinants of health that influenced disease outcomes in rural locations. Testing and disease surveillance was limited in rural populations, making it difficult to track the spread of the disease through cases, hospitalizations, and deaths (Chillag and Lee, 2020). Although the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic started in urban centers, the overall picture is that non-metropolitan residents have had a higher rate of deaths per capita than metropolitan residents (Cuadros et al., 2021, Dobis et al., 2021). Infection-induced seroprevalence among blood donors was lower in rural populations during the first few months of the pandemic but rapidly and consistently exceeded that of urban populations by the fall of 2020 (Li et al., 2022). Lower rates of vaccine-induced seroprevalence were also found among rural blood donors than among urban donors (Li et al., 2022), indicating an increased risk of infection in rural populations. Even within large counties, from August 2020 to February 2021, the overall positive seroprevalence for SARS-CoV-2 spike protein-specific IgG antibodies was lower in the urban core than in the lower-density county edges, indicating that the scale of the geographic zone matters (Smith et al., 2022). While cases and deaths were higher among the racially diverse urban poor during the first few months of the pandemic, they transitioned to rural areas, especially those with higher levels of poverty (Li et al., 2021). Barriers to testing remained a limitation to understanding the rate of infection reported by health facilities in rural populations throughout the pandemic. Rader and colleagues reported greater travel times to testing centers in rural counties (Rader et al., 2020). Given these barriers to testing, other metrics should be considered as indicators of infectious disease surveillance in rural populations. The risk of serious illness is higher among rural populations; accordingly, the hospitalization of coronavirus disease 2019 (COVID-19) cases in rural populations was estimated to be 10% higher than in urban populations with similar levels of exposure (Kaufman et al., 2020). This increased risk is due to preexisting comorbidities and older age in rural populations (Kaufman et al., 2020). A large longitudinal study of health records found the adjusted odds ratios for hospitalization to be 1.18 and 1.29 times higher among urban-adjacent rural and non-urban-adjacent rural populations, respectively, compared to urban populations (Anzalone et al., 2023). Rural patients are also more likely to require invasive mechanical ventilation, vasopressor support, and extracorporeal membrane oxygenation than urban patients (Anzalone et al., 2023). Finally, the odds of death or transfer to hospice care were approximately 1.36 times higher in rural patients than in urban patients (Anzalone et al., 2023). Similarly, the cumulative death rate per capita from COVID-19 was nearly 50% higher in rural populations (446.12 deaths per 100,000 people) compared to large central metropolitan areas (302.25 deaths per 100,000 people) (CDC, 2022a). During this pandemic, poorer health outcomes were observed in rural patients than in their urban counterparts. Many factors contribute to these outcomes, including vaccination rates, comorbidities, and a reduction in community mitigation measures (Jackson et al., 2021). 2.2 Vaccination Vaccination rates against COVID-19 differ between urban and rural populations. In a study on Arkansas residents, rural respondents reported lower trust in vaccines than urban populations (McElfish et al., 2021). In a study on Tennessee populations, access to testing and vaccination was greater in urban areas than in rural areas (Alcendor, 2021). Furthermore, Tennessee non-metropolitan residents were significantly more likely to resist vaccination (Gatwood et al., 2021). A national survey found that non-metropolitan populations were less likely to report intending to receive the COVID-19 vaccine (Salmon et al., 2021). Rural Oklahoma residents reported hesitation toward getting vaccinated as well, with skepticism and limited knowledge about the vaccine being commonly cited reasons (Hubach et al., 2022). Current vaccination rates suggest that these hesitations may have reduced the population receiving the vaccine. According to the Centers for Disease Control (2022b), in large central metropolitan areas, 70.3% of the population completed their primary COVID-19 vaccination series versus 50.1% of rural populations. The reduction in vaccine recipients among rural populations may have contributed to an increase in both mortality and case fatality rates in rural populations during the Omicron wave (Cuadros et al., 2022). Unvaccinated populations are significantly more likely to require hospitalization for COVID-19 than vaccinated populations (Winkelman et al., 2022). Lower vaccination rates in rural populations may contribute to greater vulnerability toward severe outcomes. 2.3 Access to treatment Early experimental treatments and clinical trials were conducted primarily in a few urban centers (Sharma et al., 2020, Dandachi et al., 2021). Rural populations, on average, would need to travel 85.2 min to the nearest clinical trial site compared to 18.7 min for urban populations (Khazanchi et al., 2021). Furthermore, rural healthcare providers tend to have limited resources, such as intensive care unit beds, ventilators, and key staff to support patient care (Lakhani et al., 2020). In many cases, rural areas served by critical-access hospitals must transfer patients to urban hospitals for more complex care (Diaz et al., 2020, Henning-Smith, 2020). Visitor restrictions observed early in the pandemic, along with the need to travel away from their community of support, may have led rural patients to hesitate in seeking care (Henning-Smith, 2020). Recent studies have reported that rural patients had to travel a median time of 69.6 min to a test and treatment site compared to urban populations, where the median travel time was 11 min (Khazanchi et al., 2022). Since these sites provide access to treatments such as nirmatrelvir-ritonavir and molnupiravir, which need to be initiated within 5 days of symptom onset for maximum efficacy, long travel distances to treatment centers may be a barrier for patients residing in rural areas. This is especially problematic for individuals with limited means of transportation. Overall, barriers to healthcare, including access to testing that allows timely treatment, are likely to contribute to the observed rural health disparities. 2.4 Linking a medical model of disease and contagion with a social determinants of health model A better surveillance model of health must still bump up against a lack of faith in the public health system among rural inhabitants, especially in terms of prevention. Rural residents are less likely to obtain recommended preventive healthcare services (Casey et al., 2001). Geospatial analyses on healthcare capacity have determined that local geography exacerbated the deployment of COVID-19 vaccines as far as access of residents to rural clinics, termed vaccination coldspots (Cuadros et al., 2023). The lack of faith in the public health system was also seen in an assessment of tweets for topics related to COVID-19 prevention (i.e., “vaccine” and “mandates”), which found rural social media users had a stronger negative sentiment than urban users (Liu et al., 2023). Linking a medical model of disease and contagion with a social determinants of health model is equally needed to overcome health disparities among rural residents. Many studies have noted that rural populations with weaker social determinants of health have poorer outcomes. Rural minority populations are also more likely to experience poor COVID-19 outcomes (Henning-Smith et al., 2021, Wheeler et al., 2021). In North Carolina, the test positivity rate was the highest among minority populations living in rural areas (Brandt et al., 2021), and similar findings have been documented among rural and minority populations in Indiana (Dixon et al., 2021). A greater number of comorbid diseases was observed among hospitalized rural North Carolina patients early in the pandemic (Denslow et al., 2022), with similarly increased rates of comorbid diseases in rural populations in Georgia associated with higher COVID-19 mortality (Shah et al., 2020). Further complicating the clinical outcomes observed, rural populations are more likely to be older than urban populations (Chillag and Lee, 2020), which naturally puts these populations at a higher risk for severe COVID-19 disease. Persistent poverty is more common in rural settings than in urban settings, especially among minority populations (Chillag and Lee, 2020). Employers in the agricultural and food-processing industries are often in rural locations, working conditions that can foster the spread of the disease (Behrman et al., 2021). Finally, a combination of increased misinformation, lack of trust, and reduced health literacy in rural settings may further contribute to the barriers observed in rural populations (Lakhani et al., 2020). Trust in public health messages and the perception of a reduced risk of disease have been noted among rural populations (Ridenhour et al., 2022). These social factors might increase the probability of severe outcomes in rural populations. Underpinning the SARS-CoV-2 rural health disparities is the larger issue of 'access' and density. Rural areas lack the density of both population and of health services - this is not surprising, but it does raise important issues around the location and allocation of public facilities (such as health care) and equity. DeVerteuil (2000) reviews the literature on access, density, and equity when relating to public facilities. This comes to attention where in rural areas, away from urban centers, efficiency according to facility distance, pattern, accessibility, impacts, and externalities are more sensitive to facility spacing. This is a case where wastewater surveillance may overcome health disparities among rural residents to circumvent a human services location legacy model. 3 Wastewater surveillance to improve rural health Despite this, there is an opportunity for innovation in identifying the virus in rural communities as a step toward alleviating these health disparities. The current pandemic has enabled the development and utilization of temporary mobile clinical testing units and traveling healthcare professionals to aid rural communities experiencing surges in illness. However, early warning health surveillance systems are necessary to allow states to efficiently mobilize resources and healthcare workers to rural populations during health crises. The lack of effective surveillance in rural areas leads to an increased probability of the spread of COVID-19 among rural communities, which may lead to the spread of the virus to the surrounding urban areas (Souch and Cossman, 2021). Wastewater surveillance is a tool that can be used to provide surveillance in rural and remote locations (Medina et al., 2022). This surveillance has been well-received by the public for its use in tracking diseases, environmental toxins, and terrorist threats when large populations, such as those served by wastewater treatment plants, are included (La Joie et al., 2022). However, complicating surveillance, many rural populations use on-site sanitation systems rather than piped sewer connections. Approximately 15% of United States households are served by septic systems (World Health Organization, 2021). While this is true, rural populations often travel to adjacent larger communities for school, to work, and to shop, allowing regional wastewater treatment plants to still offer an opportunity for inclusive sampling. Working with local experts to understand travel patterns among rural populations is key to establishing surveillance systems that capture rural populations that are otherwise off the grid. Currently, the National Wastewater Surveillance System excludes communities with populations of less than 3,000 (Centers for Disease Control, 2022c). In situations where travel to regional urban centers exceeding 3,000 people is not well documented, surveillance efforts can be uniquely aligned to identify the relative risk of the disease in these rural populations. Wastewater or non-sewered sanitation system surveillance is a flexible form of surveillance with the potential to provide information on many diseases circulating in a community with relatively few samples being collected. In the absence of wastewater treatment, septic tank emptying operations or building or factory-level plumbing access points can be sampled as representative of the community. Since the launch of widespread COVID-19 sanitation system surveillance in the United States, some jurisdictions have expanded their surveillance to include emerging diseases such as monkeypox, influenza, and polio (Ryerson et al., 2022, Mercier et al., 2022, Nelson, 2022, Tanne, 2022). Methods have been developed to include common circulating diseases, such as human adenovirus, measles, and norovirus (Kevill et al., 2022). Importantly, to evaluate many of these diseases, additional samples are not needed and can be assessed from the same sample volume collected to monitor SARS-CoV-2. This provides cost efficiency that cannot be matched using traditional clinical surveillance techniques. It is estimated that clinical surveillance costs approximately ten times more than wastewater surveillance (Manuel et al., 2022), and the cost savings are, in part, because wastewater represents a greater proportion of the population than clinical testing. Wastewater samples represent the pooled contribution of the community yet still provide the individual privacy of each community member. As well, wastewater surveillance may provide information on both symptomatic and asymptomatic populations (Cavany et al., 2022), including vaccinated individuals who may shed the virus during low-level infections. Furthermore, the samples of treatment-seeking and non-treatment-seeking infected individuals includes those who used at-home testing to make decisions regarding current and future healthcare visits. Given the barriers to healthcare and testing faced by rural patients, ensuring that the health surveillance system does not depend on clinical testing capacity is critical. Significantly, wastewater samples are collected by researchers, wastewater utility workers, or sanitation workers, and analyzed by science, technology, engineering, and mathematics (STEM) professionals, which avoids overtaxing healthcare workers in rural locations. This health surveillance system can then be linked to continuously (and within days) notify healthcare providers to prepare for an influx of sick patients, including requests to mobilize testing and increase the number of traveling healthcare workers in the region. As rural locations comprise approximately 97% of the land in the United States (Ratcliffe et al., 2016), finding surveillance methods that involve these widespread communities is important. Creating a more inclusive disease surveillance system increases the chances for early detection of emerging diseases and outbreaks. This allows mitigation and prevention efforts to be implemented, thereby reducing the probability of an outbreak becoming an epidemic or pandemic. A truly robust framework should also consider opportunities to connect this surveillance with conventional public health surveillance mechanisms. For example, seasonal influenza will most likely be detected in wastewater before conventional syndromic surveillance using electronic health records. Connecting geographic reporting systems from clinical settings with the associated data from environmental sampling will increase situational awareness. Wastewater surveillance also has the potential to help mitigate the economic impact of disease outbreaks by preventing their spread and proactively addressing potential ones, thus overcoming many barriers found in rural communities that contribute to health disparities. Moreover, wastewater surveillance allows the participation of more community members. One common theme among leaders at rural health workshops is that building trust among rural communities is the key to addressing rural health disparities (Cacari Stone et al., 2021). Treating rural communities as small urban communities often results in poor outcomes and frustration among the rural participants. Investing in long-term partnerships and building research capacity are critical for improving rural health (Cacari Stone et al., 2021). Partnerships among rural-serving academic institutions, healthcare providers, and communities allow for trust and collaboration among rural stakeholders. Designing health surveillance programs to meet the unique needs of rural populations, while building the trust necessary to implement changes, is critical in rural communities. 4 Learning from advancements in underserved areas of other countries Developing countries, which compare to some rural communities in the United States, also have lower sewer connection coverage. Despite this, they have been able to show usefulness of sanitation system surveillance for the health of the communities as part of their pandemic response. Only 43% of the global population has access to household toilets connected to sewers (WHO, 2021). Furthermore, approximately 6% of the world’s population does not have any access to sanitation facilities (WHO, 2021). However, a case in point is South Africa, where 61% of the population has access to toilets connected to sewer systems, although sewer infrastructure distribution is not uniform across provinces (WHO, 2021). Household toilets connected to sewers are common in more urbanized provinces, namely the Western Cape (87%) and Gauteng (84%) provinces, while in South African metropolitan areas, approximately 16.8% of the population lives in informal dwellings (Statistics South Africa, 2019). The adaptation model of the wastewater surveillance approach for pathogen detection in non-sewered settlements in South Africa is highly beneficial for developing countries that do not have fully inclusive sewer connections. Almost 40% of South Africans would not be included in surveillance for the health of their communities with wastewater surveillance , given a lack of sewered connections. Those left behind in sanitation access also typically lack access to sufficient healthcare or financial resources. In communities lacking formal sewerage networks, poorly or partially treated human excreta and graywater are often disposed into the environment, which enters a nearby stream or water source that has the potential to enter nearby rivers. Alternatively, community-level data and hotspot detection have been used to overcome the burden of individual testing by including a framework for sampling and surveillance of gray and wastewater within non-sewered communities, ensuring a timely response to an upsurge in disease detection. In contrast to wastewater, where a largely homogenous influent sample can be taken from a facility in South Africa, non-sewered informal settlement sampling becomes more challenging owing to the variety of systems and waste products generated. Previous sampling of non-sewered communities for SARS-CoV-2 carried out by Pocock et al. (2020) found that rivers downstream of the community were the most reliable sample sources in densely populated, non-sewered, settlements. The deployment of low-cost passive sampling devices into the rivers during periods of high rainfall, a process pioneered by Schang et al. (2021), overcomes the challenges of a dilute sample matrix in these vulnerable communities. Although this wastewater approach in non-sewered portions of South Africa cannot relate viral loads in surface water to a defined population or possible number of cases, river sampling still provides an alternative means to monitor the spread of respiratory and enteric pathogens within informal settlements by monitoring trends and presence or absence in viral loads. During the COVID-19 pandemic, this was evidenced in that it allowed the identification of possible spikes in infections within the communities upstream of the sample point. This presence or absence approach may provide an alternative method for early warning against infections in unsewered communities, which have a high risk of rapid spread and a low likelihood of convenient clinical testing to trigger the deployment of rapid response teams. 5 Limitations of rural wastewater surveillance programs Innovative approaches to implement wastewater surveillance in low-resource settings, such as rural locations, can be learned from these international efforts but are still not without limitations. Calibration of wastewater and clinical data that can be performed in urban settings remains an unrealistic expectation in rural or resource-limited settings. In such sampling programs, well-defined limits of detection and specific metrics employed to interpret results below quantification boundaries will be needed. This may require different approaches toward reporting data to rural public health officials or to a larger (academic) audience. Understanding the boundaries of rural wastewater surveillance programs is essential. From a public health perspective, it may be more important to focus on common health disparities rather than rare diseases. Similarly, sampling locations may not be easily accessible due to limitations in building-level locations, such as agricultural, food-processing industries, correctional or long-term care facilities, or limited environmental samples, such as surface water. Finally, not all households in the United States have complete plumbing facilities (Rural Community Assistance Partnership, 2015). It is estimated that 1.7 million people in the United States lack basic plumbing. Rural populations, especially minority populations living in rural locations, are more than twice as likely to not have basic plumbing. These populations cannot be served by wastewater surveillance because basic sanitation services are not available. Promoting equity among these populations is dire and challenging. 6 Conclusions A better surveillance model of disease detection among rural residents will overcome issues around the interactions of a disease and social determinants of health (Fig. 1). The SARS-CoV-2 pandemic has highlighted rural health disparities across the United States, but solutions require transdisciplinary approaches. Barriers to healthcare, comorbid diseases, poverty, and hesitation in prevention messaging from outside entities are some of the factors contributing to poor outcomes among rural patients infected with the disease. Wastewater surveillance has the potential to reduce rural health disparities by providing a flexible tool for assessing diseases in a wider community. Importantly, this form of surveillance does not rely on rural healthcare providers, who may be limited and overworked. Yet, this tool could provide the necessary warnings of potential outbreaks that may require coordinated responses from healthcare providers and community officials, as well as rural employers such as the agricultural and food-processing industries. The case study of South Africa, which demonstrates the ability to innovatively sample vulnerable communities, provides an alternative approach for monitoring rural communities in the United States. Furthermore, community participation and partnerships with local researchers involved in wastewater surveillance may help provide the boots-on-the-ground approach needed to encourage trust and the willingness to engage in mitigation measures that reduce the spread of diseases within the community. Building this infrastructure and trust will be key to improving not only health disparities related to COVID-19 but also a variety of other diseases. Wastewater surveillance is a critical tool that can be readily implemented to reduce future disparities in rural health. Funding This work was supported by the National Institutes of General Medical Sciences (grant numbers GM103443 and GM121341) and grants from the James Graham Brown Foundation, Owsley Brown II Family Foundation, Jon Rieger Seed Grant, and the Water Research Commission in South Africa. The content, study design, collection, analysis and interpretation of data, writing of the article, and decision to submit the article for publication are solely the responsibility of the authors and do not necessarily represent the views of the funding sources. CRediT authorship contribution statement Rochelle H. Holm: Writing – original draft, Writing – review & editing. Gina Pocock: Writing – original draft, Writing – review & editing. Marie A. Severson: Writing – original draft, Writing – review & editing. Victor C. Huber: Funding acquisition, Writing – review & editing. Ted Smith: Funding acquisition, Writing – review & editing. Lisa M. McFadden: Funding acquisition, Writing – original draft, Writing – review & editing. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability No data was used for the research described in the article. ==== Refs References Alcendor D.J. Targeting COVID vaccine hesitancy in rural communities in Tennessee: implications for extending the COVID-19 pandemic in the south Vaccines (Basel) 9 11 2021 1279 10.3390/vaccines9111279 34835210 Anzalone A.J. Horswell R. Hendricks B.M. Chu S. Hillegass W.B. Beasley W.H. Harper J.R. Kimble W. Rosen C.J. Miele L. McClay J.C. Santangelo S.L. Hodder S.L. Higher hospitalization and mortality rates among SARS-CoV-2-infected persons in rural America J. Rural Health 39 1 2023 39 54 35758856 Ashburn N.P. Snavely A.C. Angi R.M. Scheidler J.F. Crowe R.P. McGinnis H.D. Hiestand B.C. Miller C.D. Mahler S.A. Stopyra J.P. Prehospital time for patients with acute cardiac complaints: A rural health disparity Am. J. Emerg. Med. 52 2021 64 68 10.1016/j.ajem.2021.11.038 34871845 Behrman P. Fitzgibbon M.L. Dulin A. Wang M.L. Baskin M. Society of behavioral medicine statement on COVID-19 and rural health Transl. Behav. Med. 11 2 2021 625 630 10.1093/tbm/ibaa114 33289790 Brandt K. Goel V. Keeler C. Bell G.J. Aiello A.E. Corbie-Smith G. Wilson E. Fleischauer A. Emch M. Boyce R.M. SARS-CoV-2 testing in North Carolina: Racial, ethnic, and geographic disparities Health Place 69 2021 102576 10.1016/j.healthplace.2021.102576 Cacari Stone L. Roary M.C. Diana A. Grady P.A. State health disparities research in Rural America: Gaps and future directions in an era of COVID-19 J. Rural Health 37 3 2021 460 466 10.1111/jrh.12562 33881778 Casey M.M. Call K.T. Klingner J.M. Are rural residents less likely to obtain recommended preventive healthcare services? Am. J. Prev. Med. 21 3 2001 182 188 10.1016/S0749-3797(01)00349-X 11567838 Cavany S. Bivins A. Wu Z. North D. Bibby K. Perkins T.A. Inferring SARS-CoV-2 RNA shedding into wastewater relative to the time of infection Epidemiol. Infect. 150 2022 e21 35068403 Centers for Disease Control, 2022A. Trends in COVID-19 Cases and Deaths in the United States, by County-level Population Factors. https://covid.cdc.gov/covid-data-tracker/#pop-factors_totaldeaths. Centers for Disease Control, 2022B. COVID-19 Vaccination Equity. https://covid.cdc.gov/covid-data-tracker/#vaccination-equity. Centers for Disease Control, 2022C. National Wastewater Surveillance System (NWSS). https://www.cdc.gov/nwss/wastewater-surveillance/index.html?CDC_AA_refVal=https%3A%2F%2Fwww.cdc.gov%2Fhealthywater%2Fsurveillance%2Fwastewater-surveillance%2Fwastewater-surveillance.html. Centers for Medicare and Medicaid Services, 2021. Critical Access Hospitals. https://www.cms.gov/Medicare/Provider-Enrollment-and-Certification/CertificationandComplianc/CAHs. Chillag K.L. Lee L.M. Synergistic disparities and public health mitigation of COVID-19 in the rural United States J. Bioeth. Inq. 17 4 2020 649 656 10.1007/s11673-020-10049-0 33169255 Cross S.H. Califf R.M. Warraich H.J. Rural-urban disparity in mortality in the US from 1999 to 2019 J. Am. Med. Assoc. 325 22 2021 2312 2314 10.1001/jama.2021.5334 Cuadros D.F. Branscum A.J. Mukandavire Z. Miller F.D. MacKinnon N. Dynamics of the COVID-19 epidemic in urban and rural areas in the United States Ann. Epidemiol. 59 2021 16 20 10.1016/j.annepidem.2021.04.007 33894385 Cuadros D.F. Moreno C.M. Musuka G. Miller F.D. Coule P. MacKinnon N.J. Association between vaccination coverage disparity and the dynamics of the COVID-19 Delta and Omicron waves in the US Front. Med. (Lausanne) 9 2022 898101 10.3389/fmed.2022.898101 Cuadros D.F. Gutierrez J.D. Moreno C.M. Escobar S. Miller F.D. Musuka G. Omori R. Coule P. MacKinnon N.J. Impact of healthcare capacity disparities on the COVID-19 vaccination coverage in the United States: A cross-sectional study Lancet Regional Health-Am. 18 2023 100409 10.1016/j.lana.2022.100409 Dandachi D. Reece R. Wang E.W. Nelson T. Rojas-Moreno C. Shoemaker D.M. Treating COVID-19 in rural America J. Rural Health 37 1 2021 205 206 10.1111/jrh.12457 32362035 Dani S.S. Lone A.N. Javed Z. Khan M.S. Zia Khan M. Kaluski E. Virani S.S. Shapiro M.D. Cainzos-Achirica M. Nasir K. Khan S.U. Trends in premature mortality from acute myocardial infarction in the United States, 1999 to 2019 J. Am. Heart Assoc. 11 1 2022 e021682 34935456 Denslow S. Wingert J.R. Hanchate A.D. Rote A. Westreich D. Sexton L. Cheng K. Curtis J. Jones W.S. Lanou A.J. Halladay J.R. Chen R.J. Rural-urban outcome differences associated with COVID-19 hospitalizations in North Carolina PLoS One 17 8 2022 e0271755 10.1371/journal.pone.0271755 35976813 DeVerteuil G. Reconsidering the legacy of urban public facility location theory in human geography Prog. Hum. Geog. 24 1 2000 47 69 10.1191/030913200668094045 Diaz A. Chhabra K.R. Scott J.W. The COVID-19 pandemic and rural hospitals —Adding insult to injury Health Affairs Blog. 2020 https://www.healthaffairs.org/do/10.1377/hblog20200429.583513/full/ Dixon B.E. Grannis S.J. Lembcke L.R. Valvi N. Roberts A.R. Embi P.J. Shaman J. The synchronicity of COVID-19 disparities: statewide epidemiologic trends in SARS-CoV-2 morbidity, hospitalization, and mortality among racial minorities and in rural America PLoS One 16 7 2021 e0255063 10.1371/journal.pone.0255063 34297747 Dobis, E.A., Krumel, T.P., Cromartie, J., Conley, K.L., Sanders, A., Ortiz, R., 2021. Rural America at a glance. https://www.ers.usda.gov/webdocs/publications/102576/eib-230.pdf?v=7224.1. Gatwood J. McKnight M. Fiscus M. Hohmeier K.C. Chisholm-Burns M. Factors influencing likelihood of COVID-19 vaccination: A survey of Tennessee adults Am. J. Health Syst. Pharm. 78 10 2021 879 889 10.1093/ajhp/zxab099 33954426 Henning-Smith C. The Unique impact of COVID-19 on older adults in rural areas J. Aging Soc. Policy 32 4–5 2020 396 402 10.1080/08959420.2020.1770036 32475255 Henning-Smith C. Tuttle M. Kozhimannil K.B. Unequal distribution of COVID-19 risk among rural residents by race and ethnicity J. Rural Health 37 1 2021 224 226 10.1111/jrh.12463 32396220 Rural Health Information Hub, 2022. Rural Data Explorer. https://www.ruralhealthinfo.org/data-explorer. Hubach R.D. Shannon B. Morgan K.D. Alexander C. O'Neil A.M. Ernst C. Giano Z. COVID-19 vaccine hesitancy among rural Oklahomans Rural Remote Health 22 2 2022 7128 10.22605/RRH7128 35658522 Jackson S.L. Derakhshan S. Blackwood L. Lee L. Huang Q. Habets M. Cutter S.L. Spatial disparities of COVID-19 cases and fatalities in United States counties Int. J. Env. Res. Pub. He 18 16 2021 8259 10.3390/ijerph18168259 Kaufman B.G. Whitaker R. Pink G. Holmes G.M. Half of rural residents at high risk of serious illness due to COVID-19, creating stress on rural hospitals J. Rural Health 36 4 2020 584 590 10.1111/jrh.12481 32603030 Kevill J.L. Lambert-Slosarska K. Pellett C. Woodhall N. Richardson-O'Neill I. Pântea I. Alex-Sanders N. Farkas K. Jones D.L. Assessment of two types of passive sampler for the efficient recovery of SARS-CoV-2 and other viruses from wastewater Sci. Total Environ. 838 Pt 4 2022 156580 10.1016/j.scitotenv.2022.156580 Khazanchi, R., Powers, S.D., Rogawski, McQuade, E.T., McManus, K.A., 2021. Inequities in the geographic accessibility of COVID-19 biomedical therapeutic Trials in the United States. J. Gen. Intern. Med., 36, 11, 3650–3653. https://doi.org/10.1007/s11606-021-07081-0. Khazanchi R. Strumpf A. Essien U.R. Powers S.D. McManus K.A. Geographic accessibility of COVID-19 test to treat sites by race, ethnicity, age, and rurality JAMA Netw. Open 5 11 2022 e2241144 36350654 LaJoie A.S. Holm R.H. Anderson L.B. Ness H.D. Smith T. Aslan A. Nationwide public perceptions regarding the acceptance of using wastewater for community health monitoring in the United States PLoS One 17 10 2022 e0275075 10.1371/journal.pone.0275075 36219594 Lakhani H.V. Pillai S.S. Zehra M. Sharma I. Sodhi K. Systematic review of clinical insights into novel coronavirus (COVID-19) pandemic: Persisting challenges in U.S. rural population Int. J. Environ. Res. Public Health 17 12 2020 4279 10.3390/ijerph17124279 32549334 Li D. Gaynor S.M. Quick C. Chen J.T. Stephenson B.J.K. Coull B.A. Lin X. Identifying US county-level characteristics associated with high COVID-19 burden BMC Public Health 21 1 2021 1007 10.1186/s12889-021-11060-9 34049526 Li Z Lewis B Berney K Social Vulnerability and Rurality Associated With Higher Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infection-Induced Seroprevalence: A Nationwide Blood Donor Study-United States, July 2020-June 2021 Clin Infect Dis 75 1 2022 e133 e143 10.1093/cid/ciac105 35137014 Liu Y. Yin Z. Ni C. Yan C. Wan Z. Malin B. Examining rural and urban sentiment difference in COVID-19–related topics on Twitter: Word embedding–based retrospective study J. Med. Internet Res. 25 2023 e42985 36790847 Manuel, D., Amadei, C.A., Campbell, J.R., Brault, J.M., Zierler, A., 2022. Strengthening public health surveillance through wastewater testing: An essential investment for the COVID-19 pandemic and future health threats. Washington, DC: World Bank. Strengthening-Public-Health-Surveillance-Through-Wastewater-Testing-An-Essential-Investment-for-the-COVID-19-Pandemic-and-Future-Health-Threats (2).pdf. McElfish P.A. Willis D.E. Shah S.K. Bryant-Moore K. Rojo M.O. Selig J.P. Sociodemographic determinants of COVID-19 vaccine hesitancy, fear of infection, and protection self-efficacy J. Prim. Care Community Health 12 2021 10.1177/21501327211040746 21501327211040746 Medina C.Y. Kadonsky K.F. Roman F.A. Jr. Tariqi A.Q. Sinclair R.G. D'Aoust P.M. Delatolla R. Bischel H.N. Naughton C.C. The need of an environmental justice approach for wastewater-based epidemiology for rural and disadvantaged communities: A review in California Curr. Opin. Environ. Sci. Health 27 2022 100348 10.1016/j.coesh.2022.100348 Mercier E. D'Aoust P.M. Thakali O. Hegazy N. Jia J.J. Zhang Z. Eid W. Plaza-Diaz J. Kabir M.P. Fang W. Cowan A. Stephenson S.E. Pisharody L. MacKenzie A.E. Graber T.E. Wan S. Delatolla R. Municipal and neighbourhood level wastewater surveillance and subtyping of an influenza virus outbreak Sci. Rep. 12 1 2022 15777 10.1038/s41598-022-20076-z 36138059 Nelson B. What poo tells us: wastewater surveillance comes of age amid covid, monkeypox, and polio BMJ 378 2022 o1869 10.1136/bmj.o1869 Rural Community Assistance Partnership, 2015. Still living without the basics in the 21st century: Analyzing the availability of water and sanitation services in the United States. http://opportunitylinkmt.org/wp-content/uploads/2015/07/Still-Living-Without-the-Basics-Water.pdf. Pocock G. Coetzee L. Mans J. Taylor M. Genthe B. Proof of Concept Study: Application of wastewater-based surveillance to monitor SARS-CoV-2 prevalence in South African Communities WRC Report TT832/20. 2020 Rader B. Astley C.M. Sy K.T.L. Sewalk K. Hswen Y. Brownstein J.S. Kraemer M.U.G. Geographic access to United States SARS-CoV-2 testing sites highlights healthcare disparities and may bias transmission estimates J. Travel Med. 27 7 2020 taaa076 10.1093/jtm/taaa076 32412064 Ratcliffe M. Burd C. Holder K. Fields A. Defining Rural at the U.S Census Bureau. 2016 https://www.census.gov/content/dam/Census/library/publications/2016/acs/acsgeo-1.pdf Ridenhour B.J. Sarathchandra D. Seamon E. Brown H. Leung F.-Y. Johnson-Leon M. Megheib M. Miller C.R. Johnson-Leung J. Molina J.A. Effects of trust, risk perception, and health behavior on COVID-19 disease burden: Evidence from a multi-state US survey PLoS One 17 5 2022 e0268302 10.1371/journal.pone.0268302 35594254 Ryerson A.B. Lang D. Alazawi M.A. Neyra M. Hill D.T. St. George K. Fuschino M. Lutterloh E. Backenson B. Rulli S. Ruppert P.S. Lawler J. McGraw N. Knecht A. Gelman I. Zucker J.R. Omoregie E. Kidd S. Sugerman D.E. Jorba J. Gerloff N. Ng T.F.F. Lopez A. Masters N.B. Leung J. Burns C.C. Routh J. Bialek S.R. Oberste M.S. Rosenberg E.S. Anderson B.J. Anderson N. Augustine J.A. Baldwin M. Barrett K. Bauer U. Beck A. Belgasmi H. Bennett L.J. Bhatt A. Blog D. Boss H. Brenner I.R. Brister B. Brown T.W. Buchman T. Bullows J. Connelly K. Cassano B. Castro C.J. Cirillo C. Cone G.E. Cory J. Dasin A. de Coteau A. DeSimone A. Chauvin F. Dixey C. Dooling K. Doss S. Duggar C. Dunham C.N. Easton D. Egan C. Emery B.D. English R. Faraci N. Fast H. Feumba G.S. Fischer N. Flores S. Frolov A.D. Getachew H. Gianetti B. Godinez A. Gray T. Gregg W. Gulotta C. Hamid S. Hammette T. Harpaz R. Smith L.H. Hanson B. Henderson E. Heslin E. Hess A. Hoefer D. Hoffman J. Hoyt L. Hughes S. Hutcheson A.R. Insaf T. Ionta C. Miles S.J. Kambhampati A. Kappus-Kron H.R. Keys G.N. Kharfen M. Kim G. Knox J. Kovacs S. Krauchuk J. Krow-Lucal E.R. Lamson D. Laplante J. Larsen D.A. Link-Gelles R. Liu H. Lueken J. Ma K. Marine R.L. Mason K.A. McDonald J. McDonough K. McKay K. McLanahan E. Medina E. Meek H. Mustafa G.M. Meldrum M. Mello E. Mercante J.W. Mhatre M. Miller S. Migliore N. Mita-Mendoza N.K. Moghe A. Momin N. Morales T. Moran E.J. Nabakooza G. Neigel D. Ogbamikael S. O’Mara J. Ostrowski S. Patel M. Paul P. Paziraei A. Peacock G. Pearson L. Plitnick J. Pointer A. Popowich M. Punjabi C. Ramani R. Raymond S.J. Rickerman L. Rist E. Robertson A.C. Rogers S.L. Rosen J.B. Sanders C. Santoli J. Sayyad L. Schoultz L. Shudt M. Smith J. Smith T.L. Souto M. Staine A. Stokley S. Sun H. Terranella A.J. Tippins A. Tobolowsky F. Wallace M. Wassilak S. Wolfe A. Yee E. Wastewater testing and detection of poliovirus type 2 genetically linked to virus isolated from a paralytic polio case - New York, March 9-October 11, 2022 MMWR Morb. Mortal. Wkly Rep. 71 44 2022 1418 1424 36327157 Salmon D.A. Dudley M.Z. Brewer J. Kan L. Gerber J.E. Budigan H. Proveaux T.M. Bernier R. Rimal R. Schwartz B. COVID-19 vaccination attitudes, values and intentions among United States adults prior to emergency use authorization Vaccine 39 19 2021 2698 2711 10.1016/j.vaccine.2021.03.034 33781601 Schang C. Crosbie N.D. Nolan M. Poon R. Wang M. Jex A. John N. Baker L. Scales P. Schmidt J. Thorley B.R. Hill K. Zamyadi A. Tseng C.W. Henry R. Kolotelo P. Langeveld J. Schilperoort R. Shi B. Einsiedel S. Thomas M. Black J. Wilson S. McCarthy D.T. Passive sampling of SARS-CoV-2 for wastewater surveillance Environ. Sci. Tech. 55 15 2021 10432 10441 10.1021/acs.est.1c01530 Shah P. Owens J. Franklin J. Mehta A. Heymann W. Sewell W. Hill J. Barfield K. Doshi R. Demographics, comorbidities and outcomes in hospitalized COVID-19 patients in rural southwest Georgia Ann. Med. 52 7 2020 354 360 10.1080/07853890.2020.1791356 32620056 Sharma S. Cain J. Sakhuja A. Schaefer G. Krupica T. Sarwari A. Guidance for Healthcare Providers Managing COVID-19 in Rural and Underserved Areas J. Racial Ethn. Health Disparities 7 5 2020 817 821 10.1007/s40615-020-00820-9 32651882 Smith T. Holm R.H. Keith R.J. Amraotkar A.R. Alvarado C.R. Banecki K. Choi B. Santisteban I.C. Bushau-Sprinkle A.M. Kitterman K.T. Fuqua J. Hamorsky K.T. Palmer K.E. Brick J.M. Rempala G.A. Bhatnagar A. Quantifying the relationship between sub-population wastewater samples and community-wide SARS-CoV-2 seroprevalence Sci. Total Environ. 853 2022 158567 10.1016/j.scitotenv.2022.158567 Souch J.M. Cossman J.S. A commentary on rural-urban disparities in COVID-19 testing rates per 100,000 and risk factors J. Rural Health 37 1 2021 188 190 10.1111/jrh.12450 32282964 South Dakota Department of Health, 2020. South Dakota Primary Care Needs Assessment. https://doh.sd.gov/documents/Providers/RuralHealth/2016-2020_PrimaryCareNeedsAssessment.pdf. Statistics South Africa. (2019). Toilet facilities. (E. S. SA, Compiler) Pretoria. Summers-Gabr N.M. Rural-urban mental health disparities in the United States during COVID-19 Psychol Trauma-US 12 S1 2020 10.1037/tra0000871 S222–S224 Tanne J.H. Polio emergency declared in New York State over virus found in wastewater BMJ 378 2022 o2211 10.1136/bmj.o2211 Wheeler P.H. Patten C.A. Wi C.I. Bublitz J.T. Ryu E. Ristagno E.H. Juhn Y.J. Role of geographic risk factors and social determinants of health in COVID-19 epidemiology: Longitudinal geospatial analysis in a midwest rural region J. Clin. Transl. Sci. 6 1 2021 e51 35651962 Winkelman T.N.A. Rai N.K. Bodurtha P.J. Chamberlain A.M. DeSilva M. Jeruzal J. Johnson S.G. Kharbanda A. Klyn N. Mink P.J. Muscoplat M. Waring S. Yu Y. Drawz P.E. Trends in COVID-19 vaccine administration and effectiveness through October 2021 JAMA Netw. Open 5 3 2022 e225018 35357452 World Health Organization (WHO) and the United Nations Children’s Fund (UNICEF), 2021. Progress on household drinking water, sanitation and hygiene 2000-2020: Five years into the SDGs. Geneva, 2021. https://data.unicef.org/resources/progress-on-household-drinking-water-sanitation-and-hygiene-2000-2020/.
PMC010xxxxxx/PMC10292651.txt	==== Front Environ Health Perspect Environ Health Perspect EHP Environmental Health Perspectives 0091-6765 1552-9924 Environmental Health Perspectives EHP12902 10.1289/EHP12902 Science Selection Using Only as Needed: California Limits on Agricultural Antibiotics May Bring Relief from Resistant Infections https://orcid.org/0000-0001-8387-1554 Washam Cynthia 26 6 2023 6 2023 131 6 06400314 2 2023 12 5 2023 https://ehp.niehs.nih.gov/about-ehp/license EHP is an open-access journal published with support from the National Institute of Environmental Health Sciences, National Institutes of Health. All content is public domain unless otherwise noted. Young pigs on bed of straw in a pen, with one looking at the camera ==== Body pmcA California law restricting the use of antimicrobials in livestock is associated with a lower risk of drug-resistant infections in humans, according to a new study published in Environmental Health Perspectives.1 Urinary-tract infections (UTIs) resistant to a common class of antibiotics are 7.1% lower in California than a model predicted they would be if the state had not enacted the 2015 law, which was the first of its kind in the United States (Maryland enacted a related law in 2017).2 California Senate Bill 27 (SB27), which took effect 1 January 2018, bans routine preventive treatment of food-producing animals with medically important antimicrobials, meaning those also used to treat human infections. SB27 does allow necessary antimicrobial treatment of animals that are sick or at risk of illness, with a veterinarian’s prescription.3 Historically, farmers bought antimicrobials over the counter and administered them routinely in feed and water to prevent infections. This continual exposure to antimicrobials promotes resistant mutations, which can spread quickly in industrial livestock facilities, says Keeve Nachman, an associate professor at Johns Hopkins Bloomberg School of Public Health and a coauthor of the study. New evidence suggests a recently implemented policy on use of medically important antibiotics in livestock is having the desired effect: fewer antimicrobial-resistant infections in people. Image: © iStock/sircco. Young pigs on bed of straw in a pen, with one looking at the camera Resistant microbes can reach consumers who eat contaminated meat or the produce grown in fields fertilized with livestock waste, Nachman says, whereas people living near farms risk exposure to resistant pathogens through air and water. Antimicrobial-resistant infections cause approximately 48,000 deaths annually in the United States4 and are ranked among the top 10 global health threats by the World Health Organization.5 Ingested microbes can colonize a person’s gut, potentially leading to a UTI that resists treatment.6 UTIs are the most common outpatient infections, and most are caused by Escherichia coli.7 The strains of E. coli responsible for UTIs and other extraintestinal infections differ from those that cause intestinal illness, such as diarrhea, and those that coexist in the human gut with other bacteria and in some cases are even beneficial.8,9 Worldwide, almost 405 million cases were reported in 2019.10 Women are more susceptible than men to UTIs because of hormonal differences and anatomical differences in their urinary tracts,11 and the risk of infection increases in older adults.12 Serious consequences, including death, may occur if infection spreads to the kidneys,13 prostate,14 or bloodstream.15 Lead author Joan Casey, an assistant professor at the University of Washington School of Public Health, and her coauthors assessed the potential impact of SB27 by comparing actual rates in California of UTIs caused by resistant E. coli between 2013 and 2020 with rates that a “synthetic California” control model estimated would have occurred without the law. The authors created their synthetic California model by analyzing pre-SB27 state-level variables—including distribution of residents by sex, race, ethnicity, and age—that are associated with antibiotic-resistant UTIs, Casey explains. Her team also considered the spatial distribution of slaughterhouses, hypothesizing that living near one or more slaughterhouses would increase risk for an antimicrobial-resistant UTI. The researchers then used data from 7.1 million urine specimens collected at outpatient sites across 33 states and tested for E. coli resistance to four classes of antibiotics. Infection data from states with demographics that closely matched California’s were weighted more heavily than data from less similar states1 to approximate the state’s hypothetical experience more closely. Casey and her coauthors expected to see fewer actual UTIs that were resistant to tetracyclines and extended-spectrum cephalosporins (ESC) post SB27 because these classes are used in both humans and livestock. They were encouraged by the estimated 7.1% lower actual incidence of ESC-resistant infections compared with the synthetic control estimate, though tetracycline resistance did not decline, for reasons the authors were unable to pinpoint. They suggested that it might be because some of the meat sold in California is raised out of state, or that compliance with SB27 might be poor. The U.S. Food and Drug Administration (FDA) took action in 2013 to reduce use of antimicrobials for growth promotion,16 which had been widespread in industrial livestock operations.2 Unlike the state of California, the U.S. federal government does allow use of antimicrobials to prevent disease in healthy food-producing animals; however, as of 2017, a veterinarian’s prescription has been required.2 The FDA reported that sales of medically important antimicrobials for livestock decreased 38% from the peak year of 2015 through 2021.17 Sales decreased most dramatically for poultry; sales for cattle, swine, and turkeys also saw declines.2 The authors’ finding of a difference in infection rates between the actual and synthetic California models impressed Michael Martin, an associate clinical professor at the University of California, San Francisco. “When analyzing a large, complex data set, it is often difficult to show a difference between groups,” Martin says. “It is remarkable the authors could show a statistically significant difference. The likelihood of that happening by chance is very low.” The researchers did not have access to data on quantities of antimicrobials used. This lack “significantly reduced our ability to interpret our findings and place them in the larger context of why we might be seeing this change,” Casey says. Nevertheless, she adds, “Our study provides evidence that similar policies in other places could have the same effect.” Cynthia Washam writes for science and medical publications from her home in South Florida. ==== Refs References 1. Casey JA, Tartof SY, Davis MF, Nachman KE, Price L, Liu C, et al. 2023. Impact of a statewide livestock antibiotic use policy on resistance in human urine Escherichia coli isolates: a synthetic control analysis. Environ Health Perspect 131 (2 ):27007, PMID: , 10.1289/EHP11221.36821707 2. Wallinga D, Smit LAM, Davis MF, Casey JA, Nachman KE. 2022. A review of the effectiveness of current US policies on antimicrobial use in meat and poultry production. Curr Environ Health Rep 9 (2 ):339–354, PMID: , 10.1007/s40572-022-00351-x.35477845 3. State of California. California Legislative Information: Food and Agricultural Code FAC, Chapter 4.5. Livestock: Use of Antimicrobial Drugs. https://leginfo.legislature.ca.gov/faces/codes_displayText.xhtml?lawCode=FAC&division=7.&title=&part=&chapter=4.5.&article [accessed 7 June 2023]. 4. U.S. CDC (U.S. Centers for Disease Control and Prevention). National Infection & Death Estimates for Antimicrobial Resistance. https://www.cdc.gov/drugresistance/national-estimates.html [accessed 7 June 2023]. 5. World Health Organization. Antimicrobial Resistance. https://www.who.int/news-room/fact-sheets/detail/antimicrobial-resistance [accessed 7 June 2023] 6. Nordstrom L, Liu CM, Price LB. 2013. Foodborne urinary tract infections: a new paradigm for antimicrobial-resistant foodborne illness. Front Microbiol 4 :29, PMID: , 10.3389/fmicb.2013.00029.23508293 7. Flores-Mireles AL, Walker JN, Caparon M, Hultgren SJ. 2015. Urinary tract infections: epidemiology, mechanisms of infection and treatment options. Nat Rev Microbiol 13 (5 ):269–284, PMID: , 10.1038/nrmicro3432.25853778 8. Russo TA, Johnson JR. 2000. Proposal for a new inclusive designation for extraintestinal pathogenic isolates of Escherichia coli: exPEC. J Infect Dis 181 (5 ):1753–1754, PMID: , 10.1086/315418.10823778 9. American Society for Microbiology. 2011. FAQ: E. Coli: Good, Bad, & Deadly. Washington, DC: American Society for Microbiology. 10.1128/AAMCol.1-2011. 10. Zhu C, Wang D-Q, Zi H, Huang Q, Gu J-M, Li L-Y, et al. 2021. Epidemiological trends of urinary tract infections, urolithiasis and benign prostatic hyperplasia in 203 countries and territories from 1990 to 2019. Mil Med Res 8 (1 ):64, PMID: , 10.1186/s40779-021-00359-8.34879880 11. Abelson B, Sun D, Que L, Nebel RA, Baker D, Popiel P, et al. 2018. Sex differences in lower urinary tract biology and physiology. Biol Sex Differ 9 (1 ):45, PMID: , 10.1186/s13293-018-0204-8.30343668 12. Cortes-Penfield NW, Trautner BW, Jump RLP. 2017. Urinary tract infection and asymptomatic bacteriuria in older adults. Infect Dis Clin North Am 31 (4 ):673–688, PMID: , 10.1016/j.idc.2017.07.002.29079155 13. Scholes D, Hooton TM, Roberts PL, Gupta K, Stapleton AE, Stamm WE, et al. 2005. Risk factors associated with acute pyelonephritis in healthy women. Ann Intern Med 142 (1 ):20–27, PMID: , 10.7326/0003-4819-142-1-200501040-00008.15630106 14. Nickel JC. 1999. Prostatitis: evolving management strategies. Urol Clin North Am 26 (4 ):737–751, PMID: , 10.1016/s0094-0143(05)70215-9.10584615 15. Lee J-C, Lee N-Y, Lee H-C, Huang W-H, Tsui K-C, Chang C-M, et al. 2012. Clinical characteristics of urosepsis caused by extended-spectrum beta-lactamase-producing Escherichia coli or Klebsiella pneumonia and their emergence in the community. J Microbiol Immunol Infect 45 (2 ):127–133, PMID: , 10.1016/j.jmii.2011.09.029.22041167 16. U.S. CDC. 2012. CVM GFI #209 The Judicious Use of Medically Important Antimicrobial Drugs in Food-Producing Animals. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/cvm-gfi-209-judicious-use-medically-important-antimicrobial-drugs-food-producing-animals [accessed 7 June 2023]. 17. U.S. Food and Drug Administration. 2022. 2021 Summary Report on Antimicrobials Sold or Distributed for Use in Food-Producing Animals. https://www.fda.gov/media/163739/download [accessed 7 June 2023].
PMC010xxxxxx/PMC10292659.txt	==== Front Psychoneuroendocrinology Psychoneuroendocrinology Psychoneuroendocrinology 0306-4530 1873-3360 The Author(s). Published by Elsevier Ltd. S0306-4530(23)00302-5 10.1016/j.psyneuen.2023.106324 106324 Corrigendum Corrigendum to “Acute TNFα levels predict cognitive impairment 6–9 months after COVID-19 infection” [Psychoneuroendocrinology 153 (2023) 106104] Nuber-Champier A. ab Cionca A. a Breville G. b Voruz P. abc Jacot de Alcântara I. abc Allali G. d Lalive P.H. bc Benzakour L. ce Lövblad K.-O. cf Braillard O. g Nehme M. g Coen M. h Serratrice J. h Reny J.-L. h Pugin J. ci Guessous I. cg Landis B.N. cj Griffa A. bk Van De Ville D. ck Assal F. bc Péron J.A. ab⁎ a Clinical and Experimental Neuropsychology Laboratory, Faculty of Psychology, University of Geneva, Geneva, Switzerland b Neurology Division, Geneva University Hospitals, Switzerland c Faculty of Medicine, University of Geneva, Switzerland d Leenaards Memory Center, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland e Psychiatry Department, Geneva University Hospitals, Switzerland f Diagnostic and Interventional Neuroradiology Department, Geneva University Hospitals, Switzerland g Division and Department of Primary Care Medicine, Geneva University Hospitals, Switzerland h Division of General Internal Medicine, Department of Medicine, Geneva University Hospitals and Geneva University, Switzerland i Intensive Care Department, Geneva University Hospitals, Switzerland j Rhinology-Olfactology Unit, Otorhinolaryngology Department, Geneva University Hospitals, Switzerland k Institute of Bioengineering, Center for Neuroprosthetics, Ecole Polytechnique Federale de Lausanne (EPFL), Geneva, Switzerland ⁎ Correspondence to: Faculté de Psychologie et des Sciences de l′Education, 40 bd du Pont d′Arve, 1205 Geneva, Switzerland. 26 6 2023 26 6 2023 106324© 2023 The Authors 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThe authors regret “the absence of online supplementary material concerning the article "Acute TNFα levels predict cognitive impairment 6–9 months after COVID-19 infection". All supplementary material is therefore available in this document”. The authors would like to apologise for any inconvenience caused. Appendix A Supplementary material Supplementary material . Appendix A Supplementary data associated with this article can be found in the online version at doi:10.1016/j.psyneuen.2023.106324.
PMC010xxxxxx/PMC10292660.txt	==== Front J Allergy Clin Immunol J Allergy Clin Immunol The Journal of Allergy and Clinical Immunology 0091-6749 1097-6825 American Academy of Allergy, Asthma & Immunology S0091-6749(23)00804-7 10.1016/j.jaci.2023.06.008 Paradigms and Perspectives The translation potential of harnessing the resolution of inflammation Collins George MBBS, BSc, MRCP de Souza Carvalho Jhonatan DDS, MSc Gilroy Derek W. PhD ∗ Department for Experimental and Translational Medicine, Division of Medicine, University College London ∗ Corresponding author: Derek W. Gilroy, PhD, 5 University St, University College London, London WC1E 6JJ, United Kingdom. 21 6 2023 21 6 2023 1 2 2023 13 6 2023 16 6 2023 © 2023 American Academy of Allergy, Asthma & Immunology. 2023 American Academy of Allergy, Asthma & Immunology Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Key words Macrophages cell death efferocytosis maladapted inflammation ==== Body pmcThese are heady times in inflammation research. Although immunology and pharmacology have always been important, rarely have they been so popular. The pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a reminder of this, and with 4 such outbreaks in the past 20 years, including severe acute respiratory syndrome (in 2002), swine flu (in 2009), Middle East respiratory syndrome (in 2015) and coronavirus disease 2019 (COVID-19), it is not unreasonable to expect another pandemic soon. Another problem is the impact of age on our immune systems. This is often referred to as “inflammaging” or immunosenescence. Complications of aging include multimorbidities (2 or more illnesses or diseases occurring in the same person at the same time), prolonged wound healing, vaccine inefficacy, and frailty.1 , 2 With the number of persons aged 60 or older expected to rise from 962 million globally in 2017 to 2.1 billion in 2050 and 3.1 billion in 2100, this will result in financial strain on society and increased pressure being placed on national health care systems (https://www.who.int/news-room/fact-sheets/detail/ageing-and-health). In addition, we have yet to find cures for noncommunicable diseases such as rheumatoid arthritis, cancer, psoriasis, asthma, SLE, multiple sclerosis, and atherosclerosis. In short, there is a great deal to do. Understanding the pathogenesis of communicable and noncommunicable diseases has been a herculean effort that has borne comparatively little fruit despite the financial, academic, creative, and technological input. In addition to playing “safe science,” there are significant experimental hurdles in the way, including an overreliance on rodent models of inflammation. Certainly, experiments on humans offer the advantage of being directly relevant to human physiology and medicine. However, human models of inflammation are limited by ethical and practical considerations, such as the difficulty of obtaining samples, a limited number of patients suitable for study, aging, multimorbidity, polypharmacy, and the inability to glean mechanistic proof of a hypothesis in the definitive manner afforded by rodents. Nonetheless, we have been very good at treating the symptoms of chronic inflammatory diseases by blocking pathways that drive inflammation, namely, heat, swelling, redness, and pain. Such treatments primarily include nonsteroidal anti-inflammatory drugs, biologics, and steroids. However, these medicines do not bring about a cure, are ineffective in subsets of patients, and have side effects. Thus, there is a need to identify more effective and safer therapeutics to treat chronic inflammatory diseases. Consequently, attention has turned to the other end of the inflammatory spectrum, namely, resolution, to understand the endogenous processes involved in switching off inflammation. The idea is to identify novel internal counterregulatory systems that can be harnessed for therapeutic gain and have fewer side effects than current anti-inflammatory medicines have.3 Inflammatory resolution is being studied extensively and has been demonstrated to be an active process with quantifiable indexes and specific physiologic requirements.4 Key determinants of resolution include clearing pathogenic stimuli through the phagocytosis-mediated generation of reactive oxygen species in the pathogen-containing vacuole, fusion with intracellular phagocyte granules, or NETosis. Once the pathogen has been neutralized, there is a catabolism of proinflammatory mediators via adsorption onto the surface of apoptotic bodies or scavenging by D6, which is a decoy receptor expressed at low levels on leukocytes but at high levels on trophoblasts and endothelial cells of lymphatic afferent vessels in the skin, gut and lung. Deletion of D6 in mice predisposes them to development of chronic inflammation. The resolution cascade continues with proinflammatory signaling pathways being silenced, which is a surprisingly understudied aspect of resolution biology. These include LRRC33, which inhibits Toll-like receptor (TLR) 4/nuclear factor-κB activation or tristetraprolin, which destabilizes mRNA transcripts that encode several diverse proinflammatory modulators. These have all contributed to the evidence that resolution is both tractable and druggable, a process that is ripe for drug discovery, and the opportunity to understand the etiology of chronic inflammatory diseases.5 Before a discussion of cell death and clearance, in some tissues—the colon, for instance—hypoxia is a requirement for resolution through several mechanisms, including extracellular acidification, purine/adenosine biosynthesis, and generation of proresolving lipid mediators, driven by phagocyte oxygen consumption and stabilization of hypoxia-inducible factor-1α (HIF-1α). In this regard, the phagocyte effectors of host response are also key regulators of resolution. Consequently, disruptions in the microenvironment can prevent the resolution of acute inflammation and favor chronic inflammatory lesions. The pharmacologic stabilization of HIF is already showing promise as an anti-inflammatory therapy in human patients (ClinicalTrials.gov identifier NCT04353791). When these phagocytes have performed their primary function, they must be cleared from the inflamed site. For these cells, death comes in many guises, with apoptosis (a process that parcels up effete cells for nonphlogistic removal by phagocytes [notably, macrophages]) being the most studied.6 On this note, efferocytosis is the process by which phagocytes clear and recycle cellular debris and apoptotic cells. The molecular pathways involved include (1) recognition of and binding to surface markers on apoptotic cells such as phosphatidylserine via TIM4 expressed on the surface of phagocytes, which we found to be downregulated on the phagocytes of elderly people via tonically elevated p38 mitogen-activated protein kinase activity7; (2) engulfment; and (3) signal transduction, where following binding and engulfing the apoptotic cell, MerTK and STAT 3 dampen proinflammatory signaling, followed by (4) degradation, where the phagocyte then uses lysosomal enzymes to degrade the apoptotic cell, after which the remaining fragments are recycled or excreted. On this theme, recent research has shown that apoptotic cell–derived methionine is used by phagocytes to drive proresolution pathways and the secretion of immune-dampening signals such as TGF-β8 (Fig 1 ).Fig 1 Classic resolution. Following infection, locally released chemokines and cytokines, as well as upregulated cell adhesion molecules, facilitate granulocyte accumulation into tissues. Granulocytes such as neutrophils phagocytose pathogens and die as a result of apoptosis. Phosphatidylserine, expressed on the surface of apoptotic cells, engages with macrophage (Mᵩ) TIM-4 in preparation for efferocytosis. This has the dual effect of clearing dangerous granulocytes along with their dangerous cargo of histotoxic agents and endogenous antigens within their cell surface blebs in a manner that programs Mᵩs down a proresolution and/or wound-healing pathway. The resolution cascade continues with proinflammatory signaling pathways being silenced, which is a surprisingly understudied aspect of resolution biology. These include LRRC33, which inhibits TLR4/nuclear factor-κB activation, or tristetraprolin (TTP), which destabilizes mRNA transcripts that encode several diverse proinflammatory modulators. This is followed by the infiltration of regulatory T (Treg) cells, which maintain immune tolerance. Mφs clear immune debris, drive wound healing, maintain tolerance, and resolve inflammatory responses. Thus, not only does programmed cell death of granulocytes safely remove cellular entities that have a great capacity to cause harm (through the release of cytotoxic agents and secretion of antigen from their membrane blebs), but it also programs phagocytes down a proresolution pathway. So important are these events that defects in efferocytosis are believed to underpin the pathogenesis of SLE9 and prolonged inflammation in elderly individuals.7 Indeed, a significant source of tissue damage in inflammatory conditions, including asthma, rheumatoid arthritis, and inflammatory bowel disease, is attributed to neutrophil nonclearance. However, proresolution processes do not stop at clearing up dead cells and immune debris. Pain must be switched off, and the microvascular hyperreactivity that caused redness must be reversed by mechanisms yet to be fully defined. This is, essentially, damping down Celsus’ 4 cardinal signs in a manner that was for many years considered to be passive in nature. It is around this time that regulatory T cells accumulate in the resolved tissue to maintain immune tolerance. Moreover, we are increasingly appreciating the fact that resolution of immune responses to infection is followed by a prolonged phase of immune activity that imprints long-term tissue immunity.10 So far, so good. By understanding the mechanisms that switch off inflammation, we can develop new therapies promoting diseases driven by chronic inflammation down a proresolution pathway. This could lead to the development of more effective and targeted treatments for these conditions, which could improve patient outcomes and reduce adverse effects. However, proresolution pathways are likely species, tissue, stimulus, sex, and perhaps ethnicity specific. For instance, the processes required to resolve inflammation driven by infectious diseases and their postresolution sequelae10 are not the same as those required to switch off inflammation driven by, say, noncommunicable diseases. Indeed, there is little point in making resolving inflammation resolve faster. Our next step is to understand how immune responses become dysregulated, leading to chronic nonresolving inflammation—in other words, how proresolution pathways become silenced, leading to maladaptation to the point of no return. Here, tolerance is broken, and endogenous antigens feed the development of autoimmunity, whereas epigenetic changes occur in both leukocytes and stromal cells—the latter turning from innocent bystanders into pathogenic drivers of disease. Hence, inflammatory responses do not always follow the prescribed resolution script. For instance, failed pathogen clearance is the hallmark of chronic granulomatous disease, where neutrophils have impaired superoxide production due to defects in NADPH oxidase, resulting in recurrent and severe infections. Autoimmune lymphoproliferative syndrome arises from a mutation in the cell death receptor Fas (CD95), dysregulating lymphocyte homeostasis and hence increasing the risk of autoimmune diseases. Age-acquired defects in efferocytosis in humans lead to immune cell debris accumulation of M1-like macrophage in tissues.7 Besides dying as a result of apoptosis as the preferred mode of death leading to resolution, neutrophils can also die by necroptosis, ferroptosis, pyroptosis, parthanatos, necrosis, and NETosis, resulting in more complicated outcomes. Whereas apoptosis takes hours, necrosis, for example, is faster. In necrosis, intracellular damage-associated molecular patterns leak out from the damaged cell, triggering inflammation via TLR2, TLR4, TLR9, and RAGE. NETosis represents a unique form of cell death, resulting in neutrophil extracellular trap (NET) formation. Nuclear chromatin decorated with nuclear proteins, elastase, high mobility group protein B1, myeloperoxidase, proteinase-3, glycolytic enzymes, and cytoskeletal proteins are released from neutrophils into the extracellular environment. NETs help eradicate infections and promote the resolution of neutrophilic inflammation by degrading cytokines and chemokines, as well as sterile crystal-mediated inflammation. NETs, however, can also promote the posttranslational modification of proteins and other macromolecules while revealing autoantigens such as DNA and histones to the immune system, thereby increasing the risk of autoimmune disease. Consequently, a variety of inhibitors that prevent NET formation as well as molecules that degrade NETs are under investigation for the treatment of inflammatory diseases. However, as with most modulators of host defense, NET inhibition in animal models increases susceptibility to infections and decreases neutrophil functions involved in innate immunity. Hence, besides inherent or acquired defects in resolution, there are many factors that determine whether inflamed tissues resolve or progress to chronicity; they include (1) the pathogenic nature of the infection; (2) the severity of tissue injury; (3) the immune cell infiltrate veering from hypoxia to “inflammatory acidification,” triggering glycolysis and lactic acid secretion; (4) the magnitude of tissue damage; and (5) the modes of immune cell death (Fig 2 ). On this note, when a neutrophil switches from apoptosis to other forms of unwanted cell death in the evolution of chronic inflammatory disease remains unknown. Investigations have demonstrated that only a subpopulation of neutrophils undergoes NETosis, suggesting heterogeneity within neutrophil populations. In SLE, for instance, low-density granulocytes have a greater tendency to form NETs than do normal density granulocytes, such that an expansion in low-density granulocytes in lupus might explain a link between this disease and NET formation. This would suggest that a chronically activated immune system “enriches” for cells and pathways that drive toward maladaptation and multimorbidities.Fig 2 Inflammation clears infection, heals wounds, and restores homeostasis, resulting in resolution. But, it can be overexuberant from the start and fail to resolve, leading to chronic inflammation. The reasons for this are many and varied: inherent or acquired defects in pathogen clearance or efferocytosis, respectively. In addition, there are (1) the pathogenic nature of the infection; (2) the immune cell infiltrate veering from hypoxia to “inflammatory acidification,” triggering glycolysis and lactic acid secretion; (3) the magnitude of tissue damage, and (4) the modes of immune cell death leading to damage-associated molecular patterns (DAMPs), NETs, and tissue stress. Mᵩ, Macrophage, Mins, minutes. This is the “real-life” challenge of curing chronic inflammatory diseases—resetting immunity, invoking tissue homeostasis, and restoring proper function. Moreover, we need to remember that this is a challenge of the human condition, complete with multimorbidities, immunosenescence, and polypharmacy, which do not beset rodents. Therefore, we need more appropriate experimental human models of disease and approaches to identify proinflammatory pathways that suppress resolution in a context-specific manner. Additionally, early diagnosis and activation of proresolution pathways will be more effective than trying to cure maladapted chronic inflammation. In summary, the reality is that we are faced with the challenge of driving chronic inflammation down a proresolution pathway. Whether engaging proresolution mechanisms is sufficient to do so—should they be active and amenable to manipulation in the chronic disease of interest—has yet to be fully proved. Perhaps, inhibiting proinflammatory signals is sufficient to trigger spontaneous resolution. However, it is also possible that some diseases may be too advanced and maladapted, such that proresolution pathways are permanently repressed or dysregulated and some other novel approach is required. That notwithstanding, for those in the field of resolution biology, there is much to do and much to learn. It seems that although we need to continue identifying the factors driving acute resolution, we also need to identify those that subvert it. ==== Refs References 1 Kingston A. Robinson L. Booth H. Knapp M. Jagger C. project M. Projections of multi-morbidity in the older population in England to 2035: estimates from the Population Ageing and Care Simulation (PACSim) model Age Ageing 47 2018 374 380 29370339 2 Tran P.B. Kazibwe J. Nikolaidis G.F. Linnosmaa I. Rijken M. van Olmen J. Costs of multimorbidity: a systematic review and meta-analyses BMC Med 20 2022 234 35850686 3 Serhan C.N. Brain S.D. Buckley C.D. Gilroy D.W. Haslett C. O'Neill L.A. Resolution of inflammation: state of the art, definitions and terms FASEB J 21 2007 325 332 17267386 4 Bannenberg G.L. Chiang N. Ariel A. Arita M. Tjonahen E. Gotlinger K.H. Molecular circuits of resolution: formation and actions of resolvins and protectins J Immunol 174 2005 4345 4355 15778399 5 Fullerton J.N. Gilroy D.W. Resolution of inflammation: a new therapeutic frontier Nat Rev Drug Discov 15 2016 551 567 27020098 6 Morioka S. Maueroder C. Ravichandran K.S. Living on the edge: efferocytosis at the interface of homeostasis and pathology Immunity 50 2019 1149 1162 31117011 7 De Maeyer R.P.H. van de Merwe R.C. Louie R. Bracken O.V. Devine O.P. Goldstein D.R. Blocking elevated p38 MAPK restores efferocytosis and inflammatory resolution in the elderly Nat Immunol 21 2020 615 625 32251403 8 Ampomah P.B. Cai B. Sukka S.R. Gerlach B.D. Yurdagul A. Jr. Wang X. Macrophages use apoptotic cell-derived methionine and DNMT3A during efferocytosis to promote tissue resolution Nat Metab 4 2022 444 457 35361955 9 Potter P.K. Cortes-Hernandez J. Quartier P. Botto M. Walport M.J. Lupus-prone mice have an abnormal response to thioglycolate and an impaired clearance of apoptotic cells J Immunol 170 2003 3223 3232 12626581 10 Feehan K.T. Gilroy D.W. Is resolution the end of inflammation? Trends Mol Med 25 2019 198 214 30795972
PMC010xxxxxx/PMC10292661.txt	==== Front Sleep Health Sleep Health Sleep Health 2352-7218 2352-7226 National Sleep Foundation. Published by Elsevier Inc. S2352-7218(23)00087-6 10.1016/j.sleh.2023.04.008 Article Insomnia symptoms among older adults during the first year of the COVID-19 pandemic: A longitudinal study Gong Kirsten MA ab Garneau James BSc ab Grenier Sébastien PhD bc Vasiliadis Helen-Maria PhD d Dang-Vu Thien Thanh MD, PhD be Dialahy Isaora Zefania PhD b Gouin Jean-Philippe PhD ab⁎ a Department of Psychology, Concordia University, Montreal, Canada b Centre de recherche de l’Institut universitaire de gériatrie de Montréal (CRIUGM), CIUSSS Centre-Sud-de-l’île-de-Montréal, Montreal, Canada c Department of Psychology, Université de Montréal, Montreal, Canada d Department of Community Health Sciences, Université de Sherbrooke, Longueuil, Canada e Department of Health, Kinesiology and Applied Physiology, Concordia University, Montreal, Canada ⁎ Corresponding author: Jean-Philippe Gouin, PhD, Concordia University, 7141, Sherbrooke St. West, PY170-14, Montreal, QC H4B 1R6, Canada. 26 6 2023 26 6 2023 3 10 2022 22 3 2023 28 4 2023 © 2023 National Sleep Foundation. Published by Elsevier Inc. All rights reserved. All rights reserved. 2023 National Sleep Foundation Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objective To identify sociodemographic, psychological, and health factors related to trajectories of insomnia symptoms in older adults during the COVID-19 pandemic. Methods From May 2020 to May 2021, 644 older adults (mean age = 78.73, SD = 5.60) completed telephone-administered self-reported measures (ie, Insomnia Severity Index, consensus sleep diaries, UCLA Loneliness Scale, Kessler Psychological Distress Scale, Post-Traumatic Checklist, perceived health threat, and International Physical Activity Questionnaire) and provided sociodemographic data at 4 timepoints. Using the Insomnia Severity Index score at each timepoint, group-based trajectory modeling was conducted to identify groups with distinct insomnia trajectories. Results On average, there was no significant change in insomnia symptoms over time. Three groups with distinct sleep trajectories were identified: clinical (11.8%), subthreshold (25.3%), and good sleepers (62.9%). Older adults who were younger, male, had elevated psychological distress and posttraumatic stress disorder symptoms, perceived more SARS-CoV-2 health threat, spent more time in bed, and had shorter sleep duration during the first wave of the pandemic were more likely to belong to the clinical than to the good sleepers group. Those who were younger, female, had elevated psychological distress and PTSD symptoms, greater loneliness, spent more time in bed, and had reduced sleep duration during the first wave were more likely to belong to the subthreshold than to the good sleepers group. Conclusions Over 1 in 3 older adults experienced persistent subthreshold or clinically significant insomnia symptoms. Both sleep-related behaviors as well as general and COVID-19-related psychological factors were associated with insomnia trajectories. Keywords SARS-CoV-2 Sleep Aging Group-based trajectory modeling ==== Body pmcIntroduction At the beginning of the coronavirus 19 (COVID-19) pandemic in March 2020, strict lockdown and physical distancing directives were imposed around the globe to mitigate the spread of the virus, protect vulnerable populations, and reduce pressure on healthcare systems. While infection from the SARS-CoV-2 virus was widespread across all demographics, older adults were at greater risk for COVID-19-related morbidity and mortality. Although stringent confinement measures can help reduce the spread of the virus within the population, they can also impact the psychological well-being and sleep of the general population, especially older adults.1 Meta-analyses of cross-sectional studies during the first wave of the COVID-19 pandemic suggest that 18.0%-36.7% of the general population reported significant sleep disturbances.2, 3 Using retrospective self-report data, older adults reported greater insomnia severity, poorer sleep quality, longer sleep onset latency, increased sleep duration, greater time in bed, lower sleep efficiency, more daytime impairment, and more frequent use of sleep medications during the pandemic compared to before the pandemic.4, 5 However, longitudinal studies have found mixed results when comparing sleep quality and insomnia severity between the prepandemic period and the first wave of the pandemic. A longitudinal study using prepandemic data from a population-based cohort of 594 middle-aged adults found that the prevalence of insomnia symptoms increased from 25.4% to 32.2% and the prevalence of insomnia disorder increased from 16.8% to 19% during the first wave of the pandemic.6 Another longitudinal study indicated that 10.2% of healthcare providers with insomnia prior to the COVID-19 pandemic no longer had the disorder during the first wave of the pandemic, whereas 43.4% without insomnia prior to the pandemic subsequently developed insomnia.7 However, these results conflict with other longitudinal findings reflecting no change in sleep quality or prevalence of insomnia in older adults during the first wave of the pandemic compared to the prepandemic period.8, 9, 10 Thus, the change in sleep quality and insomnia in older adults during the first wave of the COVID-19 pandemic remains unclear. Although the most stringent public health mitigation measures were generally put in place during the first wave of the pandemic, changing yet protracted sanitary measures were maintained in most countries during subsequent waves of the pandemic. Yet, far less information is available on insomnia symptoms during the later waves of the pandemic. Two Italian longitudinal studies of the general population examined sleep changes over the first and second waves of the pandemic, up until autumn 2020. One study found later bed and rise times, and longer time in bed and sleep latency among adults between the ages of 18 and 79 years in the first wave compared to the second wave, with poor sleep quality in both waves compared to retrospective data reflecting the period prior to each subsequent lockdown.11 The other study found similar results within the same age group, including earlier bed and rise times, decreased sleep latency, improved subjective sleep quality, and decreased insomnia severity during the second wave of the pandemic compared to the first wave.12 In addition, 1 German longitudinal study examined insomnia severity in 267 adults with a mean age of 31 over the course of 6 months, beginning in summer 2020 and ending in early winter 2021. These findings suggested an initial prevalence of clinically significant insomnia of 10.1% and subthreshold insomnia of 27.0%, with no statistically significant change in insomnia over the 6-month period.13 How insomnia symptoms changed over the first year of the pandemic remains unclear, with no research specifically examining older adults. During the first wave of the COVID-19 pandemic, cross-sectional research has identified certain sociodemographic factors, such as being female, lower education level, and greater financial strain, to be related to greater insomnia severity and poorer subjective sleep quality among older adults.14, 15, 16 Yet, evidence for the relationship between age itself and these adverse sleep outcomes in the context of the pandemic is still mixed.14, 16 Beyond age-related biological changes that can lead to earlier shifted sleep-wake cycles, more sleep fragmentation, and increased napping behavior among older adults,17 the COVID-19 pandemic has introduced additional factors that may negatively impact sleep in these individuals.14 In the early phase of the pandemic, older adults also reported disruptions in their social rhythm, described as a loss of regularity in the timing of daily behaviors, such as bed or wake-up times, socialization, eating, physical activity, and entertainment.18 Many older adults engaged in less physical activity during the pandemic compared to the prepandemic period, which may negatively impact their sleep quality.19 Furthermore, the COVID-19 confinement measures may have created a context of increased loneliness. Accordingly, older adults who reported greater feelings of loneliness during the first wave of the pandemic also experienced worse sleep quality and increased insomnia symptoms compared to the prepandemic period.15, 18 In addition, given that older adults are at greater risk for severe COVID-19 outcomes,1 the pandemic was found to be associated with worse anxiety, depression, and posttraumatic stress symptoms, which may also be associated with a higher risk for insomnia symptoms.20, 21 Thus, the current research includes these potential predictors of insomnia throughout the later stages of the pandemic to shed light on the effect of these factors over time. Objectives and hypotheses The present study examined changes in insomnia symptoms during the first year of the COVID-19 pandemic among older adults. In addition, sociodemographic, psychological, and health factors were examined to identify potential risk or protective factors for insomnia symptoms during the first year of the pandemic. Participants and methods To be eligible for this study, participants had to be at least 60 years old and had to speak French or English. They also needed to be able to complete telephone-based semistructured interviews. They were excluded if they had hearing impairment making them unable to provide oral informed consent, or severe cognitive impairment, as assessed by a cutoff score of 14 points on the telephone-based Mini-Mental State Examination.22 While 1377 older adults were contacted to participate in this study, 644 participants were enrolled in the study (participation rate: 46.8%). Participants were recruited from ongoing cohort studies and participant pools from the Centre de Recherche de l’Institut de Gériatrie de Montréal (CRIUGM) or l’Étude sur la Santé des Aînés (ESA)-Service studies. At the second timepoint, 63 additional participants were recruited through newspaper and radio advertisements. Study design and procedure A 4-wave longitudinal study was carried out from May 6, 2020 to May 24, 2021, which covered 3 waves of the COVID-19 pandemic. Data were collected at 4 timepoints: Spring 2020 (Time 1), Summer 2020 (Time 2), Fall 2020 (Time 3), and Spring 2021 (Time 4). Participants were undergoing stricter confinement and physical distancing measures during the Time 1 and Time 4 assessments. Confinement measures were relaxed during the Time 2 assessment and gradually increased during the Time 3 assessment. The local infection rate increased over time in Quebec (roughly 50 new cases/day at T2 to around 320 new cases/day at T4). The study received ethical approval from the Centre intégré universitaire de santé et de services sociaux (CIUSSS) du Centre-Sud-de-l’Île-de-Montréal Ethics Review Board. All participants provided verbal informed consent prior to the start of the study. Measures Sleep variables The Insomnia Severity Index (ISI) measures the severity of insomnia symptoms within the last 2 weeks. It is composed of 7 items on a 5-point Likert scale (0 = no problem to 4 = very severe problem). A higher overall score signifies greater insomnia symptoms.23 The internal consistency of the ISI scores within the sample was .86. Retrospective sleep diaries adapted from the consensus sleep diary24 assessed the average duration of nap time, total sleep time, and total time in bed in the last 2 weeks. To examine the stability of social rhythms, participants provided information about the earliest and latest wake-up times in the past 2 weeks. The difference between the earliest and latest wake-up times was computed to measure the stability of social rhythm, with greater scores indicating a more unstable rhythm.25 Sociodemographic variables Sociodemographic characteristics consisted of age, gender, education level, and income level. Participants with an annual income below $25 000 were considered to live in poverty.26 Participants also indicated if they lived alone or with others. Psychological and health variables A 3-item UCLA Loneliness Scale evaluated perceived loneliness within the last 2 weeks. The items are rated on a 3-point Likert scale (1 = hardly ever to 3 = often). A higher total score reflects greater loneliness.27 The internal consistency of the UCLA Loneliness Scale scores within the sample was .77. Kessler Psychological Distress Scale 6-item version (K6) assessed psychological distress in the past 2 weeks. Each item is rated on a 5-point Likert scale (1 = none of the time to 5 = all of the time). A higher total score indicates more psychological distress.28 The internal consistency of the K6 scores within the sample was .84. The Post-Traumatic Checklist – Civilian Version (PCL-C) measured PTSD symptoms associated with the COVID-19 pandemic in the last 2 weeks. This scale consists of 17 items on a 5-point Likert scale (0 = not at all to 4 = very often). A higher total score reflects greater severity of PTSD symptoms, with a cutoff score of 30 or above suggesting the presence of probable PTSD.29 The internal consistency of the PCL-C scores within the sample was .83. The perceived health threat of COVID-19 was assessed based on the health belief model using 4 items evaluating the extent to which they believed that themselves or their close others were susceptible to getting infected by the virus and how dangerous the virus could be to them or their closer others on a 5-point Likert scale (0 = not at all to 4 = extremely). A higher total score indicated a greater perceived threat. The internal consistency of the score was .70. International Physical Activity Questionnaire (IPAQ) – short form assessed physical activity in the past week. Walking, moderate-intensity, and vigorous activity times were evaluated. The metabolic equivalent of task for the combination of all of these activities was calculated.30 Statistical analysis Hierarchical linear modeling (HLM) evaluated whether the severity of insomnia symptoms changed, on average, across the 4 timepoints within our sample. Group-based trajectory modeling (GBTM) was then conducted to identify groups of older adults who exhibited distinct patterns of changes in insomnia symptoms throughout the pandemic. Unlike growth-curve modeling techniques that assume that each individual’s trajectory evolves around 1 population mean, GBTM assumes that the population is made up of multiple groups of individuals who commonly evolve together over time. The purpose of GBTM is to decrease the within-group variability in trajectories while also increasing the between-group variability such that the trajectory for each group will be distinct.31 To select the number of groups, we examined the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), cross-validation error (CVE), and the overall group size (at least 50 individuals per group). The model with lower BIC and CVE values, reflecting better model fit, was selected. Maximum likelihood was used to estimate the probability that each individual belonged to each subgroup and to allocate each individual to the most probable subgroup. The final model included 3 subgroups with a quadratic trajectory. All participants with at least 1 timepoint were included in the modeling of the group-based trajectories. GBTM was conducted using the software package, “crimCV,” in R. Subsequently, multinomial logistic regression was performed to evaluate which sociodemographic, psychological, and behavioral factors at T1 independently predicted group membership using the “nnet” package in R. All variables were entered simultaneously in the model in order to examine their independent effects adjusting for all other variables. The odds ratio (OR) of belonging to a particular trajectory group relative to a reference group, the good sleepers group, was estimated for each predictor. An alpha was set to .05 for these analyses. Results Overall, the longitudinal study consisted of 644 older adults with a mean age of 78.73 (SD = 5.60 and age range: 64.20-97.05). In total, 537 participants (83.39%) completed T1, 454 participants (70.50%) completed T2, 429 participants (66.61%) completed T3, and 395 participants (61.34%) completed T4. In addition, 259 (40.22%) participants had data for all 4 timepoints. Sociodemographic, psychological, and health characteristics of the sample at the start of the pandemic are presented in Table 1. Out of all participants who shared their COVID-19 infection status at T1, none of them obtained a positive COVID-19 test result. However, 0.65%, 0.24%, and 6.27% of participants self-reported that they tested positive for COVID-19 at T2, T3, and T4, respectively.Table 1 Sociodemographic, psychological, and health characteristics of the sample at T1 Table 1Characteristic M (SD)/% Mean age (years) 78.73 (5.60) <65 0.33 65-74 28.78 75-84 57.07 85> 13.82 Gender Male 26.93 Female 73.07 Highest education level None/Primary school 7.51 Secondary school 24.21 College 22.54 University 45.74 Annual income <$25,000 25.26 $25,000+ 74.74 Living situation Lives alone 54.25 Lives with others 45.75 Total UCLA Loneliness Scale score 4.60 (1.73) Not lonely (3-5) 71.40 Lonely (6-9) 28.60 Total Kessler Psychological Distress Scale score 9.39 (3.25) Likely to be well (10-19) 79.07 Likely to have a mild disorder (20-24) 14.53 Likely to have a moderate disorder (25-29) 3.98 Likely to have a severe disorder (30-50) 2.42 Total Post-Traumatic Checklist score 7.79 (7.35) No PTSD (0-29) 98.53 PTSD (30-68) 1.47 Perceived health threat of COVID-19 6.83 (2.90) Total International Physical Activity Questionnaire score (MET-minutes/week) 2089.04 (1936.00) Low (≤ 599) 36.59 Moderate (600-2999) 42.79 High (≥3000) 20.62 Total Insomnia Severity Index score 4.17 (4.57) No clinically significant insomnia (0-7) 79.52 Subthreshold insomnia (8-14) 16.76 Moderate-severe insomnia (15-28) 3.73 Nap time (minutes/day) 23.13 (37.78) <15 55.35 15-29.99 12.95 30-44.99 11.07 45-59.99 3.00 ≥ 60 17.64 Sleep duration (hours/night) 7.42 (1.41) <5 3.55 5-5.99 8.22 6-6.99 13.08 ≥ 7 75.14 Time in bed (hours/night) 8.34 (1.36) <5 0.19 5-6.99 12.38 7-8.99 49.34 ≥ 9 38.09 Social rhythm (minutes)a 114.39 (80.87) <30 8.08 30-59.99 10.90 60-89.99 17.11 90-119.99 15.98 ≥ 120 47.93 Note. % represents the valid percentage. a Social rhythm scores were computed as the difference between the earliest and latest wake-up times within the past 2 weeks that the questionnaire was administered. The HLM model indicated that there was no statistically significant change in the severity of insomnia symptoms across time, β 1 = 0.10, SD = 0.07, t = 1.42, and P = .16. Furthermore, GBTM identified 3 groups with distinct sleep trajectories: clinical (G1), subthreshold (G2), and good sleepers (G3) groups. As shown in Fig. 1, the clinical group (11.78% of the sample) had stably higher total ISI scores than the other 2 groups between T1 and T4. For the subthreshold group (25.27% of the sample), the total ISI scores were fairly stable and resided within the subthreshold range of clinical insomnia between T1 and T3 but exhibited a slight decrease at T4. Finally, the good sleepers group (62.95% of the sample) had stably low ISI scores in total across time.Fig. 1 Insomnia severity group trajectories throughout the COVID-19 pandemic. Notes: N = 644 Figure 1 Sociodemographic, psychological, and health factors associated with group membership were identified ( Table 2). Older adults who were younger, male, had elevated psychological distress and PTSD symptoms, perceived more SARS-CoV-2 health threats, spent more time in bed, and had shorter sleep duration at Time 1 were more likely to belong to the clinical group than to the good sleepers group. Those who were younger, female, had elevated psychological distress and PTSD symptoms, greater loneliness, spent more time in bed, and had reduced sleep duration at Time 1 were more likely to be a part of the subthreshold group as opposed to the good sleepers group. Education level, annual income, living situation, physical activity, nap duration, and variability in wake time were not independently associated with group membership in these multivariable analyses.Table 2 Multinomial logistic regression Table 2 Clinical group Subthreshold group Predictor OR 95% CI OR 95% CI Age 0.65a [0.45, 0.86] 0.70b [0.56, 0.84] Gender (ref = male) 0.20c [0.08, 0.32] 2.10b [1.25, 2.95] Education (ref = no university) 1.66 [0.76, 2.56] 1.04 [0.69, 1.39] Annual income (ref = no poverty) 2.31 [0.98, 3.65] 0.95 [0.58, 1.32] Living situation (ref = not living alone) 0.63 [0.28, 0.98] 1.24 [0.81, 1.68] UCLA Loneliness Scale score at T1 1.36 [0.93, 1.79] 1.44c [1.17, 1.71] Kessler Psychological Distress Scale score at T1 1.46a [1.03, 1.90] 1.40c [1.15, 1.66] Post-Traumatic Checklist score at T1 3.67c [2.44, 4.89] 2.51c [1.96, 3.07] Perceived health threat of COVID-19 at T1 1.59a [1.08, 2.10] 1.18 [0.95, 1.42] International Physical Activity Questionnaire score at T1 0.83 [0.56, 1.10] 0.98 [0.80, 1.15] Nap time at T1 1.13 [0.82, 1.44] 1.16 [0.96, 1.37] Sleep duration at T1 0.02c [0.01, 0.03] 0.10c [0.07, 0.12] Time in bed at T1 6.04c [3.96, 8.12] 3.89c [2.85, 4.93] Social rhythm at T1 1.15 [0.82, 1.49] 1.08 [0.89, 1.26] Note. N = 644; ref refers to reference categories for the respective predictors; T1 refers to the first wave of the pandemic. All variables were entered simultaneously in the statistical models. For the calculation of the odd ratios, the good sleepers group served as the reference group. a P < .05; b P < .01; c P < .001. Discussion The current study investigated changes in insomnia symptoms in older individuals during the COVID-19 pandemic. The average insomnia severity within this sample remained stable during the first year of the COVID-19 pandemic. Group-based trajectory identified 3 trajectory groups with varying degrees of insomnia symptoms. Membership to the clinical group was associated with younger age, male gender, elevated psychological distress, more posttraumatic stress symptoms, greater perceived health threat, shorter sleep duration, and longer time spent in bed at Time 1 compared to the good sleepers group. Membership to the subthreshold group was predicted by younger age, female gender, elevated psychological distress, more posttraumatic stress symptoms, more loneliness, shorter sleep duration, and more time spent in bed at T1 compared to the good sleepers. Thus, both sleep and psychological factors were associated with group membership. In the present sample, insomnia symptoms of older adults were stable during the first year of the COVID-19 pandemic, consistent with results from a longitudinal study in middle-aged individuals.32 The clinical group, comprising about 11.8% of the sample, exhibited higher levels of insomnia symptoms throughout the follow-up compared to the rest of the sample. The proportion of individuals meeting the clinical threshold for clinical insomnia was somewhat lower than the 15.6% prevalence of clinical insomnia symptoms observed in a meta-analysis of cross-sectional studies with older adults in China during the pandemic.33 Furthermore, it is important to note that not all individuals in the clinical group reached the clinical cutoff for insomnia (ISI ≥15), and our sample had a relatively low percentage of severe insomnia (3.73% prevalence) compared to the 13.0% prevalence of severe insomnia in older adults from Sweden prior to the pandemic,34 suggesting possible sampling biases. Similarly, our “subthreshold” group did not reach the ISI>8 cutoff at all timepoints. The subthreshold group accounted for an additional 25.3% of older adults, suggesting that a total of 37.1% of the participants reported some level of insomnia symptoms that may be negatively impacting their well-being. This is consistent with the 36.9% prevalence of subthreshold and clinical insomnia reported before the pandemic in older adults from Sweden but higher than the 22.2% prevalence among those from Canada.34 , 35. The factors associated with the largest odds ratio of membership to the clinical and subthreshold groups were time in bed and sleep durations at Time 1. During the first wave of the COVID-19 pandemic, these older adults spent more time lying in bed while reporting shorter sleep duration than the good sleepers. This greater time in bed combined with lower sleep duration reflects poor sleep efficiency. In one study, older adults maintained their usual sleep schedule during the pandemic. Here, it could be that, although their sleep duration remained unchanged from before the pandemic, their sleep efficiency decreased by spending more time in bed than usual.36 However, given the lack of prepandemic data, it is not clear if the time spent in bed increased during the pandemic. Regardless, the extensive time in bed observed in both the clinical and subthreshold groups has been proposed to be an important factor perpetuating insomnia symptoms. Remaining in bed while awake for long periods may promote conditioned arousal, where the bed environment loses its learned association with sleep and becomes associated with wake or negative feelings.37 The second largest odds ratio of membership to the clinical and subthreshold groups consisted of elevated COVID-19-related PTSD symptoms at Time 1. This is in line with previous findings that greater PTSD symptoms were associated with poorer sleep quality and increased frequency of early awakenings during the early phase of the pandemic.21 Hyperarousal, the heightened emotional, cognitive, and physiological activation associated with acute stress symptoms, may interfere with sleep initiation or maintenance, or lead to restless sleep.38 Similarly, higher psychological distress was a significant predictor of membership to the clinical and subthreshold groups. Greater perceived health threat was a significant predictor of membership to the clinical group, whereas loneliness was a significant predictor of membership to the subthreshold group. These results are consistent with previous findings, suggesting that psychological distress20 and perceived loneliness15, 18 during the early stages of the COVID-19 pandemic were associated with higher insomnia severity and worse sleep quality. The lack of social relationships may also enhance the association between psychological distress and insomnia among older adults.39 In addition, it was previously shown that older adults who were more concerned about themselves or their kin contracting COVID-19 reported more insomnia symptoms.15 Within this cohort of older adults, individuals who were younger were more likely to have subthreshold or clinical levels of insomnia symptoms throughout the pandemic compared to the good sleepers group. One study found that increased age was associated with poorer sleep quality and greater insomnia symptoms.16 However, other studies suggest that older individuals presented with better psychological adaptation and less insomnia during the pandemic than younger individuals.14, 40 We also found that females were more likely to have subthreshold levels of insomnia symptoms. However, males were more likely to belong to the clinical group. This contradicts what has previously been reported in past literature.12, 14, 36 One explanation could be that older females who tend to have more social support than males were better equipped to cope with confinement-related stress.41 Although nap duration and annual income were positively associated with insomnia severity in bivariate analyses, they were not independent predictors in the multivariable models. They might then be associated with insomnia severity through their shared associations with other variables included in the model. In the present study, education level, living situation, variability in wake-up time, and physical activity were not associated with insomnia severity. The association between these factors and insomnia may be attenuated among older adults in the context of a pandemic with protracted confinement measures. Alternatively, sampling bias may account for some of the discrepancies with other findings observed in the literature. The most notable strength of this study is its longitudinal design, which enabled us to examine patterns of sleep changes among older adults across a 12-month period that covered 3 waves of the COVID-19 pandemic. Interpretation of these findings must be done in light of the study limitations. First, without prepandemic insomnia data, it is impossible to ascertain whether the current insomnia symptoms were triggered or intensified by the pandemic or represented preexisting issues. Furthermore, the generalizability of our findings is limited by the fact that our convenience sample consisted predominantly of white and female participants who were well-educated. Likely due to these sampling biases, the prevalence of clinical insomnia at Time 1 was lower than in other population-based studies, highlighting the importance of replicating these findings in more representative samples.42 Moreover, our sleep diary assessment was retrospective and, therefore, prone to recall biases. Finally, we did not account for any medication use in our analyses because precise information on this matter was not collected. Despite these limitations, the present results highlight that more than a third of older adults displayed persistent insomnia symptoms at a clinical or subclinical level throughout the first year of the pandemic. Many of these older adults would likely benefit from sleep-focused intervention. Cognitive-behavioral therapy for insomnia (CBTi) has been shown to be effective for older adults with insomnia and comorbid disorders.43 Accordingly, nearly all factors associated with greater insomnia severity in the current study can be addressed with different components of CBTi. For instance, the cognitive component of CBTi can target PTSD symptoms and psychological distress, whereby excessive worrying can be addressed through Socratic questioning and behavioral experiments. Notably, digital CBTi has been shown to be effective in improving both insomnia and depressive symptoms at a 1-year follow-up in older adults, providing a scalable intervention in the context of a pandemic.44 However, other forms of CBT may be needed to treat distress associated with the pandemic, for example, PTSD-related symptoms. In conclusion, this study revealed that nearly 12% of older adults maintained higher insomnia symptoms than the rest of the sample over the first year of the COVID-19 pandemic with an additional 25% experiencing less, yet subthreshold, symptoms of insomnia that may still be negatively impacting their well-being and daily functioning. These results can help identify a subgroup of older adults with subthreshold or clinically significant levels of insomnia symptoms that would benefit from psychological intervention targeting sleep behaviors and cognitions. Future research should examine the trajectories of insomnia severity in the (post)pandemic era and investigate the effects of sleep-related interventions to improve sleep health among older adults and improve their postpandemic quality of life and functioning. Declaration of conflicts of interest There are no conflicts of interest to declare. Funding This study was supported by the Centre de recherche de l’Institut universitaire de gériatrie de Montréal and the 10.13039/100013228 Quebec Network for Research on Aging , a thematic network supported by the Fonds de Recherche du Québec – Santé, and a Canadian Institutes of Health Research Project Grant. Acknowledgments We would like to acknowledge Sara Matovic who contributed to data collection, cleaning, and preprocessing. ==== Refs References 1 Brooke J. Jackson D. Older people and COVID-19: Isolation, risk and ageism J Clin Nurs 29 13–14 2020 2044 2046 10.1111/jocn.15274 32239784 2 Alimoradi Z. Broström A. Tsang H.W.H. Sleep problems during COVID-19 pandemic and its’ association to psychological distress: a systematic review and meta-analysis EClinicalMedicine 36 2021 100916 10.1016/j.eclinm.2021.100916 3 Jahrami H.A. Alhaj O.A. Humood A.M. Sleep disturbances during the COVID-19 pandemic: a systematic review, meta-analysis, and meta-regression Sleep Med Rev 62 2022 101591 10.1016/j.smrv.2022.101591 4 Kocevska D. Blanken T.F. Van Someren E.J.W. Rösler L. Sleep quality during the COVID-19 pandemic: not one size fits all Sleep Med 76 2020 86 88 10.1016/j.sleep.2020.09.029 33126035 5 Trabelsi K. Ammar A. Masmoudi L. Sleep quality and physical activity as predictors of mental wellbeing variance in older adults during COVID-19 lockdown: ECLB COVID-19 international online survey Int J Environ Res Public Health 18 8 2021 1 18 10.3390/ijerph18084329 6 Morin C.M. Vézina-Im L.A. Ivers H. Prevalent, incident, and persistent insomnia in a population-based cohort tested before (2018) and during the first-wave of COVID-19 pandemic (2020) Sleep 45 1 2022 1 6 10.1093/sleep/zsab258 7 McCall W.V. Mensah-Bonsu D. Withers A.E. Gibson R.W. Short-term insomnia disorder in health care workers in an academic medical center before and during COVID-19: rates and predictive factors J Clin Sleep Med 17 4 2021 749 755 10.5664/jcsm.9034 33226333 8 Chao A.M. Wadden T.A. Clark J.M. Changes in the prevalence of symptoms of depression, loneliness, and insomnia in US older adults with type 2 diabetes during the COVID-19 pandemic: the Look AHEAD Study Diabetes Care 45 1 2022 74 82 10.2337/dc21-1179 34753805 9 Kunzler A.M. Röthke N. Günthner L. Mental burden and its risk and protective factors during the early phase of the SARS-CoV-2 pandemic: systematic review and meta-analyses Global Health 17 1 2021 34 10.1186/s12992-021-00670-y 33781283 10 Okely J.A. Corley J. Welstead M. Change in physical activity, sleep quality, and psychosocial variables during COVID-19 lockdown: evidence from the Lothian Birth Cohort 1936 Int J Environ Res Public Health 18 1 2020 210 10.3390/ijerph18010210 33396611 11 Conte F. Cellini N. De Rosa O. Dissociated profiles of sleep timing and sleep quality changes across the first and second wave of the COVID-19 pandemic J Psychiatr Res 143 2021 222 229 10.1016/j.jpsychires.2021.09.025 34507103 12 Salfi F. D'Atri A. Tempesta D. Ferrara M. Sleeping under the waves: a longitudinal study across the contagion peaks of the COVID-19 pandemic in Italy J Sleep Res 30 5 2021 e13313 10.1111/jsr.13313 13 Werner G.G. Cludius B. Sckopke P. Stefan A. Schönbrodt F. Zygar-Hoffmann C. The predictive power of insomnia symptoms on other aspects of mental health during the COVID-19 pandemic: a longitudinal study J Sleep Res 2022 e13641 10.1111/jsr.13641 14 Mandelkorn U. Genzer S. Choshen-Hillel S. Escalation of sleep disturbances amid the COVID-19 pandemic: a cross-sectional international study J Clin Sleep Med 17 1 2021 45 53 10.5664/jcsm.8800 32900428 15 Polenick C.A. Daniel N.R. Perbix E.A. Factors associated with sleep disturbances related to the COVID-19 pandemic among older adults with chronic conditions Am J Geriatr Psychiatry 29 11 2021 1160 1165 10.1016/j.jagp.2021.03.003 33867223 16 Salfi F. Lauriola M. D’Atri A. Demographic, psychological, chronobiological, and work-related predictors of sleep disturbances during the COVID-19 lockdown in Italy Sci Rep 11 2021 11416 10.1038/s41598-021-90993-y 34075173 17 Ancoli-Israel S. Martin J.L. Insomnia and daytime napping in older adults J Clin Sleep Med 2 3 2006 333 342 10.5664/jcsm.26597 17561549 18 Murray G. Gottlieb J. Swartz H.A. Maintaining daily routines to stabilize mood: theory, data, and potential intervention for circadian consequences of COVID-19 Can J Psychiatry 66 1 2021 9 13 10.1177/0706743720957825 32909832 19 Bailey L. Ward M. DiCosimo A. Physical and mental health of older people while cocooning during the COVID-19 pandemic Q J Med 114 9 2021 648 653 10.1093/qjmed/hcab015 20 Cipriani G.E. Bartoli M. Amanzio M. Are sleep problems related to psychological distress in healthy aging during the COVID-19 pandemic? A review Int J Environ Res Public Health 18 20 2021 10676 10.3390/ijerph182010676 34682423 21 Liu N. Zhang F. Wei C. Prevalence and predictors of PTSS during COVID-19 outbreak in China hardest-hit areas: gender differences matter Psychiatry Res 287 2020 112921 10.1016/j.psychres.2020.112921 22 Newkirk L.A. Kim J.M. Thompson J.M. Validation of a 26-point telephone version of the mini-mental state examination J Geriatr Psychiatry Neurol 17 2 2004 81 87 10.1177/0891988704264534 15157348 23 Morin C.M. Belleville G. Bélanger L. Ivers H. The Insomnia Severity Index: psychometric indicators to detect insomnia cases and evaluate treatment response Sleep 34 5 2011 601 608 10.1093/sleep/34.5.601 21532953 24 Carney C.E. Buysse D.J. Ancoli-Israel S. The consensus sleep diary: standardizing prospective sleep self-monitoring Sleep 35 2 2012 287 302 10.5665/sleep.1642 22294820 25 Monk T.H. Flaherty J.F. Frank E. Hoskinson K. Kupfer D.J. The social rhythm metric. An instrument to quantify the daily rhythms of life J Nerv Ment Dis 178 2 1990 120 126 10.1097/00005053-199002000-00007 2299336 26 Statistics Canada. Table 11-10-0241-01 Low income cut-offs (LICOs) before and after tax by community size and family size, in current dollars. 2022. https://doi.org/10.25318.1110024101-eng. 27 Hughes M.E. Waite L.J. Hawkley L.C. Cacioppo J.T. A short scale for measuring loneliness in large surveys: results from two population-based studies Res Aging 26 6 2004 655 672 10.1177/0164027504268574 18504506 28 Kessler R.C. Barker P.R. Colpe L.J. Screening for serious mental illness in the general population Arch Gen Psychiatry 60 2 2003 184 189 10.1001/archpsyc.60.2.184 12578436 29 Weathers F.W. Litz B.T. Huska J.A. Keane T.M. PTSD Checklist – Civilian Version 1994 National Center for PTSD 30 Craig C.L. Marshall A.L. Sjöström M. International Physical Activity Questionnaire: 12-country reliability and validity Med Sci Sports Exerc 35 8 2003 1381 1395 10.1249/01.MSS.0000078924.61453.FB 12900694 31 Nagin D.S. Odgers C.L. Group-based trajectory modeling in clinical research Annu Rev Clin Psychol 6 2010 109 138 10.1146/annurev.clinpsy.121208.131413 20192788 32 Werner G.G. Cludius B. Sckopke P. Stefan A. Schönbrodt F. Zygar-Hoffmann C. The predictive power of insomnia symptoms on other aspects of mental health during the COVID-19 pandemic: a longitudinal study J Sleep Res 32 1 2022 e13641 10.1111/jsr.13641 33 Zhang Q.Q. Li L. Zhong B.L. Prevalence of insomnia symptoms in older Chinese adults during the COVID-19 pandemic: a meta-analysis Front Med 8 2021 779914 10.3389/fmed.2021.779914 34 Dragioti E. Levin L.Å. Bernfort L. Larsson B. Gerdle B. Insomnia severity and its relationship with demographics, pain features, anxiety, and depression in older adults with and without pain: cross-sectional population-based results from the PainS65+ cohort Ann Gen Psychiatry 16 2017 15 10.1186/s12991-017-0137-3 28250802 35 Chaput J.P. Yau J. Rao D.P. Morin C.M. Prevalence of insomnia for Canadians aged 6 to 79 Health Rep 29 12 2018 16 20 Available at: 〈https://www150.statcan.gc.ca/n1/pub/82-003-x/2018012/article/00002-eng.pdf〉 30566205 36 de Almondes K.M. Castro E.A.S. Paiva T. Sleep habits, quality of life and psychosocial aspects in the older age: before and during COVID-19 Front Neurosci 16 2022 694894 10.3389/fnins.2022.694894 37 Spielman A.J. Caruso L.S. Glovinsky P.B. A behavioral perspective on insomnia treatment Psychiatr Clin North Am 10 4 1987 541 553 10.1016/S0193-953X(18)30532-X 3332317 38 Riemann D. Spiegelhalder K. Feige B. The hyperarousal model of insomnia: a review of the concept and its evidence Sleep Med Rev 14 1 2010 19 31 10.1016/j.smrv.2009.04.002 19481481 39 Zhang C. Xiao S. Lin H. The association between sleep quality and psychological distress among older Chinese adults: a moderated mediation model BMC Geriatr 22 1 2022 35 10.1186/s12877-021-02711-y 35012479 40 World Health Organization. Mental health and COVID-19: early evidence of the pandemic’s impact: scientific brief; 2022. Available at: https://www.who.int/publications/i/item/WHO-2019-nCoV-Sci_Brief-Mental_health-2022.1. 41 Liu C. Liu D. Huang N. The combined impact of gender and age on post-traumatic stress symptoms, depression, and insomnia during COVID-19 outbreak in China Front Public Health 8 2021 620023 10.3389/fpubh.2020.620023 42 Morin C.M. Vézina-Im L.A. Ivers H. Prevalent, incident, and persistent insomnia in a population-based cohort tested before (2018) and during the first-wave of COVID-19 pandemic (2020) Sleep 45 1 2022 1 6 10.1093/sleep/zsab258 43 Hertenstein E. Trinca E. Wunderlin M. Cognitive behavioral therapy for insomnia in patients with mental disorders and comorbid insomnia: a systematic review and meta-analysis Sleep Med Rev 62 2022 101597 10.1016/j.smrv.2022.101597 44 Cheng P. Kalmbach D.A. Hsieh H.F. Castelan A.C. Sagong C. Drake C.L. Improved resilience following digital cognitive behavioral therapy for insomnia protects against insomnia and depression one year later Psychol Med 2022 1 11 10.1017/S0033291722000472
PMC010xxxxxx/PMC10292668.txt	==== Front Biocybern Biomed Eng Biocybern Biomed Eng Biocybernetics and Biomedical Engineering 0208-5216 2391-467X The Author(s). Published by Elsevier B.V. on behalf of Nalecz Institute of Biocybernetics and Biomedical Engineering of the Polish Academy of Sciences. S0208-5216(23)00037-2 10.1016/j.bbe.2023.06.003 Original Research Article Detection of various lung diseases including COVID-19 using extreme learning machine algorithm based on the features extracted from a lightweight CNN architecture Nahiduzzaman Md. a Faruq Goni Md Omaer a Robiul Islam Md. a Sayeed Abu b Shamim Anower Md. c Ahsan Mominul d Haider Julfikar e Kowalski Marcin f⁎ a Department of Electrical & Computer Engineering, Rajshahi University of Engineering & Technology, Rajshahi 6204, Bangladesh b Department of Computer Science & Engineering, Rajshahi University of Engineering & Technology, Rajshahi 6204, Bangladesh c Department of Electrical & Electronic Engineering, Rajshahi University of Engineering & Technology, Rajshahi 6204, Bangladesh d Department of Computer Science, University of York, Deramore Lane, Heslington, York YO10 5GH, UK e Department of Engineering, Manchester Metropolitan University, Chester St, Manchester M1 5GD, UK f Institute of Optoelectronics, Military University of Technology, Gen. S. Kaliskiego 2, Warsaw, Poland ⁎ Corresponding author at: Military University of Technology, Warsaw, Poland. 26 6 2023 26 6 2023 18 12 2022 4 4 2023 16 6 2023 © 2023 The Author(s) 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Around the world, several lung diseases such as pneumonia, cardiomegaly, and tuberculosis (TB) contribute to severe illness, hospitalization or even death, particularly for elderly and medically vulnerable patients. In the last few decades, several new types of lung-related diseases have taken the lives of millions of people, and COVID-19 has taken almost 6.27 million lives. To fight against lung diseases, timely and correct diagnosis with appropriate treatment is crucial in the current COVID-19 pandemic. In this study, an intelligent recognition system for seven lung diseases has been proposed based on machine learning (ML) techniques to aid the medical experts. Chest X-ray (CXR) images of lung diseases were collected from several publicly available databases. A lightweight convolutional neural network (CNN) has been used to extract characteristic features from the raw pixel values of the CXR images. The best feature subset has been identified using the Pearson Correlation Coefficient (PCC). Finally, the extreme learning machine (ELM) has been used to perform the classification task to assist faster learning and reduced computational complexity. The proposed CNN-PCC-ELM model achieved an accuracy of 96.22% with an Area Under Curve (AUC) of 99.48% for eight class classification. The outcomes from the proposed model demonstrated better performance than the existing state-of-the-art (SOTA) models in the case of COVID-19, pneumonia, and tuberculosis detection in both binary and multiclass classifications. For eight class classification, the proposed model achieved precision, recall and fi-score and ROC are 100%, 99%, 100% and 99.99% respectively for COVID-19 detection demonstrating its robustness. Therefore, the proposed model has overshadowed the existing pioneering models to accurately differentiate COVID-19 from the other lung diseases that can assist the medical physicians in treating the patient effectively. Keywords COVID-19 Convolutional Neural Network Extreme Learning Machine Pearson Correlation Coefficient Pneumonia ==== Body pmc1 Introduction In the last few decades, several lung-related diseases have become an epidemic to humans, such as pneumonia, pleural, cardiomegaly, TB, and more recently the most life-threatening disease, COVID-19. Approximately 7% of the global population (450 million) is affected by pneumonia alone, and every year approximately 2 million people lose their lives due to the pneumonia [1]. In the past three years, almost 6.27 million people died due to COVID-19, and nearly 522 million people were affected [2]. In 2020, around 1.3 million people died due to TB; hence, in total 4 million people died due to lung-related disease every year [3]. These types of diseases are generally detected using CXR images of the lung by the radiologists as it is cheap and requires only fewer steps to detect these diseases. More than 35 million CXR images are taken every year in the US alone for medical treatment [4]. Many radiologists now have to assess more than 100 CXR images every day, resulting in an increased workload and fatigue and eventually wrong diagnosis [4]. Hence, the conventional process is very time-consuming and expensive. Furthermore, the diagnosis made by the radiologists may vary due to an error in human judgement. If COVID-19, pneumonia, TB, etc. are predicted at the early stages, it may save millions of human lives. Gradually, newer variants of COVID-19 for example, Delta, Omicron, etc. are spreading at a faster rate, hence it is too difficult for the medical physicians to cope up with the high demand for treatment on time. As a result, automated technologies that are trained to forecast the symptoms of a distinct lung-related anomalies based on a specific CXR image have the potential to aid the radiologists and medical physicians in accurate and faster diagnosis with a higher level of confidence. Hence, a computer aided intelligent system based on machine learning and deep learning has been designed to detect the lung related disease automatically based on the CXR images. In the past, several efforts have been made to automatically identify lung-related anomalies using the CXR images due to the recent availability of the larger data sets. Loey et al. developed a CNN model for extracting the prominent features and finally a Bayesian model was used to classify the COVID-19 patient [5]. Moreover, they used 3,616 COVID-19 data and achieved an accuracy of 96%. Bhattacharyya et al. utilized a pre-trained CheXNet model and trained this model using 1,326 images of COVID-19 to develop a COVID-CXNet model [6]. For preprocessing they used Contrast Limited Adaptive Histogram Equalization (CLAHE) and achieved an accuracy of 87.88%. Ieracitano et al. distinguished between CXR images of patients with idiopathic pneumonias unrelated to COVID-19 and those of patients with COVID-19 pneumonia by suggesting a fuzzy logic-based deep learning (DL) technique [7]. In addition, they used 121 images of COVID-19 pneumonia patient and their model's accuracy was approximately 81% and provided an explainable artificial intelligence method to aiding doctors. Agrawal and Choudhary developed a deep CNN to detect the COVID-19 and pneumonia from the CXR images [8]. They handled the class imbalance problem using the oversampling technique, namely Synthetic Minority Over-sampling Technique (SMOTE). Moreover, they considered two (Covid-19 Vs Normal) and three classes (COVID-19 Vs Pneumonia Vs Normal), while they used 1,525 samples of COVID-19 patients and achieved an accuracy of 96% and 94.45% for the two and three classes respectively. Gayathri et al. extracted features using pre-trained CNN models, for instance InceptionResnetV2, Resnet101, etc., reduced the dimensioanlity using sparse auto-encoder and finally used a feed forward neural network to detect the COVID-19 [9]. In this study, they used 504 COVID-19 images to train their models and achieved an accuracy of 95.78% and AUC of 98.21%. Kassani et al. compared different transfer learning (TL) models to detect the COVID-19 from CXR and computed tomography (CT) images [10]. Firstly, they used eight TL models: MobileNet, DenseNet, Xception, ResNet, InceptionV3, Inception- ResNetV2, VGGNet, and NASNet for extracting the features from the images and several ML models were developed. Finally, 117 CXR and 20 CT images of COVID-19 patients were used to train their models and achieved the highest accuracy of 99% using the Bagging tree classifier with DenseNet121 as a feature extractor. Yousefi et al. segmented the lung lobes using a 2D U-Net model to detect COVID-19 from CXR images [11]. The authors reduced dimensions using several techniques, for instance, Laplacian scoring and principal component analysis (PCA). They used 704 CXR images for training their model and achieved 89.6%, 72.6% accuracy for two-class, multiclass classifications, respectively. Akter et al. used various TL models such as VGG19, GoogLeNet, etc., to detect the COVID-19 from the CXR images [12]. The author used the augmentation technique to balance the datasets and used 52,000 CXR images to train their models and achieved a high accuracy of 98% using MobileNetV2 for binary classification. At the same time, the compilation time was 2 h, 50 min and 21 s. Chowdhury et al. used several pre-trained TL models to detect the viral and COVID-19 pneumonia from CXR images [13]. This study combined several datasets for training the models and achieved an accuracy of 99.7%, 97.9% for two and three-class classification, respectively. Horry et al. proposed a suitable CNN to detect the COVID-19 from multimodal imaging data [14]. They also removed the noise from images and achieved a precision of 86%, 100%, and 84% for CXR, ultrasound, and CT scans, respectively. Rasheed et al. proposed two classifiers: LR and CNN, to diagnose COVID-19 [15]. The generative adversarial network was used for data augmentation and PCA to select the most prominent features from 308 CXR images for training their model and achieved an accuracy of 97.6% using PCA. Panwar et al. utilized the VGG19 model to fast detect the COVID-19 from CXR and CT scan images [16]. Minaee et al. used 5000 CXR images to detect the COVID-19 using four TL models and achieved a specificity of 92.9% [17]. Afshar et al. developed capsule networks named COVID-CAPS to detect COVID-19 from 13,975 CXR images and achieved an accuracy of 95.7% [18]. Abbas et al. developed a Decompose, Transfer, and Compose (DeTraC) CNN model to classify the COVID-19 and achieved an accuracy of 93.1% [19]. Arias-Londono et al. used more than 79,500 CXR images for training their CNN model to detect COVID-19 and achieved an accuracy of 91.5% [20]. Alam et al. fused histogram-oriented gradient (HOG) and CNN for extracting the features from CXR images to detect COVID-19 [21]. Pandit et al. proposed the VGG19 TL model to detect COVID-19 from CXR images [22]. They used 1428 images for training their model and achieved an accuracy of 92.53%, and 96% for two- (COVID-19 vs Normal) and three-class (Normal vs Bacterial Pneumonia vs COVID-19) classification. Sekeroglu and Ozsahin et al. used several DL and ML models to detect COVID-19 and pneumonia from CXR images and performed 38 experiments using CNN, 10 experiments using five ML models, and 14 experiments using pre-trained TL models [23]. For two classifications, they used 1808 images to train their models and achieved a mean receiver operating characteristic (ROC) of 96.51%, and for the three-class classification, they used 6100 images and achieved a macro averaged F1 score of 94.10%. Khan et al. developed a CoroNet based on the pre-trained Xception model to identify COVID-19 and pneumonia [24]. For 3 class classification, their model achieved an accuracy of 95% and for 4 class (Normal vs COVID-19 vs bacterial vs Viral Pneumonia) achieved a precision and recall of 93% and 98.2% respectively. Nahiduzzaman et al. extracted the most discriminant features using a hybrid CNN-PCA from the CXR images to detect multivariant pneumonia [25]. They used ELM to discriminate viral pneumonia from bacterial pneumonia. For enhancing the contrast of the images, they used contrast limited adaptive histogram equalization (CLAHE) and trained their model using 5857 images and achieved an accuracy of 99.83% and 98.32% for two (Normal vs Pneumonia) and three-class (Normal vs bacterial vs Viral Pneumonia) classifications. Yamac et al. developed a convolutional sparse support estimator network based on a neural network to detect COVID-19, viral, and bacterial pneumonia from CXR images [26] using 6200 images and achieved an accuracy of 87.07% for a four-class classification, while 95.90% for COVID-19. Chandra et al. developed an automatic screening model to detect COVID-19 and pneumonia from CXR images [27]. 2,088 images were used for training the ML models and achieved the highest accuracy of 91.329% using support vector machine (SVM) with linear kernel. Nahiduzzaman et al. proposed a model ChestX-ray6 based on CNN to detect pneumonia from other lung-related diseases and achieved an accuracy of 97.60% [28]. Que et al. used U-NET and DenseNet to detect cardiomegaly disease from CXR images [29] through performing segmentation using U-NET and marking two separate parts namely cardiac and thoracic. By using augmentation, 2,630 images were produced for training their models and achieved the highest area under the receiver operating characteristics (AUROC) of 93.48% and accuracy of 93.75% for U-NET. On the contrary, Robiul et al. proposed an ensemble model based on ML models to detect COVID-19 with an optimistic accuracy of 99.73% [30]. Serte and Serener used a pre-trained ResNet model to classify pleural effusions (PE) from TB, pneumonia, and COVID-19 diseases [31]. They correctly detected PE with an accuracy of 99%, 75%, and 100% from pneumonia, TB, and COVID-19, respectively. For multiclass classification, they achieved an average accuracy of 83%. Sahlol et al. extracted the 50,000 features using a pre-trained MobileNet model from CXR images to detect TB disease [32]. They used an artificial ecosystem-based optimization algorithm to select 25 and 19 relevant features from the datasets named Shenzhen (SZ) and Dataset 2, respectively. Their proposed models outperformed the SOTA methods and achieved an accuracy of 90.20% and 94.10% for SZ and Dataset 2, respectively. Chandra et al. developed a computer aided diagnosis (CAD) system to detect TB disease from CXR images [33]. First, a guided image filter was used for image de-noising followed by lung segmentation. After feature extraction, SVM was employed for classification and accuracies of 95.60% and 99.40% were achieved for Montgomery (MT) and SZ datasets, respectively. Tawsifur et al. used nine TL and two U-net models to detect TB from CXR images [34]. Several databases were merged to create a single database with 7,000 CXR images for training the proposed model. They also performed augmentation and achieved accuracies of 96.7% and 98.6% using ChexNet [35] and DenseNet201 [36] respectively. Furthermore, t-distributed stochastic neighbor embedding [37] was used for data visualization. Muhammad et al. used 7,000 CXR images for extracting the features using three TL models [38] using the eXtreme Gradient Boosting [39] package to detect TB. Ayaz et al. combined CNN based features and hand-crafted features through ensemble learning to classify TB from CXR images [40]. They achieved an AUROC [41] of 99% and 97% for the MT and SZ datasets, respectively. Lopes and Valiati et al. used three TL models to predict TB from CXR images [42] and achieved an accuracy of 83.40% and 82.60% for MT and SZ datasets, respectively. From the above literature review, it was observed that most of the studies were limited to binary classes with a small amount of data in the datasets and failed to achieve promising performance particularly for multiclass classifications with large numbers of data. Several DL and ML approaches have been conducted, but most of them require high computational times and a large number of parameters. Furthermore, most of the studies detected COVID-19 with fewer data because of the scarcity of COVID-19 CXR images. For multiclass classification, the overall accuracy is again too poor with a large number of parameters. The goal of the proposed framework was to detect lung-related diseases with novel contribution from extracting features using lightweight CNN and classifying using extreme learning machine (ELM) to obtain fewer parameters, low computational times, and high classification performance. The main contribution of this work is outlined as follows. 1. Several databases have been combined to create a more extensive database of eight diseases (CXR8), including COVID-19, with different schemes, as shown in Table 1 . In scheme 1, eight classes have been considered; in scheme 2, four classes have been considered; and so on. Binary classes such as Normal and Tuberculosis or Normal and COVID-19 were chosen mainly to verify that the binary models could still produce higher accuracy than the multiclass models [1].Table 1 Various schemes considered for lung disease classification. Scheme 1 Normal Bacterial Pneumonia Viral Pneumonia COVID-19 Cardiomegaly Pleural Lung Opacity Tuberculosis Scheme 2 Scheme 3 Normal COVID-19 Bacterial Pneumonia Viral Pneumonia Normal Bacterial Pneumonia Viral Pneumonia Scheme 4 Scheme 5 Scheme 6 Normal Pneumonia COVID-19 Normal COVID-19 Normal Tuberculosis 2. A lightweight CNN model with three layers has been proposed to extract features from 23,690 CXR8 images, including 4,192 COVID-19 CXR images. 3. To reduce the complexity and increase the speed, the PCC has been used to remove the redundant and irrelevant features. 4. An ELM has been proposed to detect multiple lung-related diseases. Consequently, the CNN-PCC-ELM model outweighs 28 SOTA models while reducing the parameters, complexities, and processing times compared to the TL models. 5. The proposed framework was developed through different schemes and achieved optimistic results in the case of every circumstance, namely balanced and imbalanced, low-resolution CXR images. 2 Proposed framework Fig. 1 presents the proposed framework for detecting lung diseases from the CXR images. First, for making multi-class classification, several image databases were combined to create suitable customized databases. Then the resolutions of the CXR images were converted to a uniform size and subsequently normalization was performed. After pre-processing, a lightweight and straightforward CNN model was employed to extract 512 features. A PCC algorithm was used to identify 195 prominent features by eliminating 317 irrelevant features. These features were standardized using the z-score normalization technique. Finally, an ELM model classified the features to identify the lung diseases.Fig. 1 Proposed framework for lung-related disease classification. 2.1 Database construction Since the number of COVID-19 images in a particular publicly available database was not so large, therefore, several databases were combined to create a customized database with a total of 4,192 COVID-19 CXR images [13], [43], [44], [45], [46], [47]. CXR images of 1,000 cardiomegaly, 1,500 pleural, and 1,500 lung opacity diseases were included in this study [48]. On the other hand, 1,037 CXR images of tuberculosis disease were collected from three different databases [49], [50], [51]. Bacterial and viral pneumonia CXR images were differentiated from the Kaggle database. Finally, the CXR images of normal patients have been collected from two databases [52], [53]. Fig. 2 demonstrates the representative CXR images from each class. Hence, a total of 23,690 CXR images were used to construct eight classes where number of CXR images for normal, bacterial pneumonia, viral pneumonia, COVID-19, cardiomegaly, pleural, lung opacity, and tuberculosis were 10,192; 2,777; 1,493; 4,192; 1,000; 1,500; 1,500 and 1,036 respectively. Though the datasets were imbalanced, the proposed framework performed satisfactorily in the case of multi-class classification, which is revealed in the result section.Fig. 2 Sample CXR images of (A) Normal, (B) Bacterial Pneumonia, (C) Viral Pneumonia, (D) COVID-19, (E) Cardiomegaly, (F) Pleural, (G) Lung Opacity, and (H) Tuberculosis. In this study, six types of ML schemes were considered to evaluate the performance of the proposed framework. Scheme 1 is a 8-class classification (normal, bacterial pneumonia, viral pneumonia, COVID-19, cardiomegaly, pleural, lung opacity, tuberculosis), Scheme 2 is a 4-class classification (normal, COVID-19, bacterial pneumonia, viral pneumonia), Scheme 3 is a 3-class classification (normal, bacterial pneumonia, viral pneumonia), Scheme 4 is a 3-class classification (normal, COVID-19, pneumonia), Scheme 5 is a 2-class classification (normal, COVID-19), and Scheme 6 is a 2-class classification (normal, tuberculosis). Since the proposed framework was designed to achieve promising accuracy in case of multiclass classification (3, 4 and 8), Scheme 5 was also implemented to check whether the same framework could be able to achieve higher classification performances in case of binary classification. The first experiment with eight-class classification was analyzed using a segregated 5-fold cross-validation (CV) procedure to ensure the model's performance was not biased and the result was demonstrated in Section 4.1. The training and testing ratios for the remaining schemes were 80:20, as the majority of SOTA models in the literature split data rather than using CV [22], [23], [25], [26], [27]. Table 2 shows the number of images per class split into training and test set.Table 2 Datasets for different ML schemes with training and testing sets. Scheme No Class type Disease Types Training Data Test Data Scheme 1 8-class Normal 8,663 1,529 Bacterial Pneumonia 2,360 417 Viral Pneumonia 1,269 224 COVID-19 3,563 629 Cardiomegaly 850 150 Pleural 1,275 225 Lung Opacity 1,275 225 Tuberculosis 881 155 Total 20,136 3,554 Scheme 2 4-class Normal 8,153 2,039 COVID-19 3,354 838 Bacterial Pneumonia 2,222 555 Viral Pneumonia 1,194 299 Total 14,923 3,731 Scheme 3 3-class Normal 8,153 2039 Bacterial Pneumonia 2,222 555 Viral Pneumonia 1,194 299 Total 11,569 2,893 Scheme 4 3-class Normal 8,153 2,039 Pneumonia 1,194 299 COVID-19 3,354 838 Total 12,701 3,176 Scheme 5 2-class Normal 8,153 2,039 COVID-19 3,354 838 Total 11,507 2,877 Scheme 6 2-class Normal 8,153 2,039 Tuberculosis 829 207 Total 8,982 2,246 Since the databases were constructed from multiple sources of CXR images, the resolutions of the images were different from each other. Hence, the CXR images were resized to a dimension of 124 × 124. An image is represented by a number of pixel values ranging from 0 to 255, which adds complexity. To remove this complexity, normalization was carried out by dividing each image by 255 and converting the range within 0 to 1. Finally, the CXR images were ready to feed into the CNN model for feature extraction. 2.2 Feature extraction A lightweight CNN model was proposed for extracting new features from the raw pixel values of the CXR images. It has only three layers, fewer parameters than the TL model and requires less processing time. On the contrary, the TL models such as VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201 have 19, 50, 152, 164, and 201 layers respectively. The proposed model has only 13.7 million parameters, which is 11 times lower than the VGG19 model. Since the proposed CNN model has fewer parameters and layers, the feature extractor CNN model is lightweight compared to the other TL models. It is easy to use an end-to-end system as it does not require a separate modules for individual tasks such as feature extraction, feature selection etc. However, it has some limitations for instance an end-to-end cannot provide good accuracy with small amount of training data. Fig. 3 shows the architecture of the CNN model used in this study.Fig. 3 A lightweight CNN model employed for extracting features. The size of the input CXR images was 124 × 124. The CNN model was comprised of three convolutional layers (CL), where each CL block consisted of a batch-normalization, an activation, and a 2 × 2 max-pooling layer. Batch normalization was utilized since it speeded up and stabilized the model by re-centering and rescaling the layers' inputs, as well as reducing overfitting. In the first CL layer, the kernel size was 5 × 5 and the padding were the’same’ to check the border elements, which might contain important information. Additionally, the kernel size was kept relatively large in case any significant lesions were detected in the CXR images that needed to be collected readily. Following the initial pooling layer with an input dimension of 62 × 62, the kernel size is kept relatively small. The kernel size of the remaining CLs was 3 × 3 and used the’valid’ padding. Basically, a custom CNN-based TL model was developed in this work. Most of the TL models used large kernel size at the beginning and then the kernel size was reduced in the proceeding layers [23] due to an initial large size of the CXR images and subsequent smaller sizes after each convolution layer. To avoid gradient fading, a ReLU activation function was utilized after each batch normalization layer [54]. Following the final CL, two fully connected (FC) layers were employed, which were determined by a trial and error procedure. The first FC layer was 1,024 nodes dense, and the feature extraction was carried out from the final FC layer, which was 512 nodes deep. To reduce the complexity and overfitting, two dropout layers were included and as a result, the training speed increased significantly [55]. One dropout with a 0.3 probability was positioned after the last CL layer and another one with a 0.5 probability before the feature extraction layer. Since the amount of training data was large, the Adam optimizer was employed as it could provide improved performance for a large training dataset [56]. In order to extract the distinctive features, the model was run with the following parameters: 50 epochs; 32 batch size and 0.001 learning rate. Finally, a total of 512 important features were extracted from the last FC layer. Table 3 shows a summary of the CNN model employed. To begin, a large number of features were retrieved in order to minimize underfitting. Additionally, PCC was employed to eliminate redundant features from the derived ones in order to account for the overfitting and to improve classification accuracy.Table 3 List of layers, output shapes and parameters for the model during feature extraction. Layer (Type) Shape of Output Parameters conv1 (Conv2D) (None, 124, 124, 16) 1,216 bn1 (BatchNormalization) (None, 124, 124, 16) 64 av1 (Activation) (None, 124, 124, 16) 0 mp1 (MaxPooling2D) (None, 62, 62, 16) 0 conv2 (Conv2D) (None, 60, 60, 32) 4,640 bn2 (BatchNormalization) (None, 60, 60, 32) 128 av2 (Activation) (None, 60, 60, 32) 0 mp2 (MaxPooling2D) (None, 30, 30, 32) 0 conv3 (Conv2D) (None, 28, 28, 64) 18,496 bn3 (BatchNormalization) (None, 28, 28, 64) 256 av3 (Activation) (None, 28, 28, 64) 0 mp3 (MaxPooling2D) (None, 14, 14, 64) 0 dp1 (Dropout) (None, 14, 14, 64) 0 ft (Flatten) (None, 12544) 0 dense (Dense) (None, 1024) 12,846,080 bn4 (BatchNormalization) (None, 1024) 4,096 av4 (Activation) (None, 1024) 0 dp2 (Dropout) (None, 1024) 0 Feature Extraction (Dense) (None, 512) 524,800 Total Parameters 13,399,776 Trainable Parameters 13,397,504 Non-trainable Parameters 2,272 2.3 Feature selection In the current era of ML, data is very important. Every piece of information which is measurable and involved in recognizing a phenomenon or a circumstance is called a feature. To recognize each phenomenon individually, there might be a large number of features available in the real world. However, all the features are not highly related to the outcome [57]. Huge feature space makes the learning process of a ML algorithm slow and increases computational complexity. Hence, it is required to find out the optimal feature subset to resolve these issues through feature selection. Among many available FS strategies, PCC based FS technique could help in identifying distinctive features. Best feature subspace was selected from the feature extracted using the CNN for reducing complexity and increasing the performance and processing speed [57]. Correlation coefficients (CC) of all the features were determined and the first one of each pair containing a CC value above the threshold were eliminated. Finally, the standard scaler has been used to standardize the extracted features performed by subtracting the mean and scaling to mean–variance. Standardization features could help in achieving better classification performance [59], [60]. The equation for the standard score for sample x [60] (1) y=x-x¯σ Where x¯ is the mean of the samples and σ is standard deviation of the samples.Algorithm 1: Feature Selection using Pearson Correlation Coefficient 1: CorrMatrix = Features.cor () 2: for i in range(len(CorrMatrix.columns)): 3: for j in range(i): 4: If abs (CorrMatrix.iloc[i, j]) > threshold: 5: 6: ColName = CorrMatrix.columns[i] ColCorr.add(ColName) 7: Features.drop(ColCorr) 2.4 Extreme learning machine Hunag proposed ELM [61], which is a neural network (NN) based on a forward feed network. A single hidden layer has been used to classify the extracted features. Normally, ELM has shown optimistic results in case of multiclass classification and since there is no backpropagation needed in ELM, hence it also requires less time compared to the traditional NN or DL models. The training time was a thousand times faster than the traditional NN and achieved better generalization power due to the absence of backpropagation and higher classification performance [61], [62], [63], [64]. The parameters between the input layer to the hidden layer were derived arbitrarily whereas the hidden layer to output layer parameters were derived using pseudoinverse. Fig. 4 demonstrates the basic architecture of ELM model used in this study.Fig. 4 The architecture of Extreme Machine Learning. The number of hidden nodes and the parameters for this study are shown in Table 4 , where the hidden layer nodes were selected through trial-and-error method. The model would get overfitted if more hidden nodes were added, while if fewer hidden nodes were utilized, then the model would be under fitted. Consequently, in this study, adequate hidden nodes had employed, which produced promising classification results that resolved both the overfitting and underfitting issues. For the eight-class classification, the total trainable parameters were 13,397,504 for extracting the features using CNN and for classification, whereas the total number of parameters of ELM were 304,500, hence the total parameters were summed up to 13,702,004.Algorithm 2: Extreme learning machine X(n,m)=x(1,1)x(1,2)⋯x(1,m)x(2,1)x(2,2)⋯x(1,m)x(3,1)x(3,2)⋯x(1,m)⋮⋮⋱⋮x(n,1)x(n,2)⋯x(n,m)Y(n,t)=y(1,1)y(1,2)⋯y(1,t)y(2,1)y(2,2)⋯y(1,t)y(3,1)y(3,2)⋯y(1,t)⋮⋮⋱⋮y(n,1)y(n,2)⋯y(n,t) Here, X and Y be the feature and target matrix. 1. Randomly generates the input weight W(m,N) and bias B(1,N) matrix. W(m,N)=w(1,1)w(1,2)⋯w(1,N)w(2,1)w(2,2)⋯w(1,N)w(3,1)w(3,2)⋯w(1,N)⋮⋮⋱⋮w(m,1)w(m,2)⋯w(m,N) B(1,N)=b(1,1)b(1,2)⋯b(1,N) 2. Determine the output H(n,N) of the hidden layer. H(n,N)=G(X(n,m)·W(m,N)+B(1,N)) H(n,N)=h(1,1)h(1,2)⋯h(1,N)h(2,1)h(2,2)⋯h(1,N)h(3,1)h(3,2)⋯h(1,N)⋮⋮⋱⋮h(n,1)h(n,2)⋯h(n,N) Here, G is the activation function. 3. Determine the output weight matrix β(N,t) β(N,t)=H(N,n)†·T(n,t) 4. Make prediction using β(N,t) Table 4 Parameters of ELM for Different Schemes. Scheme Total Nodes in Input Layer Total Nodes in Hidden Layer Total Nodes in Output Layer Total No of Parameters Scheme 1 195 1,500 8 304,500 Scheme 2 60 700 4 44,800 Scheme 3 72 400 3 30,000 Scheme 4 59 300 3 18,600 Scheme 5 45 300 1 13,800 Scheme 6 29 400 1 12,000 Activation Function ReLU 3 Results and analysis 3.1 Evaluation matrices To assess the performance of the proposed ML models, a number of metrics such as accuracy, precision, recall, f1-score, and AUC were considered. The metrics can be defined by Equation (2) to Equation (6) [65], [66]:(2) Accuracy=TP+TNTP+TN+FP+FN (3) Precision=TPTP+FP (4) Recall=TPTN+FP (5) F1-Score=2×Precision×RecallPrecision+Recall (6) AUC=12(TPTP+FN+TNTN+FP) Where TP, TN, FP and FN, denote true positives, true negatives, false positives, and false negatives, respectively. 3.2 Environmental setup Python programming language was used to write the codes, which were run on PyCharm Community Edition (2021.2.3) software. The CNN model was implemented using Keras with TensorFlow as the backend for the feature extraction. The ELM models were trained and tested on a PC with a 11th generation Intel(R) Core (TM) i9-11900 CPU @2.50 GHz, 32 GB RAM, and an NVIDIA GeForce, RTX 3090 24 GB GPU, running on a 64-bit Windows 10 Pro operating system. The code is available at https://github.com/NahiduzzamanRuet/ChestX-Ray8/blob/main/ChestX-Ray8.py. In this study, all the schemes were performed using two models. In the first model, after the pre-processing, a lightweight CNN was applied to extract the 512 features, which were standardized. After feature standardization, the ELM was used to classify different lung diseases, and the model was named as CNN-ELM. In the second model, after the extraction of 512 features and performing standardization on the features, PCC was employed on the extracted features to eliminate the redundant features and to select the most discriminant ones. Finally, classification was carried out by the ELM, and the model was termed as CNN-PCC-ELM. Table 5 shows the PCC values and the extracted features for different schemes. The PCC threshold value was selected using a trial-and-error method. For Scheme 1 (8 classes), which was a large class, a threshold value of 0.82 produced the most promising results than other PCC values either higher or lower. For instance, for large classes (8-classes) like the Scheme 1, with a higher value than the threshold value would increase the number of unnecessary redundant features causing less accurate results. Similarly, a lower PCC value than the threshold value would reduce the number of discriminant features leading to poor quality results again. In the cases of smaller classes (4, 3, and 2 classes), a slightly lower PCC value provided the best prominent features for effective and accurate detection of the diseases.Table 5 PCC value and extracted features. Scheme No PCC Value Features of CNN-ELM Features of CNN-PCC-ELM Scheme 1 0.82 512 195 Scheme 2 0.80 512 60 Scheme 3 0.80 512 72 Scheme 4 0.80 512 59 Scheme 5 0.80 512 45 Scheme 6 0.80 512 29 3.3 Results for Scheme 1 For Scheme 1, a total of 20,136 data were employed to train the CNN, CNN-ELM and CNN-PCC-ELM models, and 3,554 data were employed to assess the proposed framework’s performance. A confusion matrix (CM) was used to calculate the precision, recall, and accuracy of the CNN, CNN-ELM models as shown in Fig. 5 A.Fig. 5 CMs for scheme 1: (A) CNN-ELM and (B) CNN-PCC-ELM. The average precision, recall, and accuracy of the CNN model were 0.75, 0.73 and 85%, respectively. However, the average precision, recall, and accuracy of the CNN-ELM model were recorded as 0.95, 0.88, and 93.19%, respectively (Table 6 ). The CNN-ELM model used 512 features that were too high. Therefore, the redundant and irrelevant features were removed for improving the classification performance.Table 6 Classification performance results of Scheme 1. Type of lung diseases Precision Recall F1-score Accuracy (%) CNN CNN-ELM CNN-PCC-ELM CNN CNN-ELM CNN-PCC-ELM CNN CNN-ELM CNN-PCC-ELM CNN CNN-ELM CNN-PCC-ELM Normal 0.96 0.89 0.96 0.99 1.00 1.00 0.97 0.94 0.98 – – – Bacterial Pneumonia 0.87 0.96 0.97 0.79 0.89 0.91 0.82 0.93 0.94 – – – Viral Pneumonia 0.66 0.92 0.91 0.73 0.84 0.94 0.70 0.88 0.92 – – – COVID-19 0.91 0.99 1.00 0.93 0.96 0.99 0.92 0.97 1.00 – – – Cardiomegaly 0.64 0.98 0.97 0.45 0.79 0.93 0.53 0.88 0.95 – – – Pleural 0.55 0.95 0.92 0.41 0.84 0.90 0.47 0.89 0.91 – – – Lung Opacity 0.46 0.93 0.93 0.61 0.78 0.84 0.52 0.85 0.88 – – – Tuberculosis 0.99 1.00 1.00 0.92 0.92 0.99 0.95 0.96 1.00 – – – Average 0.75 0.95 0.96 0.73 0.88 0.94 0.74 0.91 0.95 85 93.19 96.22 After extracting the 512 features using CNN, a PCC was used to eliminate a total of 317 redundant features leaving behind a total of 195 most prominent features. Finally, the ELM model was used for the classification, and the CNN-PCC-ELM model performance was calculated using a CM as shown in Fig. 5B. It was noticed that the proposed framework demonstrated a significantly increased accuracy of 96.22%, which was approximately 11% higher than the CNN and 3% higher than the CNN-ELM model. It shows that an end-to-end model performs poorly with relatively smaller dataset. The recall of CNN-PCC-ELM also increased by 6% compared to the CNN-ELM, model demonstrating the robustness in the model’s classification performance. In biomedical engineering, the recall must be high enough so that the patient with lung diseases must be correctly detected. Since the redundant and unnecessary features confused the model and reduced the classification performance, after removing the duplicate and irrelevant features, the model's performance increased with reduced complexity. Additionally, despite imbalance in the dataset, each class contributed equally (precision and recall almost higher than 91% and 84% respectively for each class) to the final outcome. The class wise ROC values for both the models presented in Fig. 6 demonstrated that the models’ discriminant capabilities for each class indicating consistency of the models even with an unbalanced dataset. The calculated average AUCs of the CNN-ELM and CNN-PCC-ELM were recorded as 99.10% and 99.48% respectively clearly indicating the superiority of the latter.Fig. 6 ROC curve for Scheme 1: (A) CNN-ELM and (B) CNN-PCC-ELM. To demonstrate the superior performance of the proposed framework in the case of the multiclass environment, the same dataset was used for training and testing the five TL models: VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201. The ROCs of the TL models are presented in Fig. 7 .Fig. 7 ROC curves for TL models in Scheme 1: (A) VGG19, (B) ResNet50, (C) ResNet152V2, (D) InceptionResNetV2 and (E) DenseNet201. The AUCs of VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201 were 98.61%, 93.08%, 98.42%, 98.30%, and 99.07%, respectively. Table 7 shows average precision, recall, and f1-score of the TL models. The accuracies of VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201 were 74.20%, 64.02%, 73.05%, 70.45%, and 74.30%, respectively, whereas the accuracy of CNN-PCC-ELM was 96.22%, which was almost 20% higher than the TL models. The CNN-PCC-ELM model achieved a promising result because the CNN was used to extract the discriminant features. PCC was applied to find the irrelevant and redundant features that contained insignificant information, which could reduce the model’s performance. Hence, the ELM model showed a favorable result with CNN-PCC for feature extraction and selection.Table 7 Results of Scheme 1 for VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201. TL models Precision Recall F1-score Accuracy (%) VGG19 0.72 0.60 0.63 76.53 ResNet50 0.61 0.50 0.51 68.94 ResNet152V2 0.67 0.63 0.64 77.52 InceptionResNetV2 0.58 0.60 0.58 69.36 DenseNet201 0.68 0.67 0.65 77.12 The CNN has three convolutional layers, and ELM has three layers; hence there are only six layers. On the contrary, the TL models VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201 have 19, 50, 152, 164, and 201 layers respectively. The proposed model has only 13.7 million parameters, which is 11 times lower than the VGG19 model. Since the proposed CNN model has fewer parameters and layers, the features extractor CNN model is lightweight compared to the other TL models. From Table 8 , it should be noted that it took 406 s to extract the 512 features using the CNN. Only 39 s were taken to remove the features using PCC and train the ELM. Therefore, the training time of the CNN-PCC-ELM model was only 445 s, whereas some of the TL models’ training times were almost four times greater than the proposed model. The proposed framework took only 0.0156 s to test, whereas the InceptionResNetV2 TL model took 7.3505 s, which was relatively higher. The accuracy of the proposed framework (96.22%) was significantly higher than the accuracy of the TL models.Table 8 Comparing performance of CNN-PCA-ELM with VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201. Models Accuracy (%) No. of Layers Parameters (million) Training Time (sec) Testing Time (sec) VGG19 76.53 19 143.6 563 1.9710 ResNet50 68.94 50 25.6 626 2.4319 ResNet152V2 77.52 152 60.3 1512 5.8072 InceptionResNetV2 69.36 164 55.8 1888 7.3505 DenseNet201 77.12 201 20.2 1491 4.4138 CNN-PCC-ELM 96.22 3 + 3 13.7 445 0.0156 Fig. 8 shows a comparison of classification performance of the CNN-PCC-ELM model with the TL models. The recall and accuracy of the proposed framework were 0.94 and 96.22%, which were greater than that of the other TL models. From Fig. 7, it was also noticed that the ROC of CNN-PCC-ELM (99.48%) was again better than ROCs of the VGG19, ResNet50, ResNet152V2, InceptionResNetV2, and DenseNet201 models (98.61%, 93.08%, 98.42%, 98.30%, and 99.07% respectively). Since the ROC curve shows how well a classification model performs across all categorization levels, it can be concluded that the proposed framework shows its robustness in every performance criterion.Fig. 8 Classification performance comparison between CNN-PCC-ELM and the TL models for Scheme 1. From the above results, it was clear that the proposed CNN-PCC-ELM model achieved a higher classification performance with reduced processing time, parameters, and layers compared to the TL models. Therefore, it is safe to conclude that in the case of multiclass classification, the proposed CNN-PCC-ELM is superior to the other TL models based on all types of performance criteria. 3.4 Results for Scheme 2 In this scheme, COVID-19 was detected from normal, bacterial, and viral pneumonia with a four-class classification. The CNN-ELM was trained using 14,923 data with 512 features. To calculate the model's classification performance 3,731 data was used where the number of normal, COVID-19, bacterial, and viral pneumonia data were 2,039, 838, 555, and 299, respectively. The classification result was calculated from the CMs shown in Fig. 9 .Fig. 9 CMs for Scheme 2: (A) using CNN-ELM and (B) using CNN-PCC-ELM. The recall and accuracy of the CNN-ELM were 0.90 and 94.67% as shown in Table 9 . After removing the redundant features, the CNN-PCC-ELM model was trained using the same data but with only 60 features. The CNN-PCC-ELM achieved an accuracy of 96.57%, which was approximately 2% higher than the accuracy of the CNN-ELM model.Table 9 Classification performance results of Scheme 2. Type of lung diseases Precision Recall F1-score Accuracy (%) CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM Normal 0.93 0.97 0.99 0.99 0.96 0.98 – – COVID-19 0.97 0.97 0.94 0.97 0.95 0.97 – – Bacterial Pneumonia 0.95 0.95 0.91 0.94 0.92 0.95 – – Viral Pneumonia 0.97 0.94 0.76 0.83 0.85 0.88 – – Average 0.96 0.96 0.90 0.97 0.92 0.97 94.67 96.57 The AUCs of the CNN-PCC-ELM (99.57%) was slightly better than that of the CNN-ELM (99.41%) and Fig. 10 demonstrates the class-wise ROCs of both the models.Fig. 10 ROC curves for Scheme 2: (A) CNN-ELM and (B) CNN-PCC-ELM. 3.5 Results for Scheme 3 In this scheme, multivariate pneumonia was detected by training the CNN-ELM models with 11,569 data having 512 features. A total of 2,893 data was used for testing the model and calculating the classification performance. In the second stage, the CNN-PCC-ELM model was trained using the same data having 72 features, which was also used for testing the model. The accuracy of the CNN-PCC-ELM model (97.33%) was almost 3% higher than that of the CNN-ELM model (94.61%) as demonstrated in Table 10 . The average precision, recall, and F1-score of the CNN-PCC-ELM models were also found to be better than that of the CNN-ELM model.Table 10 Classification performance results of Scheme 3. Type of lung diseases Precision Recall F1-score Accuracy (%) CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM Normal 0.95 0.99 1.00 0.99 0.97 0.99 – – Bacterial Pneumonia 0.92 0.93 0.91 0.95 0.91 0.94 – – Viral Pneumonia 0.96 0.92 0.68 0.88 0.80 0.90 – – Average 0.94 0.95 0.86 0.94 0.90 0.94 94.61 97.34 Fig. 11 shows the CMs for both the CNN-ELM and CNN-PCC-ELM models.Fig. 11 CMs for Scheme 3: (A) CNN-ELM and (B) CNN-PCC-ELM. The AUC value of the CNN-PCC-ELM model (99.33%) was also superior to that of the CNN-ELM model (98.41%) as demonstrated by the class wise ROC curves presented in Fig. 12 .Fig. 12 ROC curves for Scheme 3: (A) CNN-ELM and (B) CNN-PCC-ELM. 3.6 Results for Scheme 4 In this study, the main focus was to detect COVID-19 from different other lung diseases. Scheme 4 was designed to detect COVID-19 from pneumonia with a three-class classification. A total of 12,701 data with 512 features was used for training the CNN-ELM model and for assessing the performance of the model, 3,176 data was used. In the previous studies, a small number of COVID-19 CXR images were used, whereas 4,192 CXR images were employed in this study. The CNN-ELM model achieved an accuracy of 97.42% and a precision of 98% (Table 11 ). The CNN-PCC-ELM model achieved an optimistic accuracy of 99.55%, with a precision, recall and f1-score of 99% while testing the model using the same testing data with only 59 features. The performance criterion for Scheme 4 was calculated from the CMs shown in Fig. 13 .Table 11 Classification performance results of Scheme 4. Type of lung diseases Precision Recall F1-score Accuracy (%) CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM Normal 0.97 1.00 1.00 1.00 0.98 1.00 – – Pneumonia 1.00 0.99 0.95 0.99 0.98 0.99 – – COVID-19 0.99 0.99 0.93 1.00 0.96 0.99 – – Average 0.98 0.99 0.96 0.99 0.97 0.99 97.42 99.55 Fig. 13 CMs for Scheme 4: (A) CNN-ELM and (B) CNN-PCC-ELM. The AUCs of the CNN-ELM and CNN-PCC-ELM were 99.96% and 99.97%, respectively (Fig. 14 ).Fig. 14 ROC curves for Scheme 4: (A) CNN-ELM and (B) CNN-PCC-ELM. 3.7 Results for Scheme 5 In this scheme, COVID-19 was detected from the normal patient as a binary classification. The CNN-ELM model was trained using 11,507 data with 512 features and performance was measured by testing the model using 2,877 data. The accuracy and recall of the CNN-ELM model were 96.66% and 97%, respectively as shown in Table 12 . In the second stage, PCC has been used for eliminating 467 features and the proposed CNN-PCC-ELM model achieved an accuracy of 98.82% while testing with 45 features.Table 12 Classification performance results of Scheme 5. Type of lung diseases Precision Recall F1-score Accuracy (%) CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM Normal 0.99 0.99 0.98 0.99 0.98 0.99 – – COVID-19 0.97 0.98 0.96 0.98 0.96 0.98 – – Average 0.99 0.99 0.97 0.99 0.98 0.99 96.66 98.82 The performance criterion for Scheme 5 was calculated from the CMs shown in Fig. 15 .Fig. 15 CMs for Scheme 5: (A) CNN-ELM and (B) CNN-PCC-ELM. The AUC values of the CNN-ELM and CNN-PCC-ELM models were 99.81% and 99.88%, respectively and again this showed superior performance of the CNN-PCC-ELM model (Fig. 16 ).Fig. 17. Fig. 16 ROC curves for Scheme 5: (A) CNN-ELM and (B) CNN-PCC-ELM. Fig. 17 CMs for Scheme 6 (A) CNN-ELM and (B) using CNN-PCC-ELM. 3.8 Results for Scheme 6 In this final scheme, the tuberculosis disease was detected from the normal patient. In this case, 8,982 data with 512 features was used to train the CNN-ELM model and achieved an accuracy of 98.13% and a precision of 99% (Table 13 and Fig. 16) and an AUC of 99.95% (Fig. 18 A).Table 13 Performance Classification results of Scheme 6. Type of lung diseases Precision Recall F1-score Accuracy (%) CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM CNN-ELM CNN-PCC-ELM Normal 0.99 1.00 0.99 1.00 0.99 1.00 – – Tuberculosis 0.97 0.97 0.90 0.98 0.94 0.97 – – Average 0.99 1.00 0.98 1.00 0.98 1.00 98.13 99.51 Fig. 18 ROC curves for Scheme 6: (A) CNN-ELM and (B) CNN-PCC-ELM. After removing 483 irrelevant features, the CNN-PCC-ELM model was tested using 2,246 data with only 29 features and achieved an accuracy of 99.51% and an AUC of 100% which are shown in Table 13 and Fig. 18(B). 3.9 Comparison between the six schemes From Fig. 19 , it was identified that in every scheme, the CNN-PCC-ELM performed better than the CNN-ELM model. The CNN-ELM model used 512 features while the CNN-PCC-ELM model used 195, 60, 72, 59, 45, and 29 features for Schemes 1, 2, 3, 4, 5, and 6, respectively, which were selected by eliminating the redundant features resulting in better classification performance. Hence, the results demonstrated that PCC removed the duplicate and irrelevant features efficiently and this was reflected in the improved classification performance with less complexity of the CNN-PCC-ELM model.Fig. 19 Performance comparison graph between CNN-ELM and CNN-PCC-ELM models for different schemes. 4 Performance comparison and discussions 4.1 Comparison of CNN-PCC-ELM with ResNet50-PCC-ELM and VGG19-PCC-ELM In this section, a 5-fold CV was applied to the CNN-PCC-ELM (CPE) model to determine the effect on the outcomes. The classification performance of each fold was compared with ResNet50-PCC-ELM (RPE) and VGG19-PCC-ELM (VPE) as shown in Table 14 .Table 14 Performance comparison of CNN-PCC-ELM with ResNet50-PCC-ELM and VGG19-PCC-ELM. Precision Recall Accuracy AUC RPE VPE CPE RPE VPE CPE RPE VPE CPE RPE VPE CPE Fold1 0.66 0.87 0.96 0.57 0.80 0.96 75.26 89.19 97.78 0.909 0.980 0.996 Fold2 0.65 0.88 0.96 0.57 0.81 0.95 74.91 89.82 97.36 0.915 0.979 0.995 Fold3 0.65 0.87 0.96 0.57 0.80 0.96 74.55 89.80 97.83 0.911 0.981 0.996 Fold4 0.65 0.86 0.96 0.57 0.80 0.96 74.84 88.62 97.72 0.911 0.981 0.997 Fold5 0.66 0.87 0.97 0.58 0.81 0.96 75.69 89.91 98.12 0.919 0.979 0.997 The highest accuracy of 75.69% and AUC of 91.9% were obtained from Fold 5, when employing RPE and the highest accuracy of 89.91% and AUC of 97.9% were obtained from the same fold utilizing VPE. On the other hand, the proposed CPE attained an optimistic accuracy of 98.12% (almost 9% higher than the VPE model) and AUC of 99.7% (almost 2% higher than the VPE) from the same fold. Following PCC, the RPE and VPE contained just nine and three redundant characteristics, respectively. Thus, the RPE and VPE models required 503 and 509 features, respectively, whereas the CPE model required only 195 features. As a result, the CPE model (445 s) needed shorter processing time than the VPE (609 s) and RPE (686 s) models. Since pre-trained ResNet 50 and VGG 19 were not trained for specific CXR images hence their features extraction capabilities for the CXR images were limited. Whereas the CXR images were specifically employed to train the proposed CNN model that resulted in extracting discriminant features leading to an improved the performance. As shown in Fig. 20 , the suggested CPE model was adaptable and flexible in terms of classification performance and processing time.Fig. 20 Classification performance comparison between CNN-PCC-ELM and the ResNet50-PCC-ELM, VGG19-PCC-ELM models for Scheme 1. 4.2 Comparison with SOTA models This section compared different schemes with the existing SOTA methods as shown in Table 15 . Scheme 1 was a new merged dataset; hence, it could not be compared with other SOTA methods. CNN-PCC-ELM model performed better with this new merged dataset compared with different TL models as evidenced in Section 3.3 and Section 4.1. For scheme 2, the best performances were achieved by Rashid et al. in classifying four types of diseases (normal-COVID-19-bacterial-viral pneumonia) with an accuracy, recall, and precision of 90.13%, 90.13%, and 90.13%, respectively [90]. On the contrary, the proposed framework achieved an accuracy of 96.57%, almost 6% higher than their model, and a recall of almost 7% (97%) higher, demonstrating the model’s robustness for scheme 2.Table 15 Performance comparison with previous studies. Scheme No Reference No. of Classes Precision Recall Accuracy AUC Scheme 2 [24] 4 89.84% 89.94% 89.6% – [26] – 79.79% 87.07% – [71] 90% 87.75% 89.89% – [86] 80.92% 85.66% 76.46% – [90] 90.13% 90.13% 90.13% – CNN-PCC-ELM 96% 97% 96.57% 99.45% Scheme 3 [25] 3 99% 98% 98.32% 99.01% [67] 93.7% 93.2% 93.3% 95.00% CNN-PCC-ELM 95% 94% 97.34% 99.33% Scheme 4 [27] 3 – – 93.41% – [12] – – 72.6% – [22] – 86.7% 92.53% [14] 97.95% 97.94% 97.94% – [23] 92.70% 92.70% 95.99% – [68] 93.33% 93.33% 93.3% – [69] 96.33% 93% 97.97% – [8] 94.8% – 94.4% 94.4% [73] 98.05% 98.02% 97.99% 99.41% [77] 95% 95% 97.41% 96% [79] 99.02% 98.26% 97.12% – [80] 96.67% 93.33% 96% – [76] 95.86% 97.62% 97.51% – [86] 94.27% 96.40% 95.66% – [87] – – 88.52% – [90] 96.45% 96.41% 96.41% – [93] – 97.62% 97.27% 99.60% CNN-PCC-ELM 100% 99% 99.55% 99.97% Scheme 5 [5] 2 – – 96% – [7] – – 80.9% – [8] 96% – 96% 96% [9] 95.63% – 95.78% 98.21% [22] – 92.64% 96% – [12] – – 89.6% – [24] 93% 98.2% 89.6% – [23] – 93.92% 98.39% 96.48% [13] 97% 98% 98% – [17] 94% 76% 89.47% – [73] – 98.81% 98.67% 99.10% [77] 95% 95% 98.06% 95% [86] 98.45% 98.45% 99.81% – [10] 98.00% 98.00% 98.00% – [88] 88.00% 94.00% 98.00% – [89] 99.18% 98.37% 98.78% – [91] 98% 99% 99% – [92] 96% 91.14% 93.24% 96.86% [94] – 93.62% 93.4% – CNN-PCC-ELM 99% 99% 98.82% 99.88% Scheme 6 [30] 2 – 91.94% 90.23% – [70] – 97.3% 98.7% 99% [71] 98.3% 100% 97.72% 100% [40] – – 97.59% 99% [33] 99.42% 99.40% 99.40% 99% [34] 98.57% 98.56% 98.60% – [42] – – 84.7% 92.60% [75] 95.67% 95.10% 95.10% – [76] 99.18% 99.16% 99.16% – [84] – 98.41% 97.23% – CNN-PCC-ELM 100% 100% 99.51% 100% For Scheme 3, there was not much work performed to classify bacterial pneumonia from the viral pneumonia. The authors [25] published a study to detect the multivariate pneumonia where the 99.01 % AUC was achieved from the model and this value was greater than that of prior studies. However, the accuracy (97.34%) in this work was slightly lower than the most recent work and the previous model (98.32%) [25]. The discriminating ability of the proposed model (AUC of 99.33%) was remarkable compared to the prior works demonstrating its robustness in the detection of lung diseases. The main contribution could be differentiated in a way that in contrast to two types of pneumonia in the previous work, a maximum seven types of lung diseases including COVID-19 were considered with a large amount of data in the datasets in this work. For Scheme 4, several studies were carried out to detect COVID-19 from pneumonia patients. Only the latest works were used for comparison in the table. The proposed model achieved a 100% precision and AUC of 99.97%, surpassing the seventeen SOTA methods for the three-class classification (normal-pneumonia-COVID-19). Besides this, Patro et al. developed a custom SCovNet based on CNN, which achieved the highest accuracy of 97.99% among the other SOTA models. In contrast, the proposed CNN-PCC-ELM achieved a satisfactory accuracy of 99.55%, almost 2% higher than their model [73]. Azam et al. had the best precision and recall results out of all the SOTA models, coming in at 99.02% and 98.26%, respectively [79]. The proposed CNN-PCC-ELM outweighs their precision (100%) and recall (99%). Scheme 5 has been one of the most important focuses for researchers in the last three years. For the classification of COVID-19 from the regular patients with the help of CXR images, Joshi et al. achieved an optimistic accuracy of 99.81% but the recall was 98.45% which was slightly lower than the proposed model [86]. The proposed model performed a promising result with an optimistic AUC and recall of 99.89% and 99%, respectively which outperformed the existing nineteen SOTA models. Tuberculosis, one of the most dangerous lung-related diseases, was also detected accurately in this study. The proposed model detected the tuberculosis disease from the regular patient with a precision and recall of 99% greater than the nine latest methods reported. For Scheme 6, the CNN-PCC-ELM model achieved an accuracy of 99.51% and an AUC of 100%, which were again much better than other SOTA models. Additionally, it should be noted that although the majority of SOTA models required high-quality images (227 × 227; 224 × 224) to identify certain disorders, the proposed model employed low-resolution images (124 × 124) to detect the discriminating features [13], [32], [58], [74], [80], [81], [85]. As the number of pixels in the input CXR images are decreased, the complexity of the model is proportionately decreased as well. As a result of the above discussion, it can be concluded that the suggested CNN-PCC-ELM model outperformed the selected 53 SOTA models available in the literature for various schemes that ensured efficiency and accuracy in diagnosing numerous lung illnesses with small number of features and low-resolution input CXR images. Table 15 depicts that as the number of classes decreased, the model's performance improved. However, the accuracy of Scheme 5 was worse than the Scheme 6 even though both schemes fell in the category of binary class. This could be due to the differences in disease varieties and diverse features between the schemes. Similar arguments can also be reasoned for the difference in accuracy between Scheme 3 and Scheme 4. The number of redundant features increased from higher to the lower-class schemes (Table 5) and the number of redundant features can fluctuate depending on the disease types. As a result, training models for different schemes or datasets with the same number of features was not a wise move. Using a number of deep TL models, the majority of researchers showed a favorable outcome in the detection of lung diseases in the past several years. However, the TL models required longer processing times because of their larger number of parameters and layers. The main goal of this study was to develop a lightweight CNN model with fewer parameters and layers, which would reduce the processing time compared to the conventional TL models. Therefore, the relevant features were extracted using a basic CNN model and unnecessary features were removed by employing PCC. The combination of CNN's feature extraction, PCC's most notable feature selection and ELM model's classification capabilities outperformed the earlier SOTA models and maintained high accuracy even in the presence of imbalanced datasets as well as low quality CXR images. Therefore, it can be argued that the proposed framework will ensure processing efficiency and correctness in faster detecting the lung diseases. This study also solved another challenging task of detecting lung diseases from the subtle appearance of illness signs in the CXR images, which may be difficult for the radiologists to distinguish. Sometimes these modest radiographic characteristics of various diseases such as TB, pneumonia, opacity, and others also mislead the classifier, reducing the system's diagnostic performance, as highlighted in previous studies [27]. This study considered eight lung-related diseases (Scheme 1) and the proposed model distinguished the lung diseases accurately from the extracted features while attaining an optimistic classification performance (96.22%) in the case of multi-class classifications. From the above discussion, it can be summarized that the proposed framework may assist the radiologists in detecting multiple lung-related diseases from the CXR images accurately and confidently freeing up valuable time of medical doctors to engage with other high priority tasks. In future, more lung-related diseases will be collected and used to further develop the proposed lightweight model, which can be applied to an embedded system that automatically detects different lung-related diseases from the CXR images. This can aid the medical practitioners in quickly detecting the diseases and providing appropriate treatment to the patients into real-world clinical care. 5 Conclusion This study used an extensive dataset composed of 23,690 CXR images from seven types of lung diseases, including 4,192 CXR images of COVID-19 patients, to detect these diseases using ML and DL models. A lightweight CNN model with only three layers and fewer parameters has been used to extract 512 features, and PCC has efficiently reduced unnecessary redundant features. Finally, a simple ELM with a single hidden layer has been used to classify the life-threatening diseases. The proposed CNN-PCC-ELM has successfully detected the multi-class and classified the COVID-19 disease from the lung diseases with high classification performance and reduced complexity, parameters, layers, and time. For all the schemes, the proposed CNN-PCC-ELM outperformed several SOTA methods with a high accuracy of 96.22% and an AUC of 99.483% for eight class classification. At the same time, the proposed model achieved an optimistic AUC of 99.45%, 99.33%, 99.97%, 99.88%, and 100% for the Schemes 2, 3, 4, 5, and 6 which outperformed the most recent SOTA models. Again, most studies utilized transfer learning methods for recognizing COVID-19, TB, and pneumonia, which required pre-training and also required particular resolution images (like 224 × 224, 331 × 331). On the contrary, the proposed model did not require any pre-training, and with small-resolution images (124 × 124), the model correctly detects different types of lung-related diseases. Finally, comprehensive experiments demonstrate that the CNN-PCC-ELM model can accurately diagnose several well-known lung-related diseases with a lower computational overhead than the TL models, which can help radiologists and other doctors save patients' lives. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Uncited references [72], [78], [82], [83]. CRediT authorship contribution statement Md. Nahiduzzaman: Conceptualization, Methodology, Investigation, Data curation, Formal analysis, Writing – review & editing, Writing – original draft, Project administration. Md Omaer Faruq Goni: Investigation, Formal analysis, Writing – original draft, Writing – review & editing. Md. Robiul Islam: Supervision, Formal analysis, Validation, Writing – review & editing. Abu Sayeed: Supervision, Validation, Writing – review & editing. Md. Shamim Anower: Funding acquisition, Supervision, Validation, Writing – review & editing. Mominul Ahsan: Formal analysis, Supervision, Validation, Visualization, Writing – review & editing. Julfikar Haider: Formal analysis, Supervision, Visualization, Validation, Writing – original draft, Writing – review & editing. Marcin Kowalski: Formal analysis, Supervision, Validation, Writing – review & editing. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix Appendices.Area Under Curve AUC Area Under the Receiver Operating Characteristics AUROC Chest X-Ray CXR Chest X-Ray8 CXR8 CNN-PCC-ELM CPE Computer Aided Diagnosis CAD Confusion Matrix CM Contrast Limited Adaptive Histogram Equalization CLAHE Convolutional Layers CL Convolutional Neural Network CNN Correlation Coefficients CC Cross-Validation CV Decompose, Transfer, and Compose DETRAC Deep Learning DL Extreme Learning Machine ELM Fully Connected FC Machine Learning ML Montgomery MT Neural Network NN Pearson Correlation Coefficient PCC Acknowledgement The authors would like to thank Rajshahi University of Engineering and Technology (RUET) for supporting to conduct the research. ==== Refs References: 1 Ruuskanen O. Lahti E. Jennings L.C. Murdoch D.R. Viral pneumonia Lancet 377 9773 2011 1264 1275 21435708 2 World Health Organization, “Who coronavirus (covid-19) dashboard,” 2021. Accessed on: May, 22, 2022. [Online]. Available at: https://COVID19.who.int/table 3 World Health Organization, “Global tuberculosis report, 2021”. Accessed on: May, 22, 2022. [Online]. Available at: https://www.who.int/teams/global-tuberculosis-programme/tb-reports 4 Kamel S.I. Levin D.C. Parker L. Rao V.M. Utilization trends in noncardiac thoracic imaging, 2002–2014 J Am Coll Radiol 14 3 2017 337 342 27927591 5 Loey M. El-Sappagh S. Mirjalili S. Bayesian-based optimized deep learning model to detect covid-19 patients using chest x-ray image data Comput Biol Med 142 2022 105213 6 Haghanifar A. Majdabadi M.M. Choi Y. Deivalakshmi S. Ko S. Covid-cxnet: Detecting covid-19 in frontal chest x-ray images using deep learning Multimed Tools Appl 81 21 2022 30615 30645 35431611 7 Ieracitano C. Mammone N. Versaci M. Varone G. Ali A.-R. Armentano A. A fuzzy-enhanced deep learning approach for early detection of covid-19 pneumonia from portable chest x-ray images Neurocomputing 481 2022 202 215 35079203 8 Agrawal T. Choudhary P. Focuscovid: automated covid-19 detection using deep learning with chest x-ray images Evol Syst 13 4 2022 519 533 9 Gayathri J. Abraham B. Sujarani M. Nair M.S. A computer-aided diagnosis system for the classification of covid-19 and non-covid-19 pneumonia on chest x-ray images by integrating cnn with sparse autoencoder and feed forward neural network Comput Biol Med 141 2022 105134 10 Kassania S.H. Kassanib P.H. Wesolowskic M.J. Schneidera K.A. Detersa R. Automatic detection of coronavirus disease (covid-19) in x-ray and ct images: a machine learning based approach Biocybern Biomed Eng 41 3 2021 867 879 34108787 11 Yousefi B. Kawakita S. Amini A. Akbari H. Advani S.M. Akhloufi M. Impartially validated multiple deep-chain models to detect covid-19 in chest x-ray using latent space radiomics J Clin Med 10 14 2021 3100 34300266 12 Akter S. Shamrat F. Chakraborty S. Karim A. Azam S. Covid-19 detection using deep learning algorithm on chest x-ray images Biology 10 11 2021 1174 34827167 13 Chowdhury M.E.H. Rahman T. Khandakar A. Mazhar R. Kadir M.A. Mahbub Z.B. “Can ai help in screening viral and covid-19 pneumonia?” IEEE Access 8 2020 132665 132676 14 Horry M.J. Chakraborty S. Paul M. Ulhaq A. Pradhan B. Saha M. “Covid-19 detection through transfer learning using multimodal imaging data”, IEEE Access 8 2020 149808 149824 15 Rasheed J. Hameed A.A. Djeddi C. Jamil A. Al-Turjman F. A machine learning-based framework for diagnosis of covid-19 from chest x-ray images Interdisciplinary Sciences: Computational Life Sciences 13 1 2021 103 117 33387306 16 Panwar H. Gupta P. Siddiqui M.K. Morales-Menendez R. Bhardwaj P. Singh V. A deep learning and grad-cam based color visualization approach for fast detection of covid-19 cases using chest x-ray and ct-scan images Chaos Solitons Fractals 140 2020 110190 17 Minaee S. Kafieh R. Sonka M. Yazdani S. Soufi G.J. Deep-covid: Predicting covid-19 from chest x-ray images using deep transfer learning Med Image Anal 65 2020 101794 18 Afshar P. Heidarian S. Naderkhani F. Oikonomou A. Plataniotis K.N. Mohammadi A. Covid-caps: A capsule network-based framework for identification of covid-19 cases from x-ray images Pattern Recogn Lett 138 2020 638 643 19 Abbas A. Abdelsamea M.M. Gaber M.M. Classification of covid-19 in chest x-ray images using detrac deep convolutional neural network Appl Intell 51 2 2021 854 864 20 Arias-Londono J.D. Gomez-Garcia J.A. Moro-Velazquez L. Godino-Llorente J.I. “Artificial intelligence applied to chest x-ray images for the automatic detection of covid-19. a thoughtful evaluation approach”, IEEE Access 8 2020 226811 226827 21 Alam N.-A.-A. Ahsan M. Based M.A. Haider J. Kowalski M. Covid-19 detection from chest x-ray images using feature fusion and deep learning Sensors 21 4 2021 1480 33672585 22 Pandit M.K. Banday S.A. Naaz R. Chishti M.A. Automatic detection of covid-19 from chest radiographs using deep learning Radiography 27 2 2021 483 489 33223418 23 Sekeroglu B. Ozsahin I. <? COVID19?> detection of covid-19 from chest x-ray images using convolutional neural networks SLAS TECHNOLOGY: Translating Life Sciences Innovation 25 6 2020 553 565 24 Khan A.I. Shah J.L. Bhat M.M. Coronet: A deep neural network for detection and diagnosis of covid-19 from chest x-ray images Comput Methods Programs Biomed 196 2020 105581 25 Nahiduzzaman M.d. Goni M.O.F. Anower M.S. Islam M.R. Ahsan M. Haider J. “A novel method for multivariant pneumonia classification based on hybrid CNN-PCA based feature extraction using extreme learning machine with CXR images”, IEEE Access 9 2021 147512 147526 26 Yamaç M. Ahishali M. Degerli A. Kiranyaz S. Chowdhury M.E. Gabbouj M. Convolutional sparse support estimator-based covid-19 recognition from x-ray images IEEE Trans Neural Networks Learn Syst 32 5 2021 1810 1820 27 Chandra T.B. Verma K. Singh B.K. Jain D. Netam S.S. Coronavirus disease (covid-19) detection in chest x-ray images using majority voting-based classifier ensemble Expert Syst Appl 165 2021 113909 28 Nahiduzzaman M. Islam M.R. Hassan R. ChestX-Ray6: Prediction of multiple diseases including COVID-19 from chest X-ray images using convolutional neural network Expert Syst Appl 211 2023 118576 29 Que Q. Tang Z. Wang R. Zeng Z. Wang J. Chua M. 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) IEEE 2018 2018 612 615 30 Islam M.R. Nahiduzzaman M. Complex features extraction with deep learning model for the detection of COVID19 from CT scan images using ensemble based machine learning approach Expert Syst Appl 195 2022 116554 31 Serte S. Serener A. Early pleural effusion detection from respiratory diseases including covid-19 via deep learning IEEE 2020 2020 1 4 32 Sahlol A.T. Abd Elaziz M. Tariq Jamal A. Damaševiˇcius R. Farouk Hassan O. A novel method for detection of tuberculosis in chest radiographs using artificial ecosystem-based optimisation of deep neural network features Symmetry 12 7 2020 1146 33 Chandra T.B. Verma K. Singh B.K. Jain D. Netam S.S. Automatic detection of tuberculosis related abnormalities in chest x-ray images using hierarchical feature extraction scheme Expert Syst Appl 158 2020 113514 34 Rahman T. Khandakar A. Kadir M.A. Islam K.R. Islam K.F. Mazhar R. “Reliable tuberculosis detection using chest x-ray with deep learning, segmentation and visualization”, IEEE Access 8 2020 191586 191601 35 Rajpurkar P, Irvin J, Zhu K, Yang B, Mehta H, Duan T, et al. Chexnet: Radiologist-level pneumonia detection on chest x-rays with deep learning. arXiv preprint arXiv:1711.05225; 2017. 36 Huang G, Liu Z, Van Der Maaten L, Weinberger KQ. Densely connected convolutional networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition; 2017. p. 4700–8. 37 Van der Maaten L. Hinton G. Visualizing data using t-sne J Mach Learn Res 9 11 2008 38 Rahman M. Cao Y. Sun X. Li B. Hao Y. Deep pre-trained networks as a feature extractor with xgboost to detect tuberculosis from chest x-ray Comput Electr Eng 93 2021 107252 39 Chen T. He T. Benesty M. Khotilovich V. Tang Y. Cho H. Xgboost: extreme gradient boosting R package version 0.4-2 1 4 2015 1 4 40 Ayaz M. Shaukat F. Raja G. Ensemble learning based automatic detection of tuberculosis in chest x-ray images using hybrid feature descriptors Phys Eng Sci Med 44 1 2021 183 194 33459996 41 Hanley J.A. McNeil B.J. The meaning and use of the area under a receiver operating characteristic (roc) curve Radiology 143 1 1982 29 36 7063747 42 Lopes U. Valiati J.F. Pre-trained convolutional neural networks as feature extractors for tuberculosis detection Comput Biol Med 89 2017 135 143 28800442 43 “Chest x-ray (pneumonia, covid-19, tuberculosis),” 2021. Accessed on: December, 20, 2021. [Online]. Available at: https://www.kaggle.com/datasets/jtiptj/chest-xray-pneumoniacovid19tuberculosis 44 ml-workgroup. “covid-19-image-repository,” 2019. Accessed on: December, 20, 2021. [Online]. Available at: https://github.com/ml-workgroup/covid-19-image-repository/tree/master/png 45 Joseph Paul Cohen. “covid-chestxray-dataset,” 2019. Accessed on: December, 20, 2021. [Online]. Available: https://github.com/ieee8023/covid-chestxray-dataset/tree/master/images 46 Arman Haghanifar. “Covid-cxnet,” 2019. Accessed on: December, 20, 2021. [Online]. Available at: https://github.com/armiro/COVID-CXNet/tree/master/chest_xray_images/covid19 47 BIMCV Medical Imaging Databank of the Valencia Region. “Bimcv-COVID-19,” 2020. Accessed on: December 20, 2021. [Online]. Available: https://bimcv.cipf.es/bimcv-projects/bimcv-%20covid19/ 48 Ingus T. “Neo xrays,” 2019. Accessed on: December, 20, 2021. [Online]. Available: https://www.kaggle.com/ingusterbets/neo-xrays 49 Scott Mader K. “Pulmonary chest x-ray abnormalities,” 2017. Accessed on: December, 20, 2021. [Online]. Available: https://www.kaggle.com/kmader/pulmonary-chest-xray-abnormalities 50 Jaeger S. Candemir S. Antani S. Wáng Y.-X.-J. Lu P.-X. Thoma G. Two public chest x-ray datasets for computer-aided screening of pulmonary diseases Quant Imaging Med Surg 4 6 2014 475 25525580 51 TB Portal Tuberculosis Chest X-ray dataset for Belarus. Accessed on: December, 20, 2021. [Online] Available at: https://academictorrents.com/details/509f986b456b6fce04c15f9d1de22cd4ccb2c4b7 52 Paul Mooney. “Chest x-ray images (pneumonia),” 2017. Accessed on: December, 20, 2021. [Online]. Available: https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia 53 Radiological Society of North America. “Rsna pneumonia detection challenge,” 2018. Accessed on: December, 20, 2021. [Online]. Available: https://www.kaggle.com/c/rsna-pneumonia-detection-challenge/data 54 Agarap AF. Deep learning using rectified linear units (relu). arXiv preprint arXiv:1803.08375; 2018. 55 Kovács G. Tóth L. Van Compernolle D. Ganapathy S. Increasing the robustness of cnn acoustic models using autoregressive moving average spectrogram features and channel dropout Pattern Recogn Lett 100 2017 44 50 56 Kingma DP, Ba J. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980; 2014. 57 Kira K. Rendell L.A. A practical approach to feature selection Machine learning proceedings 1992 Elsevier 249 256 58 Benesty J. Chen J. Huang Y. Cohen I. Pearson correlation coefficient Noise reduction in speech processing 2009 Springer 1 4 59 Farrell J. Saloner G. Standardization, compatibility, and innovation Rand J Econ 1985 70 83 60 Nahiduzzaman M. Nayeem M.J. Ahmed M.T. Zaman M.S.U. Prediction of heart disease using multi-layer perceptron neural network and support vector machine IEEE 2019 2019 1 6 61 Huang G.-B. Zhu Q.-Y. Siew C.-K. Extreme learning machine: theory and applications Neurocomputing 70 1–3 2006 489 501 62 Nahiduzzaman M.d. Islam M.R. Islam S.M.R. Goni M.O.F. Anower M.S. Kwak K.-S. Hybrid CNN-SVD based prominent feature extraction and selection for grading diabetic retinopathy using extreme learning machine algorithm IEEE Access 9 2021 152261 152274 63 Nahiduzzaman M. Islam M.R. Goni M.O.F. Anower M.S. Ahsan M. Haider J. Diabetic retinopathy identification using parallel convolutional neural network based feature extractor and ELM classifier Expert Syst Appl 217 2023 119557 64 Goni M.O.F. Mondal M.N.I. Islam S.R. Nahiduzzaman M. Islam M.R. Anower M.S. Diagnosis of malaria using double hidden layer extreme learning machine algorithm with CNN feature extraction and parasite inflator IEEE Access 11 2023 4117 4130 65 Swets J.A. Measuring the accuracy of diagnostic systems Science 240 4857 1988 1285 1293 3287615 66 Powers DM. Evaluation: from precision, recall and f-measure to roc, informedness, markedness and correlation. arXiv preprint arXiv:2010.16061; 2020. 67 Rahman T. Chowdhury M.E. Khandakar A. Islam K.R. Islam K.F. Mahbub Z.B. Transfer learning with deep convolutional neural network (cnn) for pneumonia detection using chest x-ray Appl Sci 10 9 2020 3233 68 Wang L. Lin Z.Q. Wong A. Covid-net: A tailored deep convolutional neural network design for detection of covid-19 cases from chest x-ray images Sci Rep 10 1 2020 1 12 31913322 69 Jain R. Gupta M. Taneja S. Hemanth D.J. Deep learning-based detection and analysis of covid-19 on chest x-ray images Appl Intell 51 3 2021 1690 1700 70 Lakhani P. Sundaram B. Deep learning at chest radiography: automated classification of pulmonary tuberculosis by using convolutional neural networks Radiology 284 2 2017 574 582 28436741 71 Duong L.T. Le N.H. Tran T.B. Ngo V.M. Nguyen P.T. Detection of tuberculosis from chest X-ray images: boosting the performance with vision transformer and transfer learning Expert Syst Appl 184 2021 115519 72 Hariri M. Avşar E. COVID-19 and pneumonia diagnosis from chest X-ray images using convolutional neural networks Netw Model Anal Health Informat Bioinformat 2023 73 Patro K.K. Prakash A.J. Hammad M. Tadeusiewicz R. Pławiak P. SCovNet: a skip connection-based feature union deep learning technique with statistical approach analysis for the detection of COVID-19 Biocybern Biomed Eng 43 1 2023 352 368 36819118 74 Yuan J. Wu F. Li Y. Li J. Huang G. Huang Q. DPDH-CapNet: a novel lightweight capsule network with non-routing for COVID-19 diagnosis using X-ray images J Digit Imaging 2023 1 13 75 Iqbal A. Usman M. Ahmed Z. Tuberculosis chest X-ray detection using CNN-based hybrid segmentation and classification approach Biomed Signal Process Control 84 2023 104667 76 Indumathi V. Siva R. An efficient lung disease classification from X-ray images using hybrid Mask-RCNN and BiDLSTM Biomed Signal Process Control 81 2023 104340 77 George G.S. Mishra P.R. Sinha P. Prusty M.R. COVID-19 detection on chest X-ray images using Homomorphic Transformation and VGG inspired deep convolutional neural network Biocybern Biomed Eng 43 1 2023 1 16 36447948 78 Alshmrani G.M.M. Ni Q. Jiang R. Pervaiz H. Elshennawy N.M. A deep learning architecture for multi-class lung diseases classification using chest X-ray (CXR) images Alex Eng J 64 2023 923 935 79 Azam S. Rafid A.R.H. Montaha S. Karim A. Jonkman M. De Boer F. Automated detection of broncho-arterial pairs using CT scans employing different approaches to classify lung diseases Biomedicines 11 1 2023 133 36672641 80 Akbulut Y. Automated pneumonia based lung diseases classification with robust technique based on a customized deep learning approach Diagnostics 13 2 2023 260 36673070 81 Shamrat F.J.M. Azam S. Karim A. Ahmed K. Bui F.M. De Boer F. High-precision multiclass classification of lung disease through customized MobileNetV2 from chest X-ray images Comput Biol Med 155 2023 106646 82 Choi Y. Lee H. Interpretation of lung disease classification with light attention connected module Biomed Signal Process Control 84 2023 104695 83 Tian J. Zhang Y. Lei J. Sun C. Hu G. Lightweight classification network for pulmonary tuberculosis based on CT images J Artif Intell Technol 2023 84 Zhang G. Luo L. Zhang L. Liu Z. Research progress of respiratory disease and idiopathic pulmonary fibrosis based on artificial intelligence Diagnostics 13 3 2023 357 36766460 85 Rajagopal R. Karthick R. Meenalochini P. Kalaichelvi T. Deep convolutional spiking neural network optimized with arithmetic optimization algorithm for lung disease detection using chest X-ray images Biomed Signal Process Control 79 2023 104197 86 Joshi R.C. Yadav S. Pathak V.K. Malhotra H.S. Khokhar H.V.S. Parihar A. A deep learning-based COVID-19 automatic diagnostic framework using chest X-ray images Biocybern Biomed Eng 41 1 2021 239 254 33518878 87 Mishra N.K. Singh P. Joshi S.D. Automated detection of COVID-19 from CT scan using convolutional neural network Biocybern Biomed Eng 41 2 2021 572 588 33967366 88 Munusamy H. Muthukumar K.J. Gnanaprakasam S. Shanmugakani T.R. Sekar A. FractalCovNet architecture for COVID-19 chest X-ray image classification and CT-scan image segmentation Biocybern Biomed Eng 41 3 2021 1025 1038 34257471 89 Murugan R. Goel T. Mirjalili S. Chakrabartty D.K. WOANet: whale optimized deep neural network for the classification of COVID-19 from radiography images Biocybern Biomed Eng 41 4 2021 1702 1718 34720309 90 Rashid N. Hossain M.A.F. Ali M. Islam Sukanya M. Mahmud T. Fattah S.A. AutoCovNet: Unsupervised feature learning using autoencoder and feature merging for detection of COVID-19 from chest X-ray images Biocybern Biomed Eng 41 4 2021 1685 1701 34690398 91 Fang L. Wang X. COVID-RDNet: a novel coronavirus pneumonia classification model using the mixed dataset by CT and X-rays images Biocybern Biomed Eng 42 3 2022 977 994 35945982 92 Meng J. Tan Z. Yu Y. Wang P. Liu S. TL-med: a two-stage transfer learning recognition model for medical images of COVID-19 Biocybern Biomed Eng 42 3 2022 842 855 35506115 93 Gour M. Jain S. Automated COVID-19 detection from X-ray and CT images with stacked ensemble convolutional neural network Biocybern Biomed Eng 42 1 2022 27 41 34908638 94 Patel R.K. Kashyap M. Automated diagnosis of COVID stages from lung CT images using statistical features in 2-dimensional flexible analytic wavelet transform Biocybern Biomed Eng 42 3 2022 829 841 35791429
PMC010xxxxxx/PMC10292739.txt	==== Front Virus Res Virus Res Virus Research 0168-1702 1872-7492 The Authors. Published by Elsevier B.V. S0168-1702(23)00119-3 10.1016/j.virusres.2023.199157 199157 Article Incidence of Epstein-Barr virus reactivation is elevated in COVID-19 patients Bernal Keishanne Danielle E. ab Whitehurst Christopher B. a⁎ a Department of Pathology, Microbiology, and Immunology, New York Medical College, Basic Medical Sciences Building, 15 Dana Rd. Valhalla, NY 10595 b Westlake High School, 825 Westlake Dr., Thornwood, NY 10594 ⁎ Corresponding author. 26 6 2023 9 2023 26 6 2023 334 199157199157 8 5 2023 13 6 2023 16 6 2023 © 2023 The Authors 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. COVID-19, an infectious respiratory illness, is caused by infection with the SARS-CoV-2 virus. Individuals with underlying medical conditions are at increased risk of developing serious illnesses such as long COVID. Recent studies have observed Epstein-Barr virus (EBV) reactivation in patients with severe illness or long COVID, which may contribute to associated symptoms. We determined the frequency of EBV reactivation in COVID-19 positive patients compared to COVID-19 negative patients. 106 blood plasma samples were collected from COVID-19 positive and negative patients and EBV reactivation was determined by detection of EBV DNA and antibodies against EBV lytic genes in individuals with previous EBV infection. 27.1% (13/48) of EBV reactivations, based on qPCR detection of EBV genomes, are from the COVID positive group while only 12.5% (6/48) of reactivations belonged to the negative group. 20/52 (42.30%) of the COVID PCR negative group had detectable antibodies against SARS-CoV-2 nucleoprotein (Np); indicative of past infection. A significantly higher SARS-CoV-2 Np protein level was found in the COVID-19 positive group. In conclusion, COVID-19 patients experienced increased reactivation of EBV in comparison to COVID negative patients. Keywords COVID-19 Epstein-Barr virus EBV Reactivation Coronavirus SARS-CoV-2 ==== Body pmc1 Introduction COVID-19, which emerged in December of 2019, became a worldwide pandemic and has claimed over 6.6 million lives (World Health Organization 2023). While many infected individuals experience mild or moderate symptoms and recover in 7-10 days it has been reported that almost 16% of patients developed severe disease in a study conducted through January 29, 2020 (Guan et al., 2020). The mortality rate among severe cases of COVID has been reported to be as high as 61.5% (Yang et al., 2020). Since the initial emergence of COVID-19, four prominent variants have emerged: Alpha, Beta, Delta, and Omicron (Callaway, 2021). Omicron, the most recent and highly mutated variant of concern, was first identified in November 2021 and quickly became the dominant strain worldwide (Tian et al., 2022). Reactivation of Epstein-Barr virus (EBV) has been reported among the critically ill and patients suffering from long COVID and EBV viremia has been correlated with COVID severity (Naendrup et al., 2022, Paolucci et al., 2021, Saade et al., 2021, Simonnet et al., 2021, Gold et al., 2021, Zubchenko et al., 2022, Vojdani et al., 2023). A longitudinal multi-omic study suggested that four main risk factors for developing long COVID are type-2 diabetes, SARS-CoV-2 RNAemia, specific auto-antibodies, and Epstein-Barr virus viremia (Su et al., 2022). Reactivation of EBV may contribute to COVID symptoms, severity, and length of illness. The mechanism by which EBV reactivation may contribute to COVID is not quite clear, however Verma et al reported that EBV lytic replication promotes ACE2 expression and therefore could facilitate entry of SARS-CoV-2 (Verma et al., 2021). Epstein-Barr virus is one of nine known human herpesviruses and infects more than 90% of the world's population (Tzellos and Farrell, 2012). It is the first human oncogenic virus discovered and is associated with the development of Burkitt's lymphoma, Hodgkin's lymphoma, nasopharyngeal carcinoma, gastric carcinoma and more (Thompson and Kurzrock, 2004, zur Hausen et al., 1970). Primary EBV infections are usually asymptomatic and mild in children but after adolescence, it commonly causes infectious mononucleosis with symptoms including extreme fatigue, fever, head and body aches, and swollen spleen and lymph nodes. When symptoms are resolved, EBV persists for a lifetime by remaining latent in memory B lymphocytes (Fujiwara et al., 2014). However, EBV can be reactivated by prolonged psychological stress, hormonal changes, infections, and other factors that result in weakened cellular immunity (Murata et al., 2021, Sausen et al., 2021, Dochi et al., 2022, Aiello et al., 2010). This reactivation is associated with autoimmune disease, chronic fatigue syndrome and various other malignancies (Kerr, 2019). Healthy individuals are mostly asymptomatic to EBV reactivation, however immunosuppressed individuals can experience the same symptoms as their primary infection of EBV (Kerr, 2019). Here, we aim to determine if COVID infection can promote EBV reactivation, which could result in complicating symptoms of COVID illness. After primary EBV infection is resolved and latency is established, antibodies against the latent EBV nuclear antigen-1 (EBNA-1) are produced (Murata et al., 2021, Vouloumanou et al., 2012). Anti-EBNA-1 IgG antibodies are not present during the acute phase of EBV but become detectable 2-4 months after infection and persist for life. Detection of EBNA-1 IgG indicates a past infection. EBV reactivation results in expression of lytic gene products such as the viral capsid antigen (VCA) and early antigen-diffuse (EA-D). Anti-VCA IgM is detectable early in infection and reactivation but falls to undetectable levels in approximately 6 weeks. Anti-EA-D IgG also appears early after infection (3-4 weeks) and reactivation and typically falls to undetectable levels in approximately 4 months (Lennette, 1987). Detection of antibodies against VCA IgM and EA-D IgG is an indicator of EBV reactivation (Gulley, 2001) (Fig. 1 ). However, diagnosing EBV reactivation on serology alone can produce different results depending on the patient's disease course and the instability of anti-EBV antibodies before the appearance of symptoms (Gold et al., 2021). EBV DNA is frequently detectable in plasma during early infection and reactivation (Lam et al., 2018). qPCR detection of EBV DNA is more sensitive than serology in terms of evaluating reactivation (She et al., 2007).Fig. 1 Study Methodology. Fig 1 This study aims to determine if COVID-19 positive patients experience increased levels of EBV reactivation compared to COVID-19 negative patients and differs from previous studies in that samples were collected at a time when the Omicron variant was the dominant strain in the area (New York City Department of Health and Mental Hygiene 2022). Past reports on EBV reactivation were conducted largely before the Omicron variant emerged (Naendrup et al., 2022, Paolucci et al., 2021, Saade et al., 2021, Simonnet et al., 2021, Zubchenko et al., 2022, Rahimi and Talebi Bezmin Abadi, 2022, Im et al., 2022, Brooks et al., 2022, Chen et al., 2021, Meng et al., 2022, Vigon et al., 2021, Xie et al., 2021). Plasma samples from hospital patients determined to be COVID negative or COVID positive using PCR-based testing were studied. EBV DNA load was quantitated and serology toward EBV lytic genes were used as determinants of EBV reactivation (Fig. 1). 2 Materials and methods 2.1 Sample collection 106 whole blood samples were collected from different individuals treated at Westchester Medical Center, Valhalla, NY, between January 13, 2022, and March 23, 2022, and placed in EDTA tubes. Whole blood samples were spun down to collect plasma. 54 samples were from patients who tested positive for COVID-19 and 52 samples were from patients who tested negative for COVID-19 as determined by PCR-based testing by hospital staff. This IRB exempt study includes samples set to be discarded that were de-identified and marked only as COVID positive or COVID negative and given to the research team. No identifying information or patient data was supplied. Samples were collected during a time frame when the Omicron variant was the most dominant in the NYC area (New York City Department of Health and Mental Hygiene 2022). 2.2 Quantification of EBV DNA DNA from 200 μL of plasma was extracted using the DNeasy Blood & Tissue Kit (Qiagen). TaqMan primer sets were used for EBV quantification targeting the BamH1W and LMP2 regions of the EBV genome and were purchased from IDT (Ryan et al., 2004). TaqMan probes include a 5’ reporter FAM (520 nm emission) and double quencher ZEN/IBFQ. BamHIW1 Forward primer 5’ GCAGCCGCCCAGTCTCT 3’ BamHIW1 reverse primer 5’ ACAGACAGTGCACAGGAGACT BamHIW1 TaqMan probe 5’- FAM-AAAAGCTGGCGCCCTTGC 3’ ZEN/IBFQ LMP2 forward primer 5’ AGCTGTAACTGTGGTTTCCATGAC 3’ LMP2 reverse primer 5’ GCCCCCTGGCGAAGA G 3’ LMP2 TaqMan probe 5’-FAM-CTGCTGCTACTGGCTTTCGTCCTCTGG 3’ ZEN/IBFQ qPCR conditions were 95°C for 2 min 95°C for 15 sec., 60°C for 1 min. for 45 cycles using QuantStudio™ 5 from Applied Biosystems. Positive and negative controls were included, and a cutoff of CT 40 was used. After cycle 45 negative controls were undetectable, indicating that samples detected at cycle 40 represented levels at least 5 cycles (or 32-fold) beyond background. Detection of anti-ENBA1 IgG antibodies was used as a marker of past infection. To determine percent reactivation, samples which were not positive for EBNA-1 were excluded (10 total samples). 2.3 Enzyme-linked immunosorbent assay (ELISA) 2.3.1 EBNA-1 IgG, EA-D IgG, VCA IgM The presence of antibodies against EBNA-1 IgG, EA-D IgG, and VCA IgM were determined using ELISA kits from Abnova and were performed according to the manufacturer's protocols. To ensure the validity of the assay, positive and negative controls provided in the kit were used. Valid runs had positive controls with an index absorbance range between 2.3-4.2 AU and negative controls below 0.9 AU. The cutoff was calculated by multiplying the mean of calibrators and the calibration factor. Cutoffs were divided from each sample absorbance to get their positivity index. Positive samples have a positivity index >1.10 AU while negatives are <0.90 AU. 2.4 C-reactive protein (CRP) The relative quantification of CRP concentration was determined using a Human C-Reactive Protein ELISA Kit (Sigma-Aldrich). Samples were diluted 200,000-fold and analysis was performed according to manufacturers’ protocol. 2.5 COVID nucleoprotein (Np) Both COVID-19 positive and negative patient samples were quantified for antibodies against COVID Np to detect past SARS-CoV-2 virus infection. The samples were prepared according to the COVID-19 N Human IgG Indirect ELISA Kit (RayBio®) protocol. A cutoff of 6.5 U/ml was established and samples with values below 6.5 were listed as “Undetected” in Table 1 and excluded from analysis in Fig. 3C.Table 1 Test results for COVID negative and COVID positive patients. Table 1Negatives Positives Sample qPCR BamHI qPCR LMP2 CRP EBNA-1 EA(D) VCA N Protein Sample qPCR BamHI qPCR LMP2 CRP EBNA-1 EA(D) VCA N Protein 46 Undetected Undetected 26.6 + - - Undetected 63 Undetected Undetected 44.8 + - - Undetected 54 Undetected Undetected 2951.5 + + + Undetected 69 Undetected Undetected 1466.5 + + - Undetected 59 Undetected Undetected 1548.4 + + - 12.948 78 26.145 29.12241936 4544.4 +\- + - 83.575 85 Undetected Undetected 56.6 + no sample + 7.718 82 Undetected Undetected 3247.5 + +\- - 16.628 95 Undetected Undetected 222.7 + - - 7.186 102 Undetected Undetected 3379.9 + - - Undetected 164 Undetected Undetected 6.7 + - - Undetected 163 36.923 Undetected 1191.2 + - - 20.2 166 Undetected Undetected 9.7 - + - 6.511 194 Undetected Undetected 154.6 + - - 23.539 168 Undetected Undetected 9.5 + - - Undetected 196 37.197 Undetected 1260.4 + - - 45.169 172 36.482 Undetected 8087.6 + - - Undetected 198 36.421 Undetected 594.5 + - - Undetected 179 Undetected Undetected 510.7 + + - 12.451 200 Undetected Undetected 58.3 + + - 22.528 65 Undetected Undetected 1083.8 + +\- - 19.653 75 Undetected Undetected 1532 + - - 6.599 68 Undetected Undetected 2122.7 + - - no sample 84 Undetected 38.75872803 553.1 + - - 372.336 73 Undetected Undetected 232.9 + + - Undetected 115 Undetected Undetected 919.2 + - - 41.382 76 Undetected Undetected 26.7 + + - Undetected 116 Undetected Undetected 645.2 + - - 34.637 80 Undetected Undetected 43.1 + + - Undetected 122 33.331 38.16422653 286.9 + - - 30.567 58 Undetected Undetected 9.5 + - - Undetected 49 Undetected Undetected 20.1 + - - Undetected 64 Undetected Undetected 1135 + - - Undetected 50 Undetected Undetected 468.8 + - - 178.322 72 30.186 34.28359222 35.4 + - - Undetected 55 30.478 34.87576675 Undetected + no sample - Undetected 74 Undetected Undetected 1024.6 + + - 27.284 118 Undetected Undetected 46.4 + - - 56.7 90 35.134 37.69936752 80.8 + - - Undetected 120 35.869 Undetected 1881.4 - - - Undetected 51 Undetected Undetected 214 + +/- - 57.682 114 Undetected Undetected 3.6 + - - 49.225 53 Undetected Undetected 1491.6 + - - Undetected 117 Undetected Undetected Undetected + no sample - 35.822 57 Undetected Undetected 1728.6 - - - Undetected 123 Undetected Undetected 15.6 + - - 84.071 62 Undetected Undetected 60 + - + 780.647 124 Undetected Undetected 120.4 + + - 35.339 79 Undetected Undetected 39.2 + - - Undetected 126 38.546 Undetected 43 + +/- - 33.74 45 Undetected Undetected 483.6 + - - Undetected 125 Undetected Undetected 20.4 + - - 279.828 48 Undetected Undetected 72.8 + - - Undetected 128 Undetected Undetected 219 + + + 47.561 56 Undetected Undetected 71.8 + + - 7.771 133 Undetected Undetected 587.6 - - - 19.985 60 Undetected Undetected 768.4 + +/- - Undetected 136 35.145 36.49794769 1049.8 + + - 54.595 61 Undetected Undetected 2133 + - - Undetected 139 36.752 Undetected 861.8 + + - 62.035 131 31.656 35.08949661 2133 - - - Undetected 129 Undetected 39.97701645 775.2 + + - Undetected 134 Undetected Undetected 664 + +/- - 8.645 138 Undetected Undetected 479.2 + + - Undetected 165 Undetected Undetected 6 + +/- - 25.352 140 Undetected Undetected 1710.8 + - - 215.617 167 37.5962944 Undetected 1438.8 + + - Undetected 143 29.645 33.68924332 2688.6 + + + 28.817 177 Undetected Undetected 54.2 + - - Undetected 147 Undetected Undetected 26.4 + - - Undetected 171 Undetected Undetected 40.6 + - - Undetected 127 Undetected Undetected 1707.6 + - - 7.849 173 Undetected Undetected 12.8 + - - 7.859 142 Undetected Undetected 281.2 + +/- - 8.746 175 Undetected Undetected 8.8 + - - 24.327 152 Undetected Undetected 19.4 + - - 65.977 176 Undetected Undetected 7.6 + - - 18.832 158 Undetected Undetected 1211.2 + - - Undetected 187 37.096 37.981 358.6 + + - 17.92 161 Undetected Undetected 34.6 + - - Undetected 170 Undetected Undetected 13.8 + - - 36.545 144 Undetected Undetected 187.4 + - - Undetected 174 Undetected Undetected 1356.2 + - - Undetected 150 Undetected Undetected 6.2 + - - Undetected 178 Undetected Undetected 315 + - - 9.457 155 Undetected Undetected 2087.2 + + - 59.019 182 Undetected Undetected 326.8 + - - 25.194 157 Undetected Undetected 6.2 + + - 26.006 189 Undetected Undetected 1899.4 + - - Undetected 159 35.39 37.545 683.4 + - - 148.646 181 37.451 38.086 1963 + - - Undetected 149 Undetected Undetected 5.4 - - - Undetected 183 Undetected Undetected 6.4 + + - Undetected 153 Undetected Undetected 583 + - - 29.974 184 Undetected Undetected 360.6 + - - Undetected 154 Undetected Undetected 553.8 + +/- +\- 77.32 185 Undetected Undetected 36.4 + - - Undetected 156 36.199 39.246 1485 + + - Undetected 188 Undetected Undetected 167.6 - - - Undetected 192 Undetected Undetected 23.6 - +/- - Undetected 190 Undetected Undetected 552.2 + - - 16.409 193 Undetected Undetected 83.6 + - - 340.319 191 Undetected Undetected 13.6 + - - Undetected 195 Undetected Undetected 29.2 - - - 17.152 201 Undetected Undetected 10.4 + - - Undetected 202 Undetected Undetected 70 + - - 29.742 2.6 Statistical analysis Welch's t-test (Figs. 2 B, 3 D, 3E, 4 B), Mann-Whitney test (Figs. 3C and 4A) and two sample proportion Z test (Figs. 2A, 3A, 3B) were used to calculate significant difference. GraphPad Prism 9 was used to construct graphs and conduct statistical analysis. Summary of raw data is shown in Table 1.Fig. 2 Increased EBV reactivation in COVID-19 patients. The presence of viral genomes in patient serum was determined using BamHI and LMP2 primers to target EBV DNA. A) Samples testing positive for the presence of EBV DNA in serum are chartered as percentage of samples where EBV was reactivated among COVID negative and positive groups using BamHI, LMP2 or both primer sets. Two sample proportion Z test: BamHI: p=0.0906, LMP2: p=0.1085, BamHI/LMP2: p=0.0364. * represents p<0.05. B) Relative quantitation of EBV genome copies using BamHI and LMP2 primers sets. Welch's t-test with BamHI primers (p=0.8223). Welch's t-test with LMP2 primers (p=0.9436). Fig 2 Fig. 3 Detection of EA-D IgG and VCA IgM as determinants of EBV reactivation. A) Samples testing positive for the presence of EBV EA-D IgG and VCA IgM are charted as percentage of samples where EBV was reactivated among COVID negative and positive groups determined by PCR assay. Two sample proportion Z test: EBNA-1 and EA-D: p=0.3821, EBNA-1 and VCA: p=0.6593, EBNA-1 and EA-D/VCA: p=0.4519. B) Samples testing positive for the presence of either EBV EA-D IgG and VCA IgM, in addition to EBNA-1 IgG, are chartered as percentage of samples where EBV was reactivated among COVID negative and positive groups where COVID status was determined by the detection of COVID anti-nucleoprotein (COVID Np ab+) (left side) or detected by either PCR or detection of antibodies against nucleoprotein (right side). Two sample proportion Z test: COVID-19 positivity determined by Np seropositivity: p=0.2815, COVID-19 positivity determined by a positive PCR test or Np seropositivity: p=0.1588. C) Relative levels of antibodies against Np among COVID negative and COVID positive patients. COVID +/- groups on X-axis represent COVID determination by PCR test; therefore, COVID negative patients with detection of antibodies against Np demonstrate past infection. COVID status determined by PCR: p=0.0002 (Mann-Whitney test). *** represents p<0.001. D) Anti-Np IgG levels in EBV reactivated patients. Samples with antibodies against COVID Np were graphed for patients found to have reactivated EBV determined by detection of EBV genomes. EBV reactivation patients with current COVID determined by PCR had higher levels of antibodies against Np than EBV reactivating patients who had a past infection (p=0.066 Welch's t-test). E) Anti-Np IgG levels in EBV PCR negative patients. COVID PCR+ patients had higher levels of antibody against Np than COVID PCR – patients in the absence of EBV reactivation (EBV PCR-) (p=0.233 Welch's t-test). Fig 3 Fig. 4 A) Measurement of CRP levels among COVID positive and negative groups: (p=0.4691 Mann-Whitney test). B) Measurement of CRP levels among COVID positive and negative groups determined via PCR among patients with EBV reactivation determined by EBV genome detection: (p=0.4508 Welch's t-test). Fig 4 3 Results 3.1 Quantitative PCR detection of EBV SARS-CoV-2 infection results in increased EBV reactivation. Detection of COVID-19 resulted in increased EBV reactivation as determined by detection of EBV DNA in plasma. To distinguish primary infection from reactivation, past infection status must first be determined. EBNA-1 IgG is an indicator of previous EBV infection. Samples that did not test positive for the presence of EBNA-1 IgG were excluded from analysis and are likely primary infections, not a result of reactivation. 19/96 samples (19.8%) showed reactivated EBV based on detection of EBV genomes with TaqMan probes and at least one primer set (BamHI and LMP2). 13/48 (27.1%) of reactivations were from the COVID positive group, while only 6/48 (12.5%) of reactivations belonged to the negative group. 17/96 (17.7%) were detected using BamHI primers with 6/48 (12.5%) reactivated in the COVID negative group, and 11/48 (22.9%) reactivated in the COVID positive group. qPCR using the LMP2 primer set showed the reactivation of 12/96 samples (12.5%) with 4/48 (8.3%) reactivated in the COVID negative group, and 8/48 (16.7%) reactivated in the COVID positive group (Fig. 2A). 10/12 samples found to be reactivated with LMP2 were also detected with BamHI primers demonstrating strong overlap. For detection of EBV genomes in the plasma samples, 45 PCR cycles were performed to detect EBV DNA using TaqMan probes. In order to distinguish true positives from background levels, a cut-off of cycle 40 was used for inclusion as EBV reactivation determined by qPCR DNA detection. This represents a minimum 32-fold increase above background levels, drastically reducing the possibility of false positives. Negative controls were undetectable at cycle 45. Taken together the data shows that the incidence of EBV reactivation is increased in COVID patients. No statistical difference was found in the amount of EBV genomes detected in the plasma of reactivated COVID negative and COVID positive patients (Fig. 2B). The mean CT value of the COVID negative group (35.66 CT) was not statistically significant (p=0.8223) compared to the COVID positive group (35.09 CT) with BamHI primers. The mean CT values of EBV reactivated samples detected with LMP2 primers were very similar between the COVID negative group (37.01 CT) and positive (37.34 CT) group (p=0.9436). This data indicates that, in this study, the number of EBV genomes produced via reactivation do not differ significantly between COVID positive and negative patients. 3.2 EBV serology 96/106 patients (90.56%) had detectable levels of anti-EBNA-1 IgG, indicative of a past EBV infection. 25/103 (24.3%) were positive for anti-EA-D IgG and only 5/106 (4.71%) had anti-VCA IgM antibodies. 23/93 (24.7%) samples were positive for both EBNA-1 and EA-D, 12/46 (26.0%) of which were COVID positive and 11/47 (23.4%) were negative. 3/48 (6.3%) were positive for EBNA-1 and VCA from the COVID negative group and 2/48 (4.2%) were positive from the COVID positive group. When either EA-D or VCA was detected in combination with EBNA-1, 12/48 (25.0%) COVID negative patients showed reactivation compared to 12/48 (25.0%) in the COVID positive group (Fig. 3A). There was no significant difference in EBV reactivation as determined by serology against VCA and EA-D antibodies when the COVID negative and positive groups were defined by PCR assay. To determine if a patient had a past SARS-CoV-2 infection, we quantified antibodies against the COVID nucleoprotein. Most COVID-19 patients develop IgG antibodies within 2–3 weeks after symptom present (Van Elslande et al., 2021). COVID-19 anti-Np IgG antibodies can be detected as early as 7 -10 days after infection and remain for at least several months, whereas PCR-based testing for COVID-19 DNA may be detected 0-4 days after symptoms begin and may remain for several weeks (Koc et al., 2022, Centers for Disease Control and Prevention 2022, Mallett et al., 2020). 20/52 of the patient samples that were determined to be COVID negative based on PCR were found to have antibodies against COVID Np, indicating they had a past infection. 36/54 COVID positive samples (as determined by PCR) were positive for antibodies against Np. Overall, 56/106 (52.8%) patients had a previous or current SARS-CoV-2 infection as determined by antibody against Np. When seropositivity for anti-Np is used as the indicator for COVID-19 detection, there is an increasing trend in EBV reactivation determined through detection of antibodies against EA-D and VCA in COVID-19 positivity samples. 9/38 (23.7%) patient samples were negative for anti-Np antibodies but had antibodies against either EA-D or VCA, whereas 16/55 (29.1%) patient samples were positive for anti-Np antibodies and had antibodies against either VCA or EA-D. When COVID-19 positivity is determined via positive PCR test or Np seropositivity EBV reactivation is also observed more in COVID-19 positive samples (5/26 (19.2%) vs 20/68 (29.4%)) (see Fig. 3B). COVID positive samples had a median of 38.60 unit/mL of anti-Np antibodies, while negatives had a median of 17.16 unit/mL as determined by ELISA (Fig. 3C). The positive and negative groups have a statistically significant difference according to the Mann-Whitney test (p=0.0002), indicating an increase in Np antibodies for COVID PCR+ positive patients. The relationship between anti-Np IgG levels in EBV reactivated patients was evaluated. Samples with antibodies against COVID Np protein for patients found to have reactivated EBV are shown in Fig. 3D. Reactivated EBV was determined by detection of EBV DNA via either BamHI or LMP2 primers. COVID +/- groups represent COVID determination by PCR test. COVID negative patients with detection of antibodies against Np indicates a past infection. Interestingly, EBV reactivation patients with current COVID, determined by PCR, had higher levels of antibodies against Np than EBV reactivating patients who had a past infection (5.0 vs 61.9 average mean (p=0.066)) (Fig. 3D). Similarly, the relationship between anti-Np IgG levels in patients that did not experience EBV reactivation was also investigated among the COVID negative and positive groups (Fig. 3E). COVID PCR + patients had higher levels of antibodies against Np than those without current COVID (28.4 vs 58.5 average mean (p=0.223)). While Fig. 3D and E both show increased detection of antibodies against Np in individuals with PCR+ COVID, the difference in Np antibody levels among COVID positive and COVID negative group is much larger in the EBV reactivated patients. However this difference may be due to one sample with a very high level of antibodies against Np. 3.3 C-reactive protein COVID-19 positive and negative patient samples were monitored for CRP levels. The COVID positive samples have a higher median of 474 AU compared to the negative group's median of 218.4 AU. The difference of medians in the two groups is not statistically significant according to the Mann-Whitney test (p=0.4691) (Fig. 4A). There was no correlation between CRP levels and COVID positive vs. negative patients. Additionally, no correlation between CRP levels and EBV reactivation between COVID -/+ groups was found (Fig. 4B). COVID negative samples had a mean of 1994 AU and the COVID positive group's mean is 956 AU (p=0.4508). 4 Discussion Several studies have investigated the relationship between EBV and COVID. A study conducted in Wuhan, China observed that 55.2% of hospitalized COVID-19 patients were positive for EBV reactivation based on the presence of antibodies against VCA IgM (Chen et al., 2021). A recent observational case-control study conducted in Italy saw 95.2% of COVID-19 ICU patients and 83.6% SICU patients were positive for EBV reactivation. Their comparison between the two groups suggested a correlation between EBV DNA load and COVID severity (Paolucci et al., 2021). Saade et al found 56.1% of EBV reactivations in severe COVID-19 patients after admission to the ICU (Saade et al., 2021) and another reported increased antibodies against EBV and detectable viremia in plasma in critical COVID patients (Vigon et al., 2021). EBV reactivation has also been examined in the context of long COVID. As of 2022, the CDC determined that 1 in 5 adults in the U.S. who were previously infected with COVID-19 experienced long COVID conditions (Centers for Disease Control and Prevention 2022). Long COVID claimed over 3500 lives in America between January 2020 and June 2022 (Centers for Disease Control and Prevention 2022). The most persistent symptoms in patients hospitalized due to long COVID are fatigue, dyspnoea, loss of memory, and sleep disorders (Garrigues et al., 2020). A 2021 retrospective study observed 66.7% of long COVID patients were positive for EBV based on the presence of EA-D and VCA IgM antibodies (Gold et al., 2021). Another study by Zubchenko et al found EBV reactivation, determined by PCR detection of EBV DNA in peripheral blood, in 42.6% of long COVID patients (Zubchenko et al., 2022). Peluso et al reported that EBV reactivation is associated with higher odds of long COVID symptoms (Peluso et al., 2023). These previous studies focused on patient populations that were infected with a variants prior to Omicron. Our samples were collected when the Omicron variant was the most dominant in NYC cases (New York City Department of Health and Mental Hygiene 2022) which may result in differences from past studies. To our knowledge this is the first report investigating EBV reactivation due to COVID during the Omicron surge. This study analyzed EBV in samples from hospital patients that are positive and negative for COVID-19 to determine whether EBV reactivation is triggered by COVID-19 irrespective of disease severity, whereas many previous studies focused on severe or long COVID. Patient samples from this study likely represent the full spectrum of COVID infection, ranging from asymptomatic, to mild cases, to severe cases and long COVID. Herein we use both serology and EBV genome detection to analyze EBV reactivation and include detection of EBNA-1 to differentiate primary EBV infection from reactivation. Additionally past SARS-Cov-2 infections were detected by detecting antibodies against the SARS-CoV-2 Np. Using two primer sets for determining EBV reactivation, it was found that the COVID positive group resulted in significantly increased reactivation of EBV (27.1% vs 12.5%) compared to the negative group. It should be noted that our COVID negative group does not represent a healthy population and rather that of patients treated for various unknown reasons at Westchester Medical Center. Interestingly, the 12.5% reactivation among non-COVID patients found in this study is similar to the amount reported in a 2016 study among a cohort of patients (12%) treated at Johns Hopkins Hospital with no current, prior, or subsequent EBV disease (Kanakry et al., 2016). While we did not have access to a healthy population for this study, other groups have reported EBV reactivation rates of 0.6% and 3% among healthy and immunocompetent individuals as determined by PCR of serum samples (Kanakry et al., 2016, Walton et al., 2014). The mean CT for EBV genome copy detection of the COVID positive (35.09 CT) and negative group (35.66 CT) was not statistically significant (p=0.8223) using primers targeting BamHI. Paolucci et al reported increased EBV genome copy number in severe COVID cases (patients in ICU) compared those in less severe cases (sub-ICU) (Paolucci et al., 2021). While we did not observe an increase in EBV DNA in the serum, we did detect approximately a 2-fold increase in the number of people with EBV reactivation among COVID positive patients. The serology of EBNA-1 antibody presence in samples is in line with what can be seen in the general population (approximately 90%) (Tzellos and Farrell, 2012). The serology for EA-D and VCA antibodies did not differ between the COVID positive and negative groups when COVID status is determined by PCR. However, when COVID status was determined by detection of Np there was a noticeable increase in EBV reactivation as determined by serology for VCA and EA-D. Differences accounting for EBV reactivation determined by PCR vs serology are likely due to the time course of disease and initial production and duration of antibodies produced. The presence of EBV DNA-containing particles is likely cleared by the immune system before EA-D and VCA antibodies are produced. EBV reactivation results in the presence of EBV particles in the extracellular serum, therefore detection of EBV DNA via qPCR is likely a better indicator of reactivation. CRP is a biochemical marker of inflammation (Pepys and Hirschfield, 2003). An increase in its concentration is also associated with the severity of COVID-19 (Ali, 2020). A 2021 study found higher CRP levels in patients with both EBV and COVID-19 compared to patients with only COVID-19 (Chen et al., 2021). However, another study conducted when the alpha strain was dominant, found no association between EBV reactivation and elevated CRP (Brooks et al., 2022). When comparing data from this study we also found there was no statistical difference in CRP levels among COVID positive patients with and without EBV reactivation. COVID positive EBV reactivated patients had a mean CRP level of 956 AU and the COVID positive population without EBV reactivation had a CRP mean of 712 AU (p=0.357). Perhaps these variations are due to differences in COVID variants and disease severity of patients included in the studies. In this study, according to the comparison between COVID positive and negative groups, there is also not a significantly higher (p=0.4691) CRP level for patients with COVID-19. Since we do not have a healthy population, this could be due to other CRP-elevating conditions. High levels of antibodies against COVID Np can indicate patients that have been infected within weeks to 3 to 6 months prior. 20/52 of the COVID negative patients (38.5%) were identified to have been previously infected with SARS-CoV-2. Therefore, if these patients experience EBV reactivation, it is reasonable to consider COVID-19 as one of the possible causes. In this study it was found that EBV reactivation patients with current COVID-19 had higher levels of antibodies against Np than EBV reactivating patients who had a past infection (Fig. 3D). A study by Imai et al found that more severe COVID correlates with elevated anti-Np antibodies (Imai et al., 2021) and this data taken together could suggest that COVID severity correlates with increased EBV reactivation. The COVID negative group's significantly lower (p=0.0002) antibodies against COVID Np is also in line with the fact that these levels gradually decrease after the infection has been resolved. Our results point to a trend suggesting that COVID-19 reactivates EBV at a higher rate than non-COVID patients. Significance of this work is heightened by studies of hospitalized COVID patients showing that reactivated EBV significantly increased mortality when compared to EBV negative patients (Xie et al., 2021, Manoharan and Ying, 2023). In addition, it was found that patients with more severe pneumonia had EBV viremia (Im et al., 2022). Results of this work may help determine the course of treatment for COVID positive patients experiencing EBV reactivation. To this end, Meng et al found that patients experiencing EBV reactivation due to COVID showed increased survival outcomes when treated with the EBV inhibitor, ganciclovir (Meng et al., 2022). Funding This work was supported by a startup grant from 10.13039/100016250 New York Medical College . CRediT authorship contribution statement Keishanne Danielle E. Bernal: Conceptualization, Methodology, Investigation, Formal analysis, Writing – review & editing. Christopher B. Whitehurst: Conceptualization, Methodology, Investigation, Formal analysis, Writing – review & editing, Supervision. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available on request. Acknowledgments The authors would like to thank the staff of the blood bank at Westchester Medical Center for the clinical samples, and Lawrence McIntyre, science research teacher at Westlake High School for supervision, review and editing. ==== Refs References Aiello AE Simanek AM Galea S. Population levels of psychological stress, herpesvirus reactivation and HIV AIDS Behav. 14 2010 308 317 18264753 Ali N. Elevated level of C-reactive protein may be an early marker to predict risk for severity of COVID-19 J. Med. Virol. 92 2020 2409 2411 32516845 Brooks B Tancredi C Song Y Mogus AT Huang MW Zhu H Phan TL Zhu H Kadl A Woodfolk J Jerome KR Zeichner SL. Epstein-Barr virus and human herpesvirus-6 reactivation in acute COVID-19 patients Viruses 14 2022 Callaway E. Beyond Omicron: what's next for COVID's viral evolution Nature 600 2021 204 207 34876665 Centers for Disease Control and Prevention Interim Guidelines for COVID-19 Antibody Testing 2022 https://www.cdc.gov/coronavirus/2019-ncov/hcp/testing/antibody-tests-guidelines.html Accessed Centers for Disease Control and Prevention Nearly One in Five American Adults Who Have Had COVID-19 Still Have “Long COVID” 2022 https://www.cdc.gov/nchs/pressroom/nchs_press_releases/2022/20220622.htm Accessed Centers for Disease Control and Prevention More Than 3,500 Americans Have Died from Long COVID-Related Illness in the First 30 Months of the Pandemic 2022 https://www.cdc.gov/nchs/pressroom/nchs_press_releases/2022/20221214.htm Accessed Chen T Song J Liu H Zheng H Chen C. Positive Epstein-Barr virus detection in coronavirus disease 2019 (COVID-19) patients Sci. Rep. 11 2021 10902 34035353 Dochi H Kondo S Murata T Fukuyo M Nanbo A Wakae K Jiang WP Hamabe-Horiike T Tanaka M Nishiuchi T Mizokami H Moriyama-Kita M Kobayashi E Hirai N Komori T Ueno T Nakanishi Y Hatano M Endo K Sugimoto H Wakisaka N Juang SH Muramatsu M Kaneda A Yoshizaki T. Estrogen induces the expression of EBV lytic protein ZEBRA, a marker of poor prognosis in nasopharyngeal carcinoma Cancer Sci. 113 2022 2862 2877 35633182 Fujiwara S Kimura H Imadome K Arai A Kodama E Morio T Shimizu N Wakiguchi H. Current research on chronic active Epstein-Barr virus infection in Japan Pediatr. Int. 56 2014 159 166 24528553 Garrigues E Janvier P Kherabi Y Le Bot A Hamon A Gouze H Doucet L Berkani S Oliosi E Mallart E Corre F Zarrouk V Moyer JD Galy A Honsel V Fantin B Nguyen Y Post-discharge persistent symptoms and health-related quality of life after hospitalization for COVID-19 J. Infect. 81 2020 e4 e6 32853602 Gold JE Okyay RA Licht WE Hurley DJ. Investigation of long COVID prevalence and its relationship to Epstein-Barr virus reactivation Pathogens 10 2021 Guan WJ Ni ZY Hu Y Liang WH Ou CQ He JX Liu L Shan H Lei CL Hui DSC Du B Li LJ Zeng G Yuen KY Chen RC Tang CL Wang T Chen PY Xiang J Li SY Wang JL Liang ZJ Peng YX Wei L Liu Y Hu YH Peng P Wang JM Liu JY Chen Z Li G Zheng ZJ Qiu SQ Luo J Ye CJ Zhu SY Zhong NS China Medical Treatment Expert Group for C. Clinical characteristics of coronavirus disease 2019 in China N. Engl. J. Med. 382 2020 1708 1720 32109013 Gulley ML. Molecular diagnosis of Epstein-Barr virus-related diseases J. Mol. Diagn. 3 2001 1 10 11227065 Im JH Nahm CH Je YS Lee JS Baek JH Kwon HY Chung MH Jang JH Kim JS Lim JH Park MH. The effect of Epstein-Barr virus viremia on the progression to severe COVID-19 Medicine (Baltimore). 101 2022 e29027 35550458 Imai K Kitagawa Y Tabata S Kubota K Nagura-Ikeda M Matsuoka M Miyoshi K Sakai J Ishibashi N Tarumoto N Takeuchi S Ito T Maesaki S Tamura K Maeda T. Antibody response patterns in COVID-19 patients with different levels of disease severity in Japan J. Med. Virol. 93 2021 3211 3218 33620098 Kanakry JA Hegde AM Durand CM Massie AB Greer AE Ambinder RF Valsamakis A. The clinical significance of EBV DNA in the plasma and peripheral blood mononuclear cells of patients with or without EBV diseases Blood 127 2016 2007 2017 26744460 Kerr JR. Epstein-Barr virus (EBV) reactivation and therapeutic inhibitors J. Clin. Pathol. 72 2019 651 658 31315893 Koc HC Xiao J Liu W Li Y Chen G. Long COVID and its Management Int J Biol Sci 18 2022 4768 4780 35874958 Lam WKJ Jiang P Chan KCA Cheng SH Zhang H Peng W Tse OYO Tong YK Gai W Zee BCY Ma BBY Hui EP Chan ATC Woo JKS Chiu RWK Lo YMD. Sequencing-based counting and size profiling of plasma Epstein-Barr virus DNA enhance population screening of nasopharyngeal carcinoma Proc. Natl. Acad. Sci. U. S. A. 115 2018 E5115 E5124 29760067 Lennette E HW Epstein-Barr virus infections: clinical and serologic features Lab. Manag. 25 1987 23 26 Mallett S Allen AJ Graziadio S Taylor SA Sakai NS Green K Suklan J Hyde C Shinkins B Zhelev Z Peters J Turner PJ Roberts NW di Ruffano LF Wolff R Whiting P Winter A Bhatnagar G Nicholson BD Halligan S. At what times during infection is SARS-CoV-2 detectable and no longer detectable using RT-PCR-based tests? A systematic review of individual participant data BMC Med. 18 2020 346 33143712 Manoharan S Ying LY. Epstein Barr virus reactivation during COVID-19 hospitalization significantly increased mortality/death in SARS-CoV-2(+)/EBV(+) than SARS-CoV-2(+)/EBV(-) Patients: a comparative meta-analysis Int. J. Clin. Pract. 2023 2023 1068000 Meng M Zhang S Dong X Sun W Deng Y Li W Li R Annane D Wu Z Chen D. COVID-19 associated EBV reactivation and effects of ganciclovir treatment Immun Inflamm Dis 10 2022 e597 35349757 Murata T Sugimoto A Inagaki T Yanagi Y Watanabe T Sato Y Kimura H. Molecular basis of Epstein-Barr virus latency establishment and Lytic reactivation Viruses 13 2021 Naendrup JH Garcia Borrega J Eichenauer DA Shimabukuro-Vornhagen A Kochanek M Boll B Reactivation of EBV and CMV in severe COVID-19-epiphenomena or trigger of hyperinflammation in need of treatment? A large case series of critically ill patients J. Intensive Care Med. 37 2022 1152 1158 34791940 New York City Department of Health and Mental Hygiene Omicron Variant: NYC Report for January 13, 2022 2022 https://www1.nyc.gov/assets/doh/downloads/pdf/covid/omicron-variant-report-jan-13-22.pdf Accessed Paolucci S Cassaniti I Novazzi F Fiorina L Piralla A Comolli G Bruno R Maserati R Gulminetti R Novati S Mojoli F Baldanti F San Matteo Pavia C-TF. EBV DNA increase in COVID-19 patients with impaired lymphocyte subpopulation count Int. J. Infect. Dis. 104 2021 315 319 33359064 Peluso MJ Deveau TM Munter SE Ryder D Buck A Beck-Engeser G Chan F Lu S Goldberg SA Hoh R Tai V Torres L Iyer NS Deswal M Ngo LH Buitrago M Rodriguez A Chen JY Yee BC Chenna A Winslow JW Petropoulos CJ Deitchman AN Hellmuth J Spinelli MA Durstenfeld MS Hsue PY Kelly JD Martin JN Deeks SG Hunt PW Henrich TJ Chronic viral coinfections differentially affect the likelihood of developing long COVID J. Clin. Invest. 2023 133 Pepys MB Hirschfield GM. C-reactive protein: a critical update J. Clin. Invest. 111 2003 1805 1812 12813013 Rahimi F Talebi Bezmin Abadi A. Omicron: a highly transmissible SARS-CoV-2 variant Gene Rep. 27 2022 101549 Ryan JL Fan H Glaser SL Schichman SA Raab-Traub N Gulley ML. Epstein-Barr virus quantitation by real-time PCR targeting multiple gene segments: a novel approach to screen for the virus in paraffin-embedded tissue and plasma J. Mol. Diagn. 6 2004 378 385 15507678 Saade A Moratelli G Azoulay E Darmon M. Herpesvirus reactivation during severe COVID-19 and high rate of immune defect Infect. Dis. Now. 51 2021 676 679 34332165 Sausen DG Bhutta MS Gallo ES Dahari H Borenstein R. Stress-induced Epstein-Barr virus reactivation Biomolecules 11 2021 She RC Stevenson J Phansalkar AR Hillyard DR Litwin CM Petti CA. Limitations of polymerase chain reaction testing for diagnosing acute Epstein-Barr virus infections Diagn. Microbiol. Infect. Dis. 58 2007 333 335 17376631 Simonnet A Engelmann I Moreau AS Garcia B Six S El Kalioubie A Robriquet L Hober D Jourdain M. High incidence of Epstein-Barr virus, cytomegalovirus, and human-herpes virus-6 reactivations in critically ill patients with COVID-19 Infect. Dis. Now. 51 2021 296 299 33495765 Su Y Yuan D Chen DG Ng RH Wang K Choi J Li S Hong S Zhang R Xie J Kornilov SA Scherler K Pavlovitch-Bedzyk AJ Dong S Lausted C Lee I Fallen S Dai CL Baloni P Smith B Duvvuri VR Anderson KG Li J Yang F Duncombe CJ McCulloch DJ Rostomily C Troisch P Zhou J Mackay S DeGottardi Q May DH Taniguchi R Gittelman RM Klinger M Snyder TM Roper R Wojciechowska G Murray K Edmark R Evans S Jones L Zhou Y Rowen L Liu R Chour W Algren HA Berrington WR Wallick JA Cochran RA Multiple early factors anticipate post-acute COVID-19 sequelae Cell 185 2022 881 895.e20 35216672 Thompson MP Kurzrock R. Epstein-Barr virus and cancer Clin. Cancer Res. 10 2004 803 821 14871955 Tian D Sun Y Xu H Ye Q. The emergence and epidemic characteristics of the highly mutated SARS-CoV-2 Omicron variant J. Med. Virol. 94 2022 2376 2383 35118687 Tzellos S Farrell PJ. Epstein-barr virus sequence variation-biology and disease Pathogens 1 2012 156 174 25436768 Van Elslande J Oyaert M Ailliet S Van Ranst M Lorent N Vande Weygaerde Y Andre E Lagrou K Vandendriessche S Vermeersch P Longitudinal follow-up of IgG anti-nucleocapsid antibodies in SARS-CoV-2 infected patients up to eight months after infection J. Clin. Virol. 136 2021 104765 Verma D Church TM Swaminathan S. Epstein-Barr virus lytic replication induces ACE2 expression and enhances SARS-CoV-2 pseudotyped virus entry in epithelial cells J. Virol. 95 2021 e0019221 Vigon L Garcia-Perez J Rodriguez-Mora S Torres M Mateos E Castillo de la Osa M Cervero M Malo De Molina R Navarro C Murciano-Anton MA Garcia-Gutierrez V Planelles V Alcami J Perez-Olmeda M Coiras M Lopez-Huertas MR Impaired antibody-dependent cellular cytotoxicity in a Spanish cohort of patients with COVID-19 admitted to the ICU Front. Immunol. 12 2021 742631 Vojdani A Vojdani E Saidara E Maes M. Persistent SARS-CoV-2 infection, EBV, HHV-6 and other factors may contribute to inflammation and autoimmunity in long COVID Viruses 15 2023 Vouloumanou EK Rafailidis PI Falagas ME. Current diagnosis and management of infectious mononucleosis Curr. Opin. Hematol. 19 2012 14 20 22123662 Walton AH Muenzer JT Rasche D Boomer JS Sato B Brownstein BH Pachot A Brooks TL Deych E Shannon WD Green JM Storch GA Hotchkiss RS. Reactivation of multiple viruses in patients with sepsis PLoS One 9 2014 e98819 24919177 World Health Organization WHO Coronavirus (COVID-19) Dashboard 2023 https://covid19.who.int/ Accessed Xie Y Cao S Dong H Lv H Teng X Zhang J Wang T Zhang X Qin Y Chai Y Yang L Liu J Wang R. Clinical characteristics and outcomes of critically ill patients with acute COVID-19 with Epstein-Barr virus reactivation BMC Infect. Dis. 21 2021 955 34525962 Yang X Yu Y Xu J Shu H Xia J Liu H Wu Y Zhang L Yu Z Fang M Yu T Wang Y Pan S Zou X Yuan S Shang Y. Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study Lancet Respir. Med. 8 2020 475 481 32105632 Zubchenko S Kril I Nadizhko O Matsyura O Chopyak V. Herpesvirus infections and post-COVID-19 manifestations: a pilot observational study Rheumatol. Int. 42 2022 1523 1530 35650445 zur Hausen H Schulte-Holthausen H Klein G Henle W Henle G Clifford P Santesson L. EBV DNA in biopsies of Burkitt tumours and anaplastic carcinomas of the nasopharynx Nature 228 1970 1056 1058 4320657
PMC010xxxxxx/PMC10292824.txt	==== Front Lancet Microbe Lancet Microbe The Lancet. Microbe 2666-5247 The Author(s). Published by Elsevier Ltd. S2666-5247(23)00115-5 10.1016/S2666-5247(23)00115-5 Personal View Viral persistence in children infected with SARS-CoV-2: current evidence and future research strategies Buonsenso Danilo PhD ab* Martino Laura MD a Morello Rosa MD a Mariani Francesco MD a Fearnley Kelly MD c Valentini Piero MD a a Department of Woman and Child Health and Public Health, Fondazione Policlinico Universitario A Gemelli IRCCS, Rome, Italy b Centro di Salute Globale, Università Cattolica del Sacro Cuore, Rome, Italy c Bradford Royal Infirmary, West Yorkshire, UK * Correspondence to: Dr Danilo Buonsenso, Department of Woman and Child Health and Public Health, Fondazione Policlinico Universitario A Gemelli IRCCS, Rome 00168, Italy 26 6 2023 26 6 2023 © 2023 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND 4.0 license 2023 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In this Personal View, we discuss current knowledge on SARS-CoV-2 RNA or antigen persistence in children infected with SARS-CoV-2. Based on the evidence that the virus can persist in adults, we have done a literature review and analysed studies that looked for SARS-CoV-2 RNA or antigens in children undergoing autopsy, biopsy, or surgery for either death from COVID-19 or multisystem inflammatory syndrome, or assessments for long COVID-19 or other conditions. Our analysis suggests that in children, independent from disease severity, SARS-CoV-2 can spread systemically and persist for weeks to months. We discuss what is known about the biological effects of viral persistence for other viral infections and highlight new scenarios for clinical, pharmacological, and basic research exploration. Such an approach will improve the understanding and management of post-viral syndromes. ==== Body pmcIntroduction SARS-CoV-2 was initially isolated in December, 2019. Since then, knowledge of this virus and its interaction with humans has advanced. Thanks to enormous efforts, researchers have been able to track the viru's variants, and develop vaccines, diagnostic tests, and pharmacological treatments that have improved management of the virus and the disease it causes—ie, COVID-19. However, several unknowns remain, which still have a major effect on the health of adults and children. One of the major challenges for patients and researchers is post-COVID-19 condition (also known as long COVID), a term that describes the persistence of otherwise unexplained signs and symptoms, which begin after SARS-CoV-2 infection, that negatively affect peoples’ daily lives.1 A similar paediatric definition has also been released.2 Hundreds of studies have been published regarding long COVID in both adults and children. Several biological abnormalities have been linked to the development of this condition, however the exact pathogenesis is still unknown.3 Currently, attention is directed towards persistence of the virus, or parts thereof, in the human body after initial infection. In this Personal View, we will discuss the current evidence on possible SARS-CoV-2 persistence in paediatric patients, how it might affect the patient, and how it might lead to long COVID. We will also discuss why these observations can inspire future research projects for both diagnostics and therapeutics. Personal experience from a doctor and patient are described in the (appendix p 1). SARS-CoV-2 persistence in studies on adults SARS-CoV-2 is known to invade both the respiratory and non-respiratory tissues, causing an infection varying in severity, from asymptomatic or mild, to severe and fatal. Several autopsy studies have documented the anatomopathological findings and related immune changes of multiple organs in patients with critical disease. These findings have shown that COVID-19 is not simply a respiratory infection, but that it potentially has major effects on the whole body,4, 5, 6 including the endothelial system.7 However, as there is evidence that SARS-CoV-2 causes a persistent illness in a subgroup of patients, researchers have looked for immunopathology and viral persistence in patients that died, for any reasons, weeks to months after the initial infection. The findings provided striking evidence that parts of the virus can persist in the body.8 A major study has been recently published by Stein and colleagues, who performed complete autopsies on 44 patients who died from COVID-19. They did extensive sampling of the CNS in 11 patients to map and quantify the distribution, replication, and cell-type specificity of SARS-CoV-2 across the human body, including the brain, from the moment of acute infection onset, to more than 7 months after symptom onset.8 In all patients who had died from COVID-19 several weeks to months after initial infection, SARS-CoV-2 RNA persistence was detected across multiple tissue groups, including in the CNS across several brain regions, despite being undetectable in the plasma. Stein and colleagues found subgenomic RNA in at least one tissue in 14 of 27 patients beyond day 14, indicating that viral replication might occur in non-respiratory tissues for several months. Although the viral RNA concentration was higher in respiratory versus non-respiratory tissue samples in the first days or weeks after initial infection, differences diminished in patients who died several weeks to months after initial infection. This observation suggests lower or less efficient viral clearance in non-respiratory tissues, leading the authors to speculate that “understanding how SARS-CoV-2 evades immune detection is essential to guide future therapeutic approaches to facilitate viral clearance”.8 Other anatomopathological and immunological studies in adults have also shown that SARS-CoV-2 RNA and antigens can persist in the lung and extra-pulmonary areas,9, 10 even in immunocompetent adults.11 For example, two patients with a clinical diagnosis of long COVID who underwent surgery for other reasons 175 and 426 days after initial infection had evidence of SARS-CoV-2 RNA in the breast, appendix, and skin.12 More recently, authors profiled the plasma of 181 individuals with or without long COVID and uninfected controls, showing persistence of viral protein in long COVID concomitant immunological perturbations, such as evidence of proinflammatory and pro-fibrotic cytokines.13 Viral persistence has also been detected in immunocompromised patients,14 with evidence that the persistent virus can replicate, evolve, and even generate new variants. Similar findings have been documented also in non-human primates infected with SARS-CoV-2,15, 16 which further reinforces the evidence for SARS-CoV-2 RNA persistence in patients. These events have been shown in patients infected with pre-omicron variants; therefore, these findings might not be translated to patients infected with the COVID-19 omicron variant. However, as there is evidence that even patients infected with omicron develop long COVID,16, 17, 18 viral persistence should be considered as a possible hypothesis even in the newly infected patients, until proven otherwise. SARS-CoV-2 persistence in the paediatric population To understand whether SARS-CoV-2 can spread throughout the body in children younger than 18 years infected with SARS-CoV-2 and how long it can persist in the body, we searched PubMed for clinical studies focused on finding SARS-CoV-2 RNA, proteins, or antigens in tissues obtained during autopsy or organ biopsy done more than 24 h after the initial diagnosis of SARS-CoV-2. The search strategy and selection criteria are given in the (appendix p 1). After screening and selection of identified articles, 21 studies were included in the review.19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39 General characteristics of the included studies are reported in Table 1, Table 2, Table 3, Table 4 . Eight studies (38%) were done in the USA, five (23·8%) in Brazil, four (19%) in European countries (ie, Spain, Germany, and France), and two (9·5%) were multinational studies—one originated in (4·7%) in Thailand, and one (4·7%) in South Africa. The age of patients who underwent autopsies or tissue biopsies ranged from 1 day to 17 years.Table 1 Presence of SARS-CoV-2 in children with critical acute illness Country Sample size Age Sex Diagnosis of COVID-19 complication Baseline severity of COVID-19 Biopsy or autopsy Organs analysed Findings Time between acute infection and biopsy or autopsy Evidence of virus, viral fragments, or antibodies Gomes et al (2021)19 Brazil 1 14 months Female Critical COVID-19 Critical Autopsy CNS SARS-CoV-2 infection of brain tissue confirmed by RT-qPCR in fragments of the choroid plexus, lateral ventricle, and cortex Acute infection Viral RNA Poisson et al (2022)20 USA 1 8 years Female Critical COVID-19 Critical Autopsy CNS High SARS-CoV-2 IgM levels in the CSF 15 days Antibodies Mabena et al (2021)21 South Africa 12 35 days (median age) 58% males, 42% females Critical COVID-19 Critical Autopsy Lung, liver, heart, and brain SARS-COV-2 N1 and N2 detected in blood of 4 patients Acute infection N1 and N2 targets of the nucleocapsid gene Freij et al (2020)22 USA 1 5 years Female Critical COVID-19 Critical Autopsy Brain Cerebellar brain biopsy positive for SARS-CoV-2 RNA 35 days Viral RNA Menger et al (2022)23 Germany 1 4 years Female Critical COVID-19 Critical Autopsy Lung Presence of SARS-CoV-2 RNA 17 days Viral RNA Bhatnagar et al (2021)24 USA 4 <1 year; 17 years Not reported Critical COVID-19 Critical Autopsy Lung and trachea Presence of SARS-CoV-2 RNA in lungs of all the patients and in the trachea of 1 patient 1 −3 days Viral RNA Ninan et al (2021)25 USA 1 8 years Female Critical COVID-19 Critical Autopsy CSF, brain, bilateral lungs, blood SARS-CoV-2 RNA tested by RT-PCR in CSF, blood, brain tissue, and bilateral lungs was negative Acute infection Not found Duarte-Neto et al (2021)26 Brazil 2 8–12 years Female Critical COVID-19 Critical Autopsy Lungs, heart, liver, kidneys, spleen, brain, skin, and muscle RT-PCR for SARS-CoV-2 positive in the lungs of all patients, and in the heart of 1 patient Acute infection Viral RNA CSF=cerebrospinal fluid. Table 2 Persistence of SARS-CoV-2 in children Country Sample size Age Sex Diagnosis of COVID-19 complication Baseline severity of COVID-19 Biopsy or autopsy Organs analysed Findings Time between the acute infection and biopsy or autopsy Evidence of virus, viral fragments, or antibodies Taweevisit et al (2022)27 Thailand 1 5 years; 7 months Male MIS-C Asymptomatic Autopsy Heart, lungs, kidneys, liver, stomach, intestine, and brain Viral particles detected in heart, kidney, and intestine 2 months and 15 days Viral RNA Dolhnikoff et al (2020)28 Brazil 1 11 years Female MIS-C Not reported Autopsy Heart, lungs, kidney, liver, stomach, brain, inguinal lymph nodes, muscle, and skin SARS-CoV-2 RNA detected on a post-mortem nasopharyngeal swab and in cardiac and pulmonary tissues by RT-PCR by use of primers and probes set for E (envelope) gene 2 Not reported Viral RNA Duarte-Neto et al (2021)26 Brazil 3 8–12 years 2 female, 1 male MIS-C Not reported Autopsy Lung, heart, liver, kidneys, spleen, brain, bone marrow, colon, skin, and muscle RT-PCR for SARS-CoV-2 positive in lung and heart of patient 1; intestine and parotid of patient 2; viral particle detected in cardiomyocytes of patient 3 Not reported Viral RNA Mayordomo-Colunga et al (2022)29 Spain 1 12 years Male MIS-C Asymptomatc Autopsy Intestine, heart, lungs, and pericecal lymph nod Spike protein detected by immunofluorescence in intestine 6 weeks Spike protein Sigal et al (2022)30 USA 3 Not reported Not reported MIS-C Not reported Blood sample Blood Antigens detected in 3 (5·7%) of 53 samples >2 weeks Antigen N or S Arostegui et al (2022)31 USA 1 11 years Female Long COVID Mild Colonscopy Intestine SARS-CoV-2 nucleocapsid proteins detected in the intestinal lamina propria 3 months SARS-CoV-2 nucleocapsid proteins Scottoni et al (2022)32 Multinational 2 1 month; 5 months 1 male, 1 female Intussusception Mild Biopsy Lymph node and Ileum Immunofluorescence staining revealed the presence of SARS-CoV-2 in both the mesenteric lymph node from patient 1 and ileum from patient 2 5 days Virus Colmenero et al (2020)33 Spain 7 11–17 years 4 males; 3 females Chiblains Asymptomatic or mild Biopsy Skin of the toes Cytoplasmic granular positivity for SARS-CoV-2 spike protein shown in endothelial cells of the capillary and postcapillary venules of the upper dermis, and in epithelial cells of the secretory portion of eccrine units in all cases 4–30 days Spike protein Miura et al (2022)35 Brazil 48 Not reported Not reported Asymptomatic Asymptomatic Tonsillectomy Tonsils and adenoids SARS-CoV-2 genome detection rate was 20% in the tonsils, 16·27% in the adenoids, 10·41% of nasal cytobrushes, and 6·25% of nasal washes. IHC confirmed the presence of SARS-CoV-2 nucleoprotein in 15 of 16 positive tonsils samples, both in epithelium and lymphoid compartment Not reported Viral RNA Araùjo et al (2021)34 Brazil 1 17 years Female Guillain-Barrè syndrome Mild CSF CSF SARS-CoV-2 RNA in CSF 8 days Viral RNA Xu et al (2022)36 USA 110 1·7–21 years Female No complication Asymptomatic or mild Biopsy Tonsils and adenoids Pharyngeal tissues from COVID-19-convalescent children showed persistent expansion of germinal centre and antiviral lymphocyte populations associated with interferon-γ-type responses, particularly in the adenoids, and viral RNA in both tissues 25–303 days Viral RNA IHC=immunohistochemistry. MIS-C=multisystem inflammatory syndrome in children. CSF= cerebrospinal fluid. Table 3 Presence of SARS-CoV-2 in neonates and fetuses Country Sample size Age (years) Sex Diagnosis of COVID-19 complication Baseline severity of COVID-19 Biopsy or autopsy Organs analysed Findings Time between the acute infection and biopsy or autopsy Evidence of virus or viral fragments or antibodies Schwartz et al (2022)37 International 6 Fetuses No reported Maternal COVID in pregnancy Critical COVID-19 Autopsy Whole body 4 of 6 autopsied stillborns babies had SARS-CoV-2 in internal organs (lung, brain, kidney, and heart) Not reported Viral RNA Lesieur et al (2021)38 France 1 Fetus Female Maternal COVID-19 in pregnancy Mild Autopsy Thymus, lung, bronchial tree, stomach, spleen, adrenal gland, kidney, oesophagus, liver, heart, pancreas, and trachea RNA and spike protein found in lungs, and liver; RNA found in spleen and thymus; spike protein found in the stomach, heart, and lymph nodes 11 days Viral RNA and spike protein Reagan-Steiner et al (2022)39 2022 1 1 day Male Critical COVID-19 Critical COVID-19 Autopsy Whole body SARS-CoV-2 RNA detected in lung, airway, heart, liver, spleen, and kidney tissue by conventional RT-PCR; subgenomic RNA, suggesting SARS-CoV-2 replication, detected by subgenomic RT-PCR in lung, airway, heart, and liver tissue, but not in spleen or kidney tissue 6 days from maternal infection Viral RNA Table 4 Detection techniques and SARS-CoV-2 fragments isolated in paediatric studies, including vaccination status of included children Detection techniques RNA, antigens, or antibodies Protein Sample type Cell type Vaccination status Gomes et al (2021)19 Immunofluorescence NA SARS-CoV-2 spike protein Lung and brain tissue Pulmonary parenchymal cells, apical region of the ChP epithelium, and ependyma of the lateral ventricle Not vaccinated Gomes et al (2021)19 Immunofluorescence Viral double-stranded RNA NA Brain Lumina of ChP capillaries and medium size blood vessels .. Gomes et al (2021)19 RT-qPCR Nucleocapsid genes N1 and N2 NA Lung, brain (ChP, lateral ventricle, basal ganglia, and cerebellum), heart, kidney, liver, stomach, trachea, and larynx .. .. Poisson et al (2022)20 NA SARS-CoV-2 IgM antibodies NA CSF .. Not vaccinated Mabena et al (2021)21 RT-PCR N1 and N2 targets of the nucleocapsid gene NA Blood and lung .. Not vaccinated Frejj et al (2020)22 RT-PCR SARS CoV-2 RNA NA Brain (cerebellum) .. Not vaccinated Menger et al (2022)23 RT-PCR SARS CoV-2 RNA NA Lung .. Not vaccinated Bhatnagar et al (2021)24 RT-PCR N gene and S gene NA Lung and trachea .. 3 patients not vaccinated, 1 unknown Ninan et al (2021)25 RT-PCR Not found NA Blood, CSF, brain, and lung .. Not vaccinated Duarte-Neto et al (2021)26 Immunohistochemistry NA SARS-CoV-2 nucleocapsid protein and spike protein Lung, heart, liver, kidney, spleen, brain, fat tissue, bone marrow, and parotid Bronchiolar cells, type II pneumocytes, pulmonary megakaryocytes, cardiomyocytes, cardiac endothelial cells, hepatocytes and biliary tract epithelium, renal epithelial tubular cells, spleen (mononuclear cells in the red or white pulp), brain endothelial cells, perivascular astrocytes, sweat glands and subcutaneous nerves, microglia, bone marrow, mononuclear cells, parotid ductal, and acinar cells Not vaccinated Duarte-Neto et al (2021)26 RT-PCR SARS-CoV-2 RNA, nucleocapsid N gene, and envelope E gene NA Lung, heart, intestine, parotid .. .. Taweevisit et al (2022)27 Electron microscopy Viral particles NA Heart, kidney, and small bowel Cardiomyocytes, proximal tubular epithelial cells, and enterocytes Not vaccinated Dolhnikoff (2020)28 Electron microscopy Viral particles NA Heart Cardiomyocytes, endocardial endothelial cells, fibroblasts, and neutrophils Not vaccinated Dolhnikoff (2020)28 RT-PCR SARS-CoV-2 RNA, envelope gene NA Heart, kidney, and intestine .. .. Mayordomo-Colunga et al (2022)29 Immunofluorescence NA SARS-CoV-2 spike protein Intestine Cecum cells Not vaccinated Sigal et al (2022)30 Electrochemiluminescence immunoassay NA SARS-CoV-2 N and S Protein Blood sample .. Not vaccinated Arostegui et al (2022)31 Immunohistochemistry NA SARS-CoV-2 nucleocapsid proteins Colon (intestinal lamina propria) .. Not vaccinated Scottoni et al (2022)32 Immunofluorescence Viral double-stranded RNA Angiotensin-converting enzyme 2 SARS-CoV-2 nuclear protein Mesenteric lymph node and ileum .. Not vaccinated Colmenero et al (2020)33 Immunohistochemistry NA SARS-CoV-2 spike protein Skin Endothelial cells of the capillary and post-capillary venules of the upper dermis, epithelial cells of the secretory portion of eccrine units Not vaccinated Miura et al (2022)35 RT-PCR SARS-CoV-2 RNA .. SARS-CoV-2 genome detection rate: 20% in the tonsils, 16·27% in the adenoids, 10·41% of nasal cytobrushes, and 6·25% of nasal washes .. Not reported Miura et al (2022)35 Immunohistochemistry NA SARS-CoV-2 nucleoprotein Tonsils and adenoids Epithelium and lymphoid compartment As above Araùjo et al (2021)34 RT-PCR SARS-CoV-2 RNA NA Cerebrospinal fluid .. Not vaccinated Xu et al (2023)36 Droplet digital PCR SARS-CoV-2 nucleocapsid RNA NA Tonsils and adenoids .. Not vaccinated Schwartz et al (2022)37 RT-PCR SARS-CoV-2 RNA NA Lung, brain, kidney, and heart .. Not vaccinated Lesieur et al (2022)38 Immunohistochemistry NA SARS-CoV-2 envelope protein Lung and stomach Desquamated cells in alveolar spaces, alveolar cells, fibroblasts, granulocytic cells, and desquamated cells Not vaccinated Lesieur et al (2022)38 Immunohistochemistry NA SARS-CoV-2 spike protein Stomach, liver, and heart Cytoplasm of the stromal cells of the submucosae and the granulocytic cells, macrophages of the liver and lymph node, cytoplasm of pericardium, endothelial, and granulocytic cells As above Lesieur et al (2022)38 RT-PCR SARS-CoV-2 RNA NA Lung, spleen, liver, and trachea .. As above Reagan-Steiner (2022)39 Immunohistochemistry NA SARS-CoV-2 nucleocapsid and spike protein Lung and trachea Alveolar macrophages, type II pneumocytes, and hyaline membranes Not vaccinated Reagan-Steiner (2022)39 In situ hybridisation SARS-CoV-2 RNA NA Lung, heart, and liver Alveolar macrophages and pneumocyte, bronchiolar and submucosal gland epithelium, macrophages in lymphoid follicles in airway submucosa, and endothelial cells in myocardium vessel walls As above Reagan-Steiner (2022)39 RT-PCR SARS-CoV-2 RNA NA Lung, airway, heart, and liver .. As above CFS=cerebrospinal fluid. ChP=choroid plexus. NA=not applicable. Eight autopsy studies19, 20, 21, 22, 23, 24, 25, 26 described the anatomopathological findings and the laboratory tests done to detect SARS-CoV-2 in tissues of children and adolescents who died because of acute illness caused by COVID-19. Five studies (23·8%)2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 reported post-mortem findings in paediatric patients diagnosed with multisystem inflammatory syndrome in children. We included two articles (9·5%)31, 32 describing the presence of virus or its fragments in gastrointestinal systems of two patients who developed two different COVID-related complications—long-COVID and intussusception. One original article (4·7%)34 described a paediatric case of Guillain-Barrè syndrome with detection of SARS-CoV-2 in cerebrospinal fluid (CSF). Two autopsy studies (9·5%)37, 38 were done on stillborn babies who died because of maternal COVID-19 infection during pregnancy, and one (4·7%)39 was done on a neonate. According to two studies (4·7%),35, 36 the persistence of viral SARS-CoV-2 RNA was found in tonsils and adenoids of children who had asymptomatic acute infection and recovered. Presence of SARS-CoV-2 in children with critical acute illness Eight publications discussed the post-mortem histopathological findings and the detection of SARS-CoV-2 RNA by RT-PCR in tissues of children who underwent autopsy because of complications of fatal acute infection.19, 20, 21, 22, 23, 24, 25, 26 In six studies viral RNA was detected in various organs and tissues, including the CNS.19, 24 Gomes and colleagues19 described SARS-CoV-2 RNA-positive cells in fragments of choroid plexus, lateral ventricle, and cortex of a child aged 14 months who died of COVID-19 pneumonitis. Brain tissue infection was also reported by Freij and colleagues,22 who described the case of a girl aged 5 years with SARS-CoV-2 RNA and Mycobacterium tuberculosis complex DNA found in a cerebellar biopsy. Poisson and colleagues20 reported the case of a paediatric patient who developed a cerebral vasculitis secondary to SARS-CoV-2 infection; this hypothesis finding was supported by the anatomopathological evidence of parenchymal infarct, multifocal haemorrhages, and perivascular inflammatory infiltrates along with the presence of SARS-CoV-2 IgM in the CSF. We included a case report describing the post-mortem examination of a paediatric patient who died because of acute fulminant cerebral oedema in the setting of acute SARS-CoV-2 infection.25 In the case of this girl patient, no viral RNA was identified in samples of the CSF, blood, brain tissue, or lungs. Three studies documented viral infection of respiratory tissues, in particular lung and trachea, confirmed by RT-PCR in paediatric patients with COVID-19-related pneumonia.23, 24, 29 Persistence of SARS-CoV-2 in children 11 studies reported the persistence of virus, or parts of it, in children's tissues or biological fluids for weeks to months after the acute SARS-CoV-2 infection.26, 27 Time between the initial infection and the anatomopathological and microbiological examination ranged from 4 to 303 days.33, 36 Three studies described the post-mortem detection of SARS-CoV-2 RNA in tissues of paediatric patients diagnosed with multisystem inflammatory syndrome.26, 27, 28 In two studies on children diagnosed with multisystem inflammatory syndrome the laboratory tests done could detect only viral fragments: Mayordomo-Colunga and colleagues identified spike protein through immunofluorescence in the intestine of a boy aged 12 years; Sigal and colleagues29, 30 detected SARS-CoV-2 nucleocapsid and spike antigen in plasma of three patients, corresponding to 5·7% of the samples of multisystem inflammatory syndrome in paediatric patients analysed. We included one case report31 showing the presence of SARS-CoV-2 nucleocapsid proteins in the intestinal lamina propria of a girl with persistent gastrointestinal symptoms 3 months after the acute SARS-CoV-2 infection. In this patient, the mucosal biopsies of the colon identified a widespread lymphocytic infiltrate that could be related to the persistent viral infection. According to Scottoni and colleagues,32 SARS-CoV-2 was identified through immunofluorescence in a mesenteric lymph node and in the ileum of two young patients who underwent surgery for ileocaecal and ileocolic intussusception. Two studies described the persistence of SARS-CoV-2, confirmed by RT-PCR, in palatine tonsils and adenoids of children diagnosed with asymptomatic or mild acute infection.35, 36 Miura and colleagues35 reported that SARS-CoV-2 RNA was detected in 20% of the tonsils analysed, 16·27% of the adenoids, 10·4% of nasal cytobrushes, and 6·2% of nasal washes. They also did immunohistochemistry, neutralisation assay, and flow cytometry, which revealed CD123+ dendritic cells as the most common infected cells. Araùjo and colleagues reported the detection of SARS-CoV-2 RNA by RT-PCR in the CSF of a girl aged 17 years diagnosed with Guillain-Barrè syndrome related to the acute infection.34 Presence of SARS-CoV-2 in neonates and stillborn babies In two included studies,37, 38 the authors looked for SARS-CoV-2 RNA in tissues of stillborn fetuses who died after maternal SARS-CoV-2 infection during pregnancy. In a multinational case-based retrospective study,37 in four of six autopsied stillborn babies the viral RNA was present in their organs. In these stillborn babies, the most common anatomopathological findings were related to intrauterine hypoxia and asphyxia. Lesieur and colleagues38 described an in-utero fetal death at 24 weeks of gestation that occurred 7 days after the diagnosis of symptomatic acute infection in the mother. The anatomopathological examination revealed hepatocellular damage and hemosiderosis. Microbiological tests confirmed the presence of viral RNA in lung tissue, liver, spleen, and trachea. Furthermore, the immunohistochemistry for spike protein on stomach, liver, lymph node, and heart sample gave positive results. Steiner and colleagues39 reported autopsy findings from an extremely premature neonate who died 4 days after birth whose mother had severe acute COVID-19. Viral RNA was detected in neonatal heart and liver vascular endothelium through in-situ hybridisation and detected in different neonatal and placental samples by RT-PCR. The subgenomic RT-PCR positivity was suggestive of viral replication in the lungs, heart, and liver of the baby. Persistence of SARS-CoV-2 or its fragments and its possible biological effects Whether the persistence of SARS-CoV-2 RNA has biological effects is unknown; however, there is preliminary evidence that these particles can stimulate immune responses. Xu and colleagues36 collected blood, tonsils, and adenoids from 110 children who underwent tonsillectomy or adenoidectomy between September, 2020, and January, 2021, and had negative SARS-CoV-2 PCR results. They found expanded populations of lymphocytes expressing CXCR5 , which were located in the germinal centres. The populations included CXCR5+ CD8+ T cells and were similar to the progenitor cells that maintain antiviral responses in chronic viral infections.40, 41 The authors also found enrichment of various CD57+ T-cell populations, which are developed after repeated antigen exposure in chronic infections.42 Researchers have also shown that SARS-CoV-2 can infect monocytes and monocyte-derived macrophages without production of the infectious virus but preserving its infectivity.43, 44 These findings led the authors to speculate that these cells might act as spreaders for the virus in different body areas concealing the virus,43, 44 or that these infected immune cells might be a source of inflammation in long COVID.45 These findings raise the intriguing hypothesis that SARS-CoV-2 fragments can chronically stimulate local immune responses46 and, through unknown mechanisms, contribute to or be the major pathological event leading to symptoms, including myalgic encephalomyelitis or chronic fatigue syndrome, pains, and other symptoms that characterise long COVID, or even lead to uncontrolled inflammatory events of multisystem inflammatory syndrome in children.36 These data, although preliminary, are in line with the mounting evidence of dysregulated and persistent multiorgan inflammation in long COVID.47, 48 Studies to date have documented: highly activated innate immune signatures and higher expression of both type I and type III interferons, as well as CXCL9 and CXCL10 8 months after non-severe SARS-CoV-2 infection;49 a predictive signature of long COVID chronicity included elevated concentrations of plasma type II interferon and IL-2;50 self-reactive immune responses;51 and clonal and mutational maturation of memory B cells.10, 52 Chronic inflammatory events, which are subtle and undetected by routine laboratory tests such as those measuring C-reactive protein, might lead to tissue damage (including of the CNS), hypoxia-induced injury, abnormal activation of immune cells, and endothelial dysregulation.47, 53 Several studies by Pretorius and colleagues have shown the presence of circulating microclots entrapped with pro-inflammatory molecules in patients with long COVID which might play a key role in the severe spectrum of long COVID.54 Many other infections (including Ebola, Lassa, chikungunya, and influenza viruses) have been linked to long lasting immune stimulation; therefore, similar events might occur following infection with SARS-CoV-2.55 In children, data about immunological profile in long COVID compared with recovered groups are more scarce; however, we have published one study on this topic.56 In this pilot study, we documented that a subgroup of children with long COVID had a compromised ability to switch from the innate to adaptive immune response, which was shown by a contraction of the naive and switched B-cell compartments, and an unstable balance of regulatory T lymphocytes. Additionally, the expression of pro-inflammatory cytokines was not significantly different between children with long COVID and recovered children. In this cohort, viral persistence was not investigated, although the described immunological features could not allow us to rule out the presence of chronic immunological stimuli. Other discoveries made from 2022, to 2023, in children with multisystem inflammatory syndrome might further reinforce a possible biological role of SARS-CoV-2 persistence. Multisystem inflammatory syndrome in children is a well-defined, delayed-onset, COVID-19-related hyperinflammatory condition.57 Boribond and colleagues found that children with this condition have neutrophils characterised by a distinct granulocytic myeloid-derived suppressor cell signature with highly altered metabolism. These children also have extensive spontaneous neutrophil extracellular trap formation with neutrophil activation and degranulation signatures, all triggered by SARS-CoV-2 immune complexes. These findings suggest that persistent SARS-CoV-2 antigenaemia can trigger hyperinflammatory presentation during multi-system inflammatory syndrome in children57 and, as a consequence, possibly also in long COVID. This hypothesis is coherent with observation of myeloid-derived suppressor cells as part of the dysregulated immune responses observed in children with severe COVID-19,58 in other inflammatory conditions and viral respiratory infections,59 and also detected in mild and asymptomatic COVID-19 convalescents.60 In addition to previous immunological studies, observational studies have found an increase in newly onset immune-mediated diseases in children previously infected with SARS-CoV-2, such as type 1 diabetes.61 Although a clear link and causal effect between COVID-19 and type 1 diabetes has not been clearly proven, there is clinical plausibility for a diabetogenic effect of COVID-19. Therefore, further reinforcing a possible chronic effect of SARS-CoV-2 infection on the immune system needs further investigation. Consideration of antivirals for pharmacological trials Nirmatrelvir is an orally administered antiviral agent targeting the SARS-CoV-2 3-chymotrypsin-like cysteine protease enzyme, Mpro,62 which plays a pivotal role in the viral replication cycle (ie, in processing viral polyproteins into functional units).62 A placebo-controlled randomised trial documented an 89% risk reduction of progression to severe COVID-19, compared with placebo, in non-hospitalised patients at high risk. This finding led to the approval of nirmatrelvir in the USA in December, 2021. Since then, the medication has been used for millions of patients with acute SARS-CoV-2 infection. However, given the medication's antiviral activity, authors have questioned whether nirmatrelvir might affect development of long COVID, given the growing evidence of possible SARS-CoV-2 RNA spread and persistence in the body after acute infection. Although studies to address this question have not yet been done, the large numbers of patients already treated during acute SARS-CoV-2 infection are providing early indirect evidence of the plausibility of this hypothesis. Xie and colleagues in 2023, published the estimates of the effect of nirmatrelvir on a prespecified panel of 12 long COVID outcomes in 9217 treated patients versus 47 123 untreated controls. Both groups had risk factors for progression to severe disease.63 Treatment with nirmatrelvir was associated with reduced risk of long COVID (hazard ratio 0·74, 95% CI 0·69–0·81; absolute risk reduction 2·32, 1·73–2·91) including reduced risk of ten of 12 long COVID sequelae (including cardiovascular, coagulation, and haematological disorders, fatigue, liver disease, acute kidney disease, muscle pain, neurocognitive impairment, and shortness of breath). This study reinforces the hypothesis that better clearance of initial viral infection would be linked to lower long-term sequelae. Similarly, other authors have speculated that even in patients who have already developed long COVID (who were infected months before) it would be worth investigating the role of nirmatrelvir in the elimination of possible viral reservoir should be investigated. Peluso and colleagues published a case series of four patients whose symptoms improved with nirmatrelvir, suggesting that systematic study of antiviral therapy for long COVID is warranted.64 However, Peluso and colleague's finding should be interpreted with caution, because although RNA molecules are likely to be present months after infection, replicating virus is not. A study of patients at high risk of developing long COVID should be given priority over a systematic study of antiviral therapy. A trial with early treatment of those patients at higher risk of developing long COVID has a stronger rationale. Other treatment options are also promising, such as immune modulatory agents aimed at reducing chronic inflammation. The GNC-501 study (NCT05497089) will enrol 200 patients from Swiss and EU study centres, who have severe neuropsychiatric syndromes after COVID-19. After SARS-CoV-2 infection, some patients might have a chronic expression of the pathogenic W-ENV protein triggered by the SARS-CoV-2 infection. This expression is suspected to have a major role in the persistence of inflammation in many patients with long COVID patients (NCT05497089). Immunoglobulin G4 monoclonal antibody that targets the MSRV-Env protein can neutralise its action. The treatment has already shown promising results in a trial on multiple sclerosis, which is a well known disease, in which immunity in the CNS—and between 2021, and 2023, researchers showed that previous Epstein-Barr virus infection—plays a pathogenetic role. Although the immunoglobulin G4 monoclonal antibodies role in post-acute sequelae of SARS-CoV-2 infection is still preliminary, if proved successful, the use of immune-modulating agents for treating virus-induced chronic inflammation would further reinforce the hypothesis that viral persistence (or persistence of viral fragments) might play a pathogenetic role in long COVID. Other options including steroids,65 immunoglobulins,66 and plasmapheresis, have been tested in case series or are under investigations, although the evidence behind their effectiveness is weak and highly debated. Although vaccination might not be considered an immune modulatory treatment, it is worth nothing that a 2023 systematic review found that COVID-19 vaccines might have therapeutic effects on long COVID by boosting and rebalancing the immune system.67 However, the results of this study were based on low-quality studies and there are currently no studies addressing this possible role of COVID-19 vaccines in children. No studies reporting similar effects in children have been published. However, there are two publications describing three children with long COVID and chronic gastrointestinal symptoms, with evidence of persistent viral shedding. In two children, a gastrointestinal anti-inflammatory and immune stimulating agent was successful in eliminating the virus in the stool and symptoms resolution.29, 68 Even children with multisystem inflammatory syndrome, who are expected to have been infected with SARS-CoV-2 1–3 months before diagnosis, have been found to have persistent fecal SARS-CoV-2 positivity.69 These findings are not conclusive, but reinforce the hypothesis on the possible role of viral persistence and highlight the need for future studies on the topic. Conclusions from other viruses Chronic, unexplained persistence of signs and symptoms that negatively affect daily life of infected people are well known. Long-term effects of infetctions with several viruses, have been described long before COVID-19 emergence.70 Choutka and colleagues have analysed in detail the similarities between symptoms reported by patients with long COVID and those who survived other infections, including Lyme, Ebola, influenza, chikungunya, and many others. All patients reported a higher number of symptoms compared with control groups.70 These clinical observations provide evidence that the health of millions of people who have survived infections can be negatively affected for years, decades, or lifelong. Nevertheless, these patients were historically labelled as having psychiatric or psychosomatic conditions, with the negative consequences of negated access to care or research and absence of attention of funding agencies and companies. This situation continues, despite increasing evidence supporting the pathogenic role of neuroinflammation, which might be triggered by viral infections, in psychiatric diseases.71 Hopefully, long COVID will serve as a model to understand many of the currently unexplained chronic post-viral conditions. In addition to these subtle and uncharacterised post-viral conditions, there are several other better characterised complications, including Guillain-Barré syndrome, multisystem inflammatory syndrome in children, post-measles immune deficiency, and subacute sclerosing panencephalitis. Although subacute sclerosing panencephalitis has a very severe course and, so far, does not seem to share any neurological complications described after SARS-CoV-2 infection, the model of subacute scleoring panencephalitis is interesting and worthy of consideration. There is overwhelming evidence that most patients infected with measles do recover.72 Measles RNA has been detected in the peripheral blood mononuclear cells, urine, and the respiratory secretions in naturally infected children and animal models for several months after initial infection.72 These molecules can stimulate the immune system contributing to both life-long immunity after natural infection and immune system dysfunction.72 Persistent but anomalous measles particles have been detected in patients with subacute sclerosing panencephalitis—a condition that can be diagnosed years after initial measles infection.72 Measles virus is widely distributed in neurons and inflammation has been incontrovertibly documented in patients with subacute sclerosing panencephalitis.72 Evidence of neuronal loss can be seen on MRI and PET, and signs of focal metabolic abnormalities can also be identified with PET, even when conventional imaging is normal.73 However, measles persistence is also hypothesised to promote maturation of the immune response and development of life-long immunity.72 There is no evidence that SARS-CoV-2 infection can lead to consequences as severe as post-measles subacute sclerosing panencephalitis. However, we believe that researchers should not negate some similarities between measles and SARS-CoV-2. As in measles, SARS-CoV-2 RNA persistence has been documented, as has been its ability to stimulate the immune system. Also, brain MRI changes before and after infection, as well as PET abnormalities, have been shown in adults and children infected with SARS-CoV-2.73 Such similarities cannot be ignored, including the potential positive role of viral persistence in immune development and maturation, and should inspire future research—from long-term clinical follow-up to studying mechanisms of pathogenesis. Future perspectives RNA persistence after viral infection is well documented;55, 74 however, its contribution to health, immunity, and chronic diseases has not been fully elucidated. The immense scientific interest in SARS-CoV-2, including from funding agencies and companies, gives a unique opportunity to better understand the effect of viral persistence on humans from biological, clinical, and therapeutic perspectives (Figure 1, Figure 2 ).Figure 1 Proposed basic follow-up and clinical or diagnostic studies for children with SARS-CoV-2 infection or long COVID CPET=cardiopulmonary exercise testing. POTS= postural ortostatic tachicardia syndrome. Figure 2 Proposed placebo-controlled trials to treat or prevent long COVID in children, according to the viral persistence hypothesis Trials are ongoing in the USA. Patients can be treated if they have developed long COVID, or they have acute SARS-CoV-2 infection and risk factors to develop long COVID. From a clinical perspective, a randomised, placebo-controlled trial of early antivirals in adults and children at risk of long COVID seems justified by both available scientific evidence and the severe negative effect of long COVID on daily life of patients. Trials for children who have developed severe life-limiting long COVID would also be justified (currently registered trials in adults and children with PASC are available in a WHO platform).75 From a biological perspective, preclinical models should be developed to better understand the possible long-term effects of SARS-CoV-2 RNA persistence in the human body, and how these molecules can drive subtle, chronic inflammatory responses leading to disease. Only between 2021 and 2023, a strong link between Epstein-Barr virus and multiple sclerosis has been found, probably through reprogramming of latently infected B lymphocytes and the chronic presentation of viral antigens.76 The Epstein-Barr virus infected cells, the free virus, and its gene products have been found in the CNS,76 providing a good model for basic scientific research in the field of SARS-CoV-2 and PASC. Another key question is why only a small portion of people infected with common viruses are not able to recover from the virus and can develop long term sequelae. However, what is needed most urgently is a paradigm shift. Long COVID has been considered a psychosomatic condition just as several other conditions, including myalgic encephalomyelitis and chronic fatigue syndrome. The evidence of biological events in patients with long COVID is overwhelming and cannot be ignored. Rather, this evidence should inspire future research and a scientific and medical revolution in the whole field of post-viral chronic conditions. Recognition of long COVID should have practical consequences and be translated in real-world clinical settings, with patients being able to receive a formal diagnosis of long COVID by their family doctor or specialist. This step should be straightforward as a formal definition of PASC has been released by the WHO for both adults1 and children.2 Long COVID diagnosis might support patients to gain access to specialised centres, but also provide social support in terms of access to dedicated insurance policies or specific support for young people (eg, school certificates for personalised schedules for children with neurocognitive problems who struggle to return to their pre-COVID-19 performance and attendance). In our paediatric long COVID unit in Rome, we issue formal diagnosis of long COVID as well as certificates for school directors, teachers, and all other extracurricular activities of our patients. This is a pivotal step in the care of children with long COVID according to our experience, since having a supportive social environment is currently most important to support a child living with a chronic condition. Conclusions Evidence exists for the possible spread of SARS-CoV-2 spread into different organs and persistence for weeks to months after initial infection, even in children independently from severity of the acute disease. Viral RNA has been documented in children who have died from critical acute disease, but also in paediatric patients diagnosed with multisystem inflammatory syndrome weeks to months after previous asymptomatic or mild infection with SARS-CoV-2. Whether these events can also occur with new variants of SARS-CoV-2 or in previously vaccinated children is still unknown. Although the biological significance of the possibility for viral spread and persistence in children is unknown, the substantial evidence for it should not be neglected, and should inform future clinical, biological, and pharmacological studies. Declaration of interests DB has received grants from Pfizer and Roche Italy to study long COVID in children. DB has participated in a peer-to-peer teaching programme on COVID-19 vaccines and long COVID in children, sponsored by Pfizer. All other authors declare no competing interests. Supplementary Material Supplementary appendix Contributors DB conceptualised the report. FM implemented the research strategy. LM and RM did the literature review. PV supervised the study team. KF was the doctor with living long COVID experience, wrote her own perspective of living with long COVID, and did the language corrections. DB, PV, and LM wrote the first draft of the manuscript. All authors read and approved the final version of the manuscript. ==== Refs References 1 WHO A clinical case definition of post COVID-19 condition by a Delphi consensus https://www.who.int/publications/i/item/WHO-2019-nCoV-Post_COVID-19_condition-Clinical_case_definition-2021.1 2021 2 WHO A clinical case definition for post COVID-19 condition in children and adolescents by expert consensus https://www.who.int/publications/i/item/WHO-2019-nCoV-Post-COVID-19-condition-CA-Clinical-case-definition-2023-1 2023 3 Davis HE McCorkell L Vogel JM Topol EJ Long COVID: major findings, mechanisms and recommendations Nat Rev Microbiol 21 2023 133 146 36639608 4 Bhatnagar J Gary J Regan-Steiner S Evidence of severe acute respiratory syndrome coronavirus 2 replication and tropism in the lungs, airways, and vascular endothelium of patients with fatal coronavirus disease 2019: an autopsy case series J Infect Dis 223 2021 752 764 33502471 5 Dorward DA Russel CD Un IH Tissue-specific immunopathology in fatal COVID-19 Am J Respir Crit Care Med 203 2021 192 201 33217246 6 Schurink B Roos E Radonic T Viral presence and immunopathology in patients with lethal COVID-19: a prospective autopsy cohort study Lancet Microbe 1 2020 e290 e299 33015653 7 Varga Z Flammer AJ Steiger P Endothelial cell infection and endotheliitis in COVID-19 Lancet 395 2020 1417 1418 32325026 8 Stein SR Ramelli SC Grazioli A SARS-CoV-2 infection and persistence in the human body and brain at autopsy Nature 612 2022 758 763 36517603 9 Van Cleemput J van Snippenberg W Lambrechts L Organ-specific genome diversity of replication-competent SARS-CoV-2 Nat Commun 12 2021 6612 10 Gaebler C Wang Z Lorenzi JCC Evolution of antibody immunity to SARS-CoV-2 Nature 591 2021 639 644 33461210 11 Ceulemans LJ Khan M Yoo S-J Persistence of SARS-CoV-2 RNA in lung tissue after mild COVID-19 Lancet Respir Med 9 2021 e78 e79 34118186 12 Goh D Lim JCT Fernaındez SB Case report: persistence of residual antigen and RNA of the SARS-CoV-2 virus in tissues of two patients with long COVID Front Immunol 13 2022 939989 13 Schultheiß C Willscher E Paschold L Liquid biomarkers of macrophage dysregulation and circulating spike protein illustrate the biological heterogeneity in patients with post-acute sequelae of COVID-19 J Med Virol 95 2023 e28364 14 Hettle D Hutchings S Muir P Moran E Persistent SARS-CoV-2 infection in immunocompromised patients facilitates rapid viral evolution: retrospective cohort study and literature review Clin Infect Pract 16 2022 100210 15 Rutkai I Mayer MG Hellmers LM Neuropathology and virus in brain of SARS-CoV-2 infected non-human primates Nat Commun 13 2022 1745 16 Morioka S Tsuzuki S Suzuki M Post COVID-19 condition of the Omicron variant of SARS-CoV-2 J Infect Chemother 28 2022 1546 1551 35963600 17 Taquet M Sillett R Zhu L Neurological and psychiatric risk trajectories after SARS-CoV-2 infection: an analysis of 2-year retrospective cohort studies including 1 284 437 patients Lancet Psychiatry 9 2022 815 827 35987197 18 Magnusson K Kristoffersen DT Dell'Isola A Post-covid medical complaints following infection with SARS-CoV-2 Omicron vs Delta variants Nat Commun 13 2022 7363 19 Gomes I Karmirian K Oliveira JT SARS-CoV-2 infection of the central nervous system in a 14-month-old child: a case report of a complete autopsy Lancet Reg Health Am 2 2021 100046 20 Poisson KE Zygmunt A Leino D Lethal pediatric cerebral vasculitis triggered by severe acute respiratory syndrome coronavirus 2 Pediatr Neurol 127 2022 1 5 34864371 21 Mabena FC Baillie VL Hale MJ Clinical characteristics and histopathology of coronavirus disease 2019–related deaths in African children Pediatr Infect Dis J 40 2021 e323 e332 34397776 22 Freij BJ Gebara BM Tariq R Fatal central nervous system co-infection with SARS-CoV-2 and tuberculosis in a healthy child BMC Pediatr 20 2020 429 32907595 23 Menger J Apostolidou S Edler C Fatal outcome of SARS-CoV-2 infection (B1.1.7) in a 4-year-old child Int J Legal Med 136 2022 189 192 34510268 24 Bhatnagar J Gary J Reagan-Steiner S Evidence of severe acute respiratory syndrome coronavirus 2 replication and tropism in the lungs, airways, and vascular endothelium of patients with fatal coronavirus disease 2019: an autopsy case series J Infect Dis 223 2021 752 764 33502471 25 Ninan S Thompson P Gershon T Fatal pediatric COVID-19 case with seizures and fulminant cerebral edema Child Neurol Open 8 2021 2329048X21102253 26 Duarte-Neto AN Caldini E Gomes-Gouvea MS An autopsy study of the spectrum of severe COVID-19 in children: from SARS to different phenotypes of MIS-C EClinicalMedicine 35 2021 100850 27 Taweevisit M Chindamporn A Sujjavorakul K Samransamruajkit R Thorner PS Multisystem inflammatory syndrome in children (MIS-C) showing disseminated aspergillosis, cytomegalovirus reactivation and persistent SARS-CoV-2: case report with autopsy review Pathol Res Pract 238 2022 154106 28 Dolhnikoff M Ferreira Ferranti J de Almeida Monteiro RA SARS-CoV-2 in cardiac tissue of a child with COVID-19-related multisystem inflammatory syndrome Lancet Child Adolesc Health 4 2020 790 794 32828177 29 Mayordomo-Colunga J Vivanco-Allende A Lopez-Alonso S SARS-CoV-2 spike protein in intestinal cells of a patient with coronavirus disease 2019 multisystem inflammatory syndrome J Pediatr 243 2022 214 18e215 34843710 30 Sigal GB Novak T Mathew A Overcoming COVID-19 investigators. Measurement of severe acute respiratory syndrome coronavirus 2 antigens in plasma of pediatric patients with acute coronavirus disease 2019 or multisystem inflammatory syndrome in children using an ultrasensitive and quantitative immunoassay Clin Infect Dis 75 2022 1351 1358 35213684 31 Arostegui D Castro K Schwarz S Vaidy K Rabinowitz S Wallach T Persistent SARS-CoV-2 nucleocapsid protein presence in the intestinal epithelium of a pediatric patient 3 months after acute infection JPGN Reports 3 2022 e152 37168753 32 Scottoni F Giobbe GG Zambaiti E Intussusception and COVID-19 in infants: evidence for an etiopathologic correlation Pediatrics 149 2022 e2021054644 33 Colmenero I Santonja C Alonso-Riaño M SARS-CoV-2 endothelial infection causes COVID-19 chilblains: histopathological, immunohistochemical and ultrastructural study of seven paediatric cases Br J Dermatol 183 2020 729 737 32562567 34 Araújo NM Ferreira LC Dantas DP First report of SARS-CoV-2 detection in cerebrospinal fluid in a child with Guillain-Barré syndrome Pediatr Infect Dis J 40 2021 e274 e276 33990525 35 Miura CS Lima TM Martins RB Asymptomatic SARS-CoV-2 infection in children's tonsils Rev Bras Otorrinolaringol 88 2022 9 36 Xu Q Milanez-Almeida P Martins AJ Adaptive immune responses to SARS-CoV-2 persist in the pharyngeal lymphoid tissue of children Nat Immunol 24 2023 186 199 36536106 37 Schwartz DA Avvad-Portari E Babál P placental tissue destruction and insufficiency from COVID-19 causes stillbirth and neonatal death from hypoxic-ischemic injury Arch Pathol Lab Med 146 2022 660 676 35142798 38 Lesieur E Torrents J Fina F Congenital infection of severe acute respiratory syndrome coronavirus 2 with intrauterine fetal death: a clinicopathological study with molecular analysis Clin Infect Dis 75 2022 e1092 e1100 34553751 39 Reagan-Steiner S Bhatnagar J Martines RB Detection of SARS-CoV-2 in neonatal autopsy tissues and placenta Emerg Infect Dis 28 2022 510 517 35138244 40 He R Hou S Liu C Follicular CXCR5-expressing CD8+ T cells curtail chronic viral infection Nature 537 2016 412 416 27501245 41 Yu D Ye L A portrait of CXCR5+ follicular cytotoxic CD8+ T cells Trends Immunol 39 2018 965 979 30377045 42 Focosi D Bestagno M Burrone O Petrini M CD57+ T lymphocytes and functional immune deficiency J Leukoc Biol 87 2010 107 116 19880576 43 Percivalle E Sammartino JC Cassaniti I Macrophages and monocytes: “Trojan Horses” in COVID-19 Viruses 13 2021 2178 44 Matveeva O Nechipurenko Y Lagutkin D Yegorov YE Kzhyshkowska J SARS-CoV-2 infection of phagocytic immune cells and COVID-19 pathology: antibody-dependent as well as independent cell entry Front Immunol 13 2022 1050478 45 Patterson BK Francisco EB Yogendra R Persistence of SARS-CoV-2 S1 protein in CD16+ monocytes in post-acute sequelae of COVID-19 (PASC) up to 15 months post-infection Front Immunol 12 2022 746021 46 Kim TS Hufford MM Sun J Fu YX Braciale TJ Antigen persistence and the control of local T cell memory by migrant respiratory dendritic cells after acute virus infection J Exp Med 207 2010 1161 1172 20513748 47 Newell KL Waickman AT Inflammation, immunity, and antigen persistence in post-acute sequelae of SARS-CoV-2 infection Curr Opin Immunol 77 2022 102228 48 Peluso MJ Lu S Tang AF Markers of immune activation and inflammation in individuals with postacute sequelae of severe acute respiratory syndrome coronavirus 2 infection J Infect Dis 224 2021 1839 1848 34677601 49 Phetsouphanh C Darley DR Wilson DB Immunological dysfunction persists for 8 months following initial mild-to-moderate SARS-CoV-2 infection Nat Immunol 23 2022 210 216 35027728 50 Patterson BK Guevara-Coto J Yogendra R Immune-based prediction of COVID-19 severity and chronicity decoded using machine learning Front Immunol 12 2021 700782 51 Son K Jamil R Chowdhury A Circulating anti-nuclear autoantibodies in COVID-19 survivors predict long COVID symptoms Eur Respir J 61 2023 2200970 52 Poon MML Rybkina K Kato Y SARS-CoV-2 infection generates tissue-localized immunological memory in humans Sci Immunol 6 2021 eabl9105 53 Chioh FW Fong S-W Young BE Convalescent COVID-19 patients are susceptible to endothelial dysfunction due to persistent immune activation eLife 10 2021 e64909 54 Kruger A Vlok M Turner S Proteomics of fibrin amyloid microclots in long COVID/post-acute sequelae of COVID-19 (PASC) shows many entrapped pro-inflammatory molecules that may also contribute to a failed fibrinolytic system Cardiovasc Diabetol 21 2022 190 36131342 55 Hirschenberger M Hunszinger V Sparrer KMJ Implications of innate immunity in post-acute sequelae of non-persistent viral infections Cells 10 2021 2134 56 Buonsenso D Valentini P De Rose C Recovering or persisting: the immunopathological features of SARS-CoV-2 infection in children J Clin Med 11 2022 4363 57 Boribong BP LaSalle TJ Bartsch YC Neutrophil profiles of pediatric COVID-19 and multisystem inflammatory syndrome in children Cell Rep Med 3 2022 100848 58 Bline K Andrews A Moore-Clingenpeel M Myeloid-derived suppressor cells and clinical outcomes in children with COVID-19 Front Pediatr 10 2022 893045 59 Koushki K Salemi M Miri SM Arjeini Y Keshavarz M Ghaemi A Role of myeloid-derived suppressor cells in viral respiratory infections; hints for discovering therapeutic targets for COVID-19 Biomed Pharmacother 144 2021 112346 60 Siemińska I Węglarczyk K Surmiak M Mild and asymptomatic COVID-19 convalescents present long-term endotype of immunosuppression associated with neutrophil subsets possessing regulatory functions Front Immunol 12 2021 748097 61 Kendall EK Olaker VR Kaelber DC Xu R Davis PB Association of SARS-CoV-2 infection with new-onset type 1 diabetes among pediatric patients from 2020 to 2021 JAMA Netw Open 5 2022 e2233014 62 Owen DR Allerton CMN Anderson AS An oral SARS-CoV-2 Mpro inhibitor clinical candidate for the treatment of COVID-19 Science 374 2021 1586 1593 34726479 63 Xie Y Choi T Al-Aly Z Association of treatment with nirmatrelvir and the risk of post-COVID-19 condition JAMA Intern Med 2023 e230743 64 Peluso MJ Anglin K Durstenfeld MS Effect of oral nirmatrelvir on long COVID symptoms: 4 cases and rationale for systematic studies Pathog Immun 7 2022 95 103 35800257 65 Goel N Goyal N Nagaraja R Kumar R Systemic corticosteroids for management of ‘long-COVID’: an evaluation after 3 months of treatment Monaldi Arch Chest Dis 92 2021 66 Thompson JS Thornton AC Ainger T Garvy BA Long-term high-dose immunoglobulin successfully treats Long COVID patients with pulmonary, neurologic, and cardiologic symptoms Front Immunol 13 2023 1033651 67 Byambasuren O Stehlik P Clark J Alcorn K Glasziou P Effect of covid-19 vaccination on long covid: systematic review BMJ Med 2 2023 e000385 68 Morello R De Rose C Cardinali S Valentini P Buonsenso D Lactoferrin as possible treatment for chronic gastrointestinal symptoms in children with long COVID: case series and literature review Children 9 2022 1446 69 Parodi E Carpino A Franchitti E Detection of faecal SARS-CoV-2 RNA in a prospective cohort of children with multisystem inflammatory syndrome (MIS-C) Epidemiol Prev 45 2021 522 527 34841834 70 Choutka J Jansari V Hornig M Iwasaki A Unexplained post-acute infection syndromes Nat Med 28 2022 911 923 35585196 71 Yao XP Ye J Feng T Adaptor protein MyD88 confers the susceptibility to stress via amplifying immune danger signals Brain Behav Immun 108 2023 204 220 36496170 72 Griffin DE Measles virus persistence and its consequences Curr Opin Virol 41 2020 46 51 32387998 73 Cocciolillo F Di Giuda D Morello R De Rose C Valentini P Buonsenso D Orbito-frontal cortex hypometabolism in children with post-COVID condition (long COVID): a preliminary experience Pediatr Infect Dis J 41 2022 663 665 35839175 74 Griffin DE Why does viral RNA sometimes persist after recovery from acute infections? PLoS Biol 20 2022 e3001687 75 Fawzy NA Shaar BA Taha R A systematic review of trials currently investigating therapeutic modalities for post-acute COVID-19 syndrome and registered on World Health Organization international clinical trials platform Clin Microbiol Infect 29 2023 570 577 36642173 76 Soldan SS Lieberman PM Epstein-Barr virus and multiple sclerosis Nat Rev Microbiol 21 2023 51 64 35931816
PMC010xxxxxx/PMC10292826.txt	==== Front Lancet Microbe Lancet Microbe The Lancet. Microbe 2666-5247 The Author(s). Published by Elsevier Ltd. S2666-5247(23)00185-4 10.1016/S2666-5247(23)00185-4 Correspondence Ongoing sporadic monkeypox cases: neutralising antibody detection in asymptomatic individuals Moschese Davide a Giacomelli Andrea a Mileto Davide b on behalf of the Sacco Mpox Study GroupANTINORI SPINELLO MD BIANCHI MICOL CAPETTI AMEDEO FERIDNANDO MD CARUSO FRANCESCO MD COLOMBO MARTINA LAURA MD COSSU MARIA VITTORIA MD CUTRERA MIRIAM GISMONDO MARIA RITA GORI ANDREA MD LAZZARIN SAMUEL MD LOMBARDI ALESSANDRA MICHELI VALERIA PASTENA ANDREA MD POZZA GIACOMO MD RIZZARDINI GIULIANO MD RIZZO ALBERTO SABAINI FEDERICO MD SALARI FEDERICA a Department of Infectious Diseases, Luigi Sacco University Hospital, ASST Fatebenefratelli Sacco, 20157 Milan, Italy b Laboratory of Clinical Microbiology, Virology and Bioemergencies, Luigi Sacco University Hospital, ASST Fatebenefratelli Sacco, 20157 Milan, Italy 26 6 2023 26 6 2023 © 2023 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND 4.0 license 2023 Elsevier has created a Monkeypox Information Center (https://www.elsevier.com/connect/monkeypox-information-center) in response to the declared public health emergency of international concern, with free information in English on the monkeypox virus. The Monkeypox Information Center is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its monkeypox related research that is available on the Monkeypox Information Center - including this research content - immediately available in publicly funded repositories, with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the Monkeypox Information Center remains active. ==== Body pmcIn their Correspondence, Silvia Accordini and colleagues1 described the case of a 35-year-old man who used pre-exposure prophylaxis (PrEP) for HIV who, after a household contact with a person with mpox (formerly known as monkeypox), did not develop any skin lesion, but tested positive for MPXV and subsequently had seroconversion (indirect immunofluorescence) at 30 days follow-up.1 To support the authors speculation, we present the case of a 35-year-old man who has sex with men who uses PrEP without a previous mpox or smallpox vaccination history, whose polyamorous partner was diagnosed with mpox. The day after the diagnosis, while completely asymptomatic, our patient underwent a full physical examination that did not detect any sign of infection. The patient was told to seek medical attention at any sign or symptom of infection in the 21 days following the last contact with their partner and to refrain from sexual activity. After 4 days, the patient returned reporting pubic pustules in absence of any other symptom. The presence of MPXV DNA was investigated on pharynx and lesion swabs by RT-PCR test. He also underwent a full sexually transmitted infections screen and HSV-1/2 DNA on lesion sample. All tests came back negative with the exception of HSV-1 DNA on lesion swab and Neisseria gonorrhoeae on anal sample; the patient was thus prescribed with oral valacyclovir and ceftriaxone, and reported full healing of the lesions. The patient did not develop any other symptoms for the whole surveillance period. After 1 month, a serological test was proposed to look for neutralising antibodies against MPXV using the in vitro plaque reduction neutralisation test. A serum sample of the patient was diluted and mixed with MPXV suspension on Vero E6 cell culture. The test was positive for mpox-virus neutralising antibodies with a titre of 1:160. Reda and colleagues2 called for greater attention on infection control, including individuals who are asymptomatic and paucisymptomatic. Nevertheless, Van Dijck and colleagues3 underscored how unrecognised signs or unreported symptoms account for almost all the cases of what were believed to be asymptomatic infections. Our experience, corroborated by an active surveillance and accurate physical examination, supports the findings by Accordini and colleagues1 in reinforcing the speculations that, not only individuals with atypical presentations, but also people who are completely asymptomatic might have a role in spreading and sustaining MPXV circulation, justifying the persistent sporadic occurrence of mpox cases nowadays, in a phase of apparent remission of the mpox epidemic. We declare no competing interests. ==== Refs References 1 Accordini S Cordioli M Pomari E Tacconelli E Castilletti C People with asymptomatic or unrecognised infection potentially contribute to monkeypox virus transmission Lancet Microbe 4 2023 e209 36563704 2 Reda A El-Qushayri AE Shah J Asymptomatic monkeypox infection: a call for greater control of infection and transmission Lancet Microbe 4 2023 e15 e16 36209756 3 Van Dijck C De Baetselier I Kenyon C Mpox screening in high-risk populations finds no asymptomatic cases Lancet Microbe 4 2023 e132 e133 36509096
PMC010xxxxxx/PMC10292827.txt	==== Front Lancet Rheumatol Lancet Rheumatol The Lancet. Rheumatology 2665-9913 Elsevier Ltd. S2665-9913(23)00127-3 10.1016/S2665-9913(23)00127-3 Articles Post-COVID condition in patients with inflammatory rheumatic diseases: a prospective cohort study in the Netherlands Boekel Laura BSc a* Atiqi Sadaf MD a Leeuw Maureen MSc a Hooijberg Femke MD a Besten Yaëlle R. BSc a Wartena Rosa BSc a Steenhuis Maurice PhD d Vogelzang Erik PhD e Webers Casper PhD fg Boonen Annelies Prof PhD fg Gerritsen Martijn PhD a Lems Willem F Prof MD ab Tas Sander W Prof MD c van Vollenhoven Ronald F Prof MD c Voskuyl Alexandre E Prof PhD b van der Horst-Bruinsma Irene Prof MD h Nurmohamed Mike Prof MD ab Rispens Theo PhD d Wolbink Gertjan PhD ad a Department of Rheumatology, Amsterdam Rheumatology and Immunology Center, Reade, Amsterdam University Medical Center, Amsterdam, Netherlands b Department of Rheumatology and Clinical Immunology, Amsterdam Rheumatology and Immunology Center, VU University Medical Center, Amsterdam University Medical Center, Amsterdam, Netherlands c Department of Rheumatology and Clinical Immunology, Amsterdam Rheumatology and Immunology Center, University of Amsterdam, Amsterdam University Medical Center, Amsterdam, Netherlands d Department of Immunopathology, Sanquin Research and Landsteiner Laboratory, Amsterdam University Medical Center, Amsterdam, Netherlands e Department of Medical Microbiology and Infection Control, Amsterdam University Medical Center, Amsterdam, Netherlands f Department of Internal Medicine, Division of Rheumatology, Maastricht University Medical Center, Maastricht, Netherlands g Care and Public Health Research Institute (CAPHRI), Maastricht University, Maastricht, Netherlands h Department of Rheumatic Diseases, Radboud, University Medical Center, Nijmegen, Netherlands * Correspondence to: Laura Boekel, Department of Rheumatology, Amsterdam Rheumatology and Immunology Center, Reade, Amsterdam University Medical Center, 1056 AB, Amsterdam, Netherlands 31 5 2023 7 2023 31 5 2023 5 7 e375e385 © 2023 Elsevier Ltd. All rights reserved. 2023 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Studies on long-term consequences of COVID-19, commonly referred to as post-COVID condition, in patients with inflammatory rheumatic diseases are scarce and inconclusive. Furthermore, classifying patients with inflammatory rheumatic diseases as having post-COVID condition is complicated because of overlapping symptoms. Therefore, we investigated the risk of post-COVID condition and time until recovery, and compared the prevalence of symptoms seen in post-COVID condition, between patients with inflammatory rheumatic diseases and healthy controls, with and without a history of COVID-19. Methods In this substudy we used data from an ongoing prospective cohort study in the Netherlands. All adult patients with inflammatory rheumatic diseases from the Amsterdam Rheumatology and Immunology Center in Amsterdam, the Netherlands, were invited to participate in the study between April 26, 2020, and March 1, 2021. All patients were asked, but not obliged, to recruit their own control participant of the same sex, of comparable age (< 5 years), and without an inflammatory rheumatic disease. Demographic and clinical data, including data on the occurrence of SARS-CoV-2 infections, were collected via online questionnaires. On March 10, 2022, all study participants received a questionnaire on the occurrence, onset, severity, and duration of persistent symptoms during the first 2 years of the COVID-19 pandemic, independent of their history of SARS-CoV-2 infection. Additionally, we prospectively monitored a subset of participants who had a PCR or antigen confirmed SARS-CoV-2 infection in the 2-month period surrounding the questionnaire in order to assess COVID-19 sequelae. In line with WHO guidelines, post-COVID condition was defined as persistent symptoms that lasted at least 8 weeks, started after the onset and within 3 months of a PCR or antigen-confirmed SARS-CoV-2 infection, and could not be explained by an alternative diagnosis. Statistical analyses included descriptive statistics, logistic regression analyses, logistic-based causal mediation analyses, and Kaplan-Meier survival analyses for time until recovery from post-COVID condition. In exploratory analyses, E-values were calculated to investigate unmeasured confounding. Findings A total of 1974 patients with inflammatory rheumatic disease (1268 [64%] women and 706 [36%] men; mean age 59 years [SD 13]) and 733 healthy controls (495 [68%] women and 238 [32%] men; mean age 59 years [12]) participated. 468 (24%) of 1974 patients with inflammatory rheumatic disease and 218 (30%) of 733 healthy controls had a recent SARS-CoV-2 omicron infection. Of those, 365 (78%) of 468 patients with inflammatory rheumatic disease and 172 (79%) of 218 healthy controls completed the prospective follow-up COVID-19 sequelae questionnaires. More patients than controls fulfilled post-COVID condition criteria: 77 (21%) of 365 versus 23 (13%) of 172 (odds ratio [OR] 1·73 [95% CI 1·04–2·87]; p=0·033). The OR was attenuated after adjusting for potential confounders (adjusted OR 1·53 [95% CI 0·90–2·59]; p=0·12). Among those without a history of COVID-19, patients with inflammatory diseases were more likely to report persistent symptoms consistent with post-COVID condition than were healthy controls (OR 2·52 [95% CI 1·92–3·32]; p<0·0001). This OR exceeded the calculated E-values of 1·74 and 1·96. Recovery time from post-COVID condition was similar for patients and controls (p=0·17). Fatigue and loss of fitness were the most frequently reported symptoms in both patients with inflammatory rheumatic disease and healthy controls with post-COVID condition. Interpretation Post-COVID condition after SARS-CoV-2 omicron infections was higher in patients with inflammatory rheumatic disease than in healthy controls based on WHO classification guidelines. However, because more patients with inflammatory rheumatic disease than healthy controls without a history of COVID-19 reported symptoms that are commonly used to define a post-COVID condition during the first 2 years of the pandemic, it is likely that the observed difference in post-COVID condition between patients and controls might in part be explained by clinical manifestations in the context of underlying rheumatic diseases. This highlights the limitations of applying current criteria for post-COVID condition in patients with inflammatory rheumatic disease, and suggests it might be appropriate for physicians to keep a nuanced attitude when communicating the long-term consequences of COVID-19. Funding ZonMw (the Netherlands organization for Health Research and Development) and Reade foundation. ==== Body pmcIntroduction Effects of inflammatory rheumatic diseases and treatment with immunosuppressive agents on the acute infection phase of SARS-CoV-2 have been extensively investigated, but there is still a paucity of data on long-term consequences of SARS-CoV-2 infection. Long-term consequences of COVID-19 are commonly referred to as post-COVID condition (also known as long COVID), and are currently seen as conditions that might affect a substantial proportion of the general population.1, 2 However, post-COVID condition is still poorly defined; it can include any subjective symptom persisting beyond the acute infection phase of SARS-CoV-2, if other possible causes can be ruled out.3 A large amount of heterogeneity exists among studies that have investigated post-COVID condition, and findings of epidemiological studies are still largely inconclusive. Pathophysiological mechanisms underlying post-COVID condition also remain incompletely understood, although recent studies suggest that persistent inflammation after the acute infection phase and auto-immune reactions might play a crucial role in its development.4 The presence of underlying rheumatic diseases that are characterised by chronic inflammation and immune dysregulation might therefore increase the risk of post-COVID condition or affect the clinical presentation and severity.5 However, whether patients with inflammatory rheumatic diseases are more susceptible to post-COVID condition, and whether the clinical phenotype differs from people in the general population, remains unclear. Additionally, studies have not yet investigated whether correctly classifying patients with inflammatory rheumatic diseases as having post-COVID condition cases is complicated by increased background noise; for example, pre-existing symptoms or new symptoms that are not caused by post-COVID conditions (eg, caused by the underlying rheumatic disease). Epidemiological studies investigating these research questions might contribute to understanding post-COVID condition and help rheumatologists to inform and guide their patients during the ongoing COVID-19 pandemic. Therefore, we aimed to compare the incidence and characteristics of post-COVID condition in patients with inflammatory rheumatic diseases and healthy controls during the period when the omicron variant (BA.1 and BA.2 lineages) was dominant, as part of a large prospective COVID-19 focused cohort study, while accounting for shared symptoms between post-COVID condition and inflammatory rheumatic diseases. Methods Study design and participants In this study, we collected data from participants enrolled in a Dutch prospective cohort study that was designed to compare the disease severity of COVID-19 between patients with inflammatory rheumatic diseases and healthy controls (Netherlands Trial Register, trial ID NL 8513). The design of the study has been previously described.6, 7, 8, 9 Briefly, patients aged 18 years and older with inflammatory rheumatic diseases from the Amsterdam Rheumatology and Immunology Center (Amsterdam, Netherlands), were invited to participate in the study between April 26, 2020, and March 1, 2021. All participants were asked (but not obliged) to recruit their own healthy control participant of the same sex, comparable age (difference of <5 years) and without an inflammatory rheumatic disease. The research protocol was approved by the medical ethical committee of the VU University Medical Center (registration number, 2020.169). All participants provided written informed consent. Research in context Evidence before this study We searched PubMed and Google Scholar on March 8, 2023, for studies published in English since Jan 1, 2020, that describe long-term consequences of COVID-19 in patients with inflammatory rheumatic diseases. Long-term consequences are commonly referred to as post-COVID condition or long-COVID and could affect a substantial proportion of the population. However, post-COVID conditions are still poorly defined, and pathophysiological mechanisms underlying post-COVID condition remain incompletely understood. Hence, it is still unknown whether patients with inflammatory rheumatic diseases are more susceptible to post-COVID condition and whether the clinical phenotype differs from people from the general population. Added value of this study To our knowledge, this is the first prospective study to compare long-term consequences of COVID-19 between patients with rheumatic inflammatory diseases and healthy controls. A unique aspect of this study is that participants were enrolled before infection with SARS-CoV-2, which minimises the risk of selection bias. Additionally, we addressed and quantified systematic differences in the reporting of persistent symptoms used to define post-COVID condition between patients and controls. This is relevant because overlap in the clinical picture of post-COVID condition and inflammatory rheumatic diseases could complicate the correct identification of post-COVID condition in patients with inflammatory rheumatic diseases, and thus bias comparisons with healthy controls. We found that more patients with inflammatory rheumatic diseases than healthy controls developed post-COVID condition when we applied WHO criteria, but symptomology and time to recovery from post-COVID condition were similar between the groups. Furthermore, we observed that patients with inflammatory rheumatic diseases without a history of COVID-19 were more likely to report persistent symptoms that are used to identify post-COVID condition than were healthy controls. Implications of all the available evidence Our data emphasise the limitations of applying current WHO criteria of post-COVID condition in patients with inflammatory rheumatic diseases, which implies that studies on this topic should be interpreted with caution. Additionally, we believe it might be appropriate for rheumatologists to have a nuanced attitude when communicating information regarding long-term consequences of COVID-19 to their patients. Procedures Demographic data were collected at baseline and included age, sex, height, weight, smoking status, autoimmune disease type, and educational level. At baseline and during follow-up, participants reported their disease activity, medication use, and COVID-19-related clinical characteristics. Demographic and clinical data were collected with online questionnaires distributed via email. Sex data were self-reported, participants could choose between “male”, “female”, and “other”. In case of “other”, participants could explain their answer in an open text field. The baseline questionnaire was sent to participants when they were included in the study. Until the start of the primary vaccination campaign in the Netherlands (January, 2021), the first and second follow-up questionnaires were sent to participants 1–4 months and 5–9 months after completion of the baseline questionnaire. Following this, questionnaires were then sent to participants at fixed timepoints on April 26, Aug 24, and Dec 10, 2021, and on March 10, 2022. Data collected included vaccination dates, vaccine type, information on COVID-19 symptoms, and admissions to a hospital due to COVID-19. In the questionnaire sent on March 10, 2022, participants were also asked about the occurrence, onset, severity, and duration of a series of one or more pre-defined symptoms characteristic of post-COVID condition during the first 2 years of the COVID-19 pandemic (from Jan 1, 2020, to Dec 31, 2021), independent of their history with SARS-CoV-2 infections (appendix pp 11–12). Additionally, participants of the March 10, 2022 questionnaire with a PCR or antigen-confirmed SARS-CoV-2 infection between Jan 1 and April 25, 2022, a time period that coincided with the omicron dominant period in the Netherlands, were included for prospective monitoring of the occurrence of post-COVID condition and recovery (appendix pp 11–12). These participants received additional questionnaires at fixed timepoints: on June 25 and Sept 20, 2022. The questions were similar to those in the March 10, 2022 questionnaire, except for the obligatory reporting of new onset of symptoms with SARS-CoV-2 infection; participants were specifically asked about new onset symptoms after their SARS-CoV-2 infection. The second questionnaire was only sent to participants whose symptoms had not yet resolved at the time of completing the first questionnaire. Serum samples were collected multiple times during follow-up for analyses of SARS-CoV-2 antibodies via regular blood sampling at the local research institute or via a finger prick at home.10, 11 Sampels were collected in October, 2020, and January, 2021, after the first SARS-CoV-2 vaccination, after the second SARS-CoV-2 vaccination, and in October–November, 2021. Serum samples that were collected before the first SARS-CoV-2 vaccination were used to identify participants with a history of COVID-19 before vaccination. Shortly before the start of the Dutch COVID-19 vaccination campaign, all participants were invited for blood sampling to cross-sectionally screen the cohort for COVID-19 cases.7 All pre-vaccination serum samples were analysed for the presence of SARS-CoV-2 specific antibodies with a receptor binding domain antibody bridging ELISA (in house) with a 98·1% sensitivity and a 99·5% specificity.10, 11 Outcomes The primary outcome of this study was to compare the risk of post-COVID condition after a SARS-CoV-2 omicron infection between patients with inflammatory rheumatic diseases and healthy controls, using WHO guidelines for classification of cases with post-COVID condition. The primary outcome was asssessed in all participants with a confrimed SARS-CoV-2 infection during the omicron-dominant period (Jan 1–April 25, 2022), with available follow-up data after the infection. A SARS-CoV-2 omicron infection was defined as a self-reported positive SARS-CoV-2 antigen or PCR test. SARS-CoV-2 infections after Jan 1, 2022, were assumed to be due to the omicron variant, because it was the dominant variant during the study period in the Netherlands. In line with WHO guidelines,3 participants with post-COVID condition were defined as those who reported any symptom that lasted at least 8 weeks, started after the onset and within 3 months of a PCR or antigen-confirmed SARS-CoV-2 infection, and could not be explained by an alternative diagnosis. Secondary objectives were to compare disease characteristics of post-COVID condition, recovery from post-COVID condition (defined as time to complete resolution of symptoms), and health-care utilisation after SARS-CoV-2 omicron infections between patients with inflammatory rheumatic diseases and healthy controls. As a further secondary outcome, we aimed to investigate whether correctly classifying patients with inflammatory rheumatic diseases as having post-COVID condition is complicated by the occurrence of persistent symptoms that could be attributed to both post-COVID condition and inflammatory rheumatic disease. To investigate this, we compared the prevalence of persistent symptoms (≥8 weeks) that are observed in post-COVID condition between patients with inflammatory rheumatic disease and healthy controls with and without a history of COVID-19. Participants could select the following symptoms: loss of fitness, shortness of breath, memory problems, concentration problems, depression, cough, chest pain, palpitations, fatigue, sleeping problems, headache, myalgia, increased arthralgia, skin rash, excessive sweating, fever, loss of taste or smell, general malaise, rhinorrhea, and other (a free text box for participants to list symptoms not in the questionnaire). Statistical analysis Participants were included for the retrospective study if they completed the questionnaire sent on March 10, 2022. Characteristics of participants are presented as mean (SD), median (IQR), or frequencies and proportions, depending on the type and distribution of the data. Analyses on post-COVID condition after SARS-CoV-2 omicron infections only included participants who were prospectively monitored after a SARS-CoV-2 omicron infection and completed the first COVID-19 sequelae questionnaire (sent on June 25, 2022). Univariable and multivariable logistic regression analyses were used to compare the risk of developing post-COVID condition after a SARS-CoV-2 omicron infection between patients with inflammatory rheumatic disease and healthy controls. Confounding was investigated for age, sex, BMI, cardiovascular disease, diabetes, chronic pulmonary disease, vaccination status, history of COVID-19 before Jan 1, 2022, and disease severity of the acute infection phase of SARS-CoV-2. We hypothesised that disease severity of the acute infection phase of SARS-CoV-2 could be a mediator in the association between participant status (patient with inflammatory rheumatic disease vs healthy control) and post-COVID condition (appendix p 9), as it has been demonstrated that disease severity was an important risk factor for post-COVID condition in people with mild COVID-19.12 We used a regression-based approach for causal mediation analyses to investigate this. Loglinear regression models were used for the outcome variable because the rare-outcome assumption (prevalence <0·10) was violated. Both crude and adjusted direct effects (the effect of participant status on post-COVID condition not mediated through disease severity), indirect effects (the effect of participant status on post-COVID condition mediated through disease severity), and total effects (the combined effect of the direct and indirect effect) were estimated. Adjusted models included age, sex, and presence of comorbidities (obesity; cardiovascular disease; pulmonary disease or diabetes) as covariates. In each model, age was conditioned at 55 years (the median), and the presence of comorbidities was conditioned on absent. For descriptive purposes, models with sex conditioned on males and females were reported separately. There was no determinant-mediator interaction, which meant that the association between participant status and post-COVID condition did not differ for different values of disease severity (WHO COVID-19 severity scores <3 vs ≥3) For each model, the proportion mediated, which quantifies the contribution of mediation to the total effect, was calculated once as: pure natural indirect effect/total effect. Exploratory analyses were done to investigate unmeasured confounding. We hypothesised that the occurrence of clinical manifestations that could be attributed to both post-COVID condition and rheumatic diseases can introduce systematic (unmeasured) differences in the reporting of persistent symptoms between patients with inflammatory rheumatic disease and healthy controls, with patients reporting more symptoms than controls. The E-value is an approach to help investigate the robustness of the main study result by calculating the minimum strength that unmeasured confounding must have to negate the association between the exposure and outcome, and evaluating whether that magnitude is plausible.13 We therefore calculated E-values for the association between participant status and post-COVID condition. Subsequently, we used our retrospective data asking about persistent symptoms during the first 2 years of the COVID-19 pandemic to quantify the difference between patients with inflammatory rheumatic disease and healthy controls. For both patients and healthy controls, three separate models using logistic regression analyses were created: (1) a model including all study participants, (2) participants with a history of COVID-19, and (3) participants who had not yet been infected with SARS-CoV-2 virus before Jan 1, 2022. All models were a priori adjusted for age and sex. If the estimated ORs were similar to the calculated E-values, it would be plausible to assume that systematic differences in the reporting of persistent symptoms could account for differences in post-COVID condition between patients with inflammatory rheumatic disease and healthy controls. In participants who met WHO criteria for post-COVID condition, Kaplan-Meier survival analyses were used to compare time until recovery from post-COVID condition following a recent SARS-CoV-2 omicron infection between patients with inflammatory rheumatic disease and healthy controls during the first 26 weeks after the onset of infection. The symptomology of post-COVID condition is presented in bar charts. Descriptive statistics, bar charts, and line graphs were used to compare characteristics of persistent symptoms between patients with inflammatory rheumatic disease and healthy controls with and without a history of COVID-19. P values less than 0·05 were considered statistically significant. SPSS (version 27.0) and R (version 4.0.3) were used for statistical analyses. GraphPad Prism (version 6.0) was used to create the figures. Role of the funding source The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Results In total, 1974 patients with inflammatory rheumatic disease and 733 healthy controls completed the questionnaire sent on March 10, 2022, and were included in this study. Characteristics of patients with inflammatory rheumatic disease and healthy controls are shown in table 1 . Among patients with inflammatory rheumatic disease, mean age was 59 years (SD 13), 1268 (64%) of 1974 were women and 706 (36%) were men; and among healthy controls, mean age was 59 years (SD 12), 495 (68%) of 733 were women and 238 (32%) were men (table 1). Cardiovascular disease, chronic pulmonary disease, diabetes, and obesity were more frequent in patients with inflammatory rheumatic disease than in healthy controls. Rheumatoid arthritis was the most common underlying disease in patients with inflammatory rheumatic disease (1065 [54%] of 1974), and patients were most frequently being treated with methotrexate (803 [41%] of 1974) and tumour necrosis factor inhibitors (606 [31%] of 1974). History of COVID-19 and SARS-CoV-2 vaccination status (number of doses and vaccine type) were similar for patients with inflammatory rheumatic disease and healthy controls.Table 1 Characteristics of patients with inflammatory rheumatic disease and healthy controls Patients with inflammatory rheumatic disease (n=1974) Healthy controls (n=733) Patient characteristics Age, years 59 (13) 59 (12) Female 1268 (64%) 495 (68%) Male 706 (36%) 238 (32%) BMI, kg/m2 26 (5) 25 (4) Coexisting conditions Cardiovascular disease 256 (13%) 51 (7%) Chronic pulmonary disease 231 (12%) 43 (6%) Diabetes 108 (5%) 24 (3%) Obesity 343 (17%) 75 (10%) Type of rheumatic disease Rheumatoid arthritis 1065 (54%) .. Psoriatic arthritis 306 (16%) .. Ankylosing spondylitis 276 (14%) .. Axial or peripheral spondylarthritis 38 (2%) .. Juvenile idiopathic arthritis 34 (2%) .. Systemic lupus erythematosus 105 (5%) .. Vasculitis 40 (2%) .. Polymyalgia rheumatica 96 (5%) .. Sjögren's disease 106 (5%) .. Systemic sclerosis 45 (2%) .. Mixed connective tissue disease 11 (1%) .. Gout 79 (4%) .. Other rheumatic diseases 143 (7%) .. Immunosuppressants* No immunosuppressive medication 390 (20%) 723 (99%) Conventional synthetic DMARDs 1087 (55%) 4 (1%) Methotrexate 803 (41%) 2 (<1%) Hydroxychloroquine 262 (13%) 3 (<1%) Sulfasalazine 96 (5%) 0 Azathioprine 50 (3%) 0 Biological DMARDs 783 (40%) 0 TNF inhibitor 606 (31%) 0 Anti-CD20 therapy 49 (2%) 0 IL-6 inhibitor 33 (2%) 0 Other immunosuppressants 301 (15%) 5 (1%) Prednisone 278 (14%) 5 (1%) SARS-CoV-2 vaccination Number of vaccine doses None 80 (4%) 30 (4%) One 18 (1%) 11 (2%) Two 239 (12%) 117 (16%) Three 1407 (71%) 524 (71%) More than three 230 (12%) 51 (7%) Vaccine type primary vaccination† AstraZeneca 358/1869 (19%) 157/694 (23%) Pfizer–BioNTech 1238/1869 (66%) 396/694 (57%) Moderna 206/1869 (11%) 104/694 (15%) Janssen 36/1869 (2%) 33/694 (5)% Mix 31/1869 (2%) 4/694 (1%) SARS-CoV-2 infections Before Jan 1, 2022 Wildtype (alpha) variant 262 (13%) 115 (16%) Delta variant 93 (5%) 32 (4%) After Jan 1, 2022 (omicron infection) 468 (24%) 218 (30%) PCR-confirmed 169/468 (36%) 82/218 (38%) Antigen-confirmed 92/468 (20%) 32/218 (15%) PCR and antigen test confirmed 210/468 (45%) 104/218 (48%) Data are mean (SD), n (%), or n/N (%). DMARD=disease-modifying anti-rheumatic drug. TNF=tumour necrosis factor. IL=interleukin. * Participants could be treated with multiple immunosuppressants. † Vaccine type was unknown for 25 patients with inflammatory rheumatic disease and nine healthy controls. In total, 365 (78%) of 466 patients with inflammatory rheumatic disease and 172 (79%) of 218 healthy controls with a SARS-CoV-2 omicron infection completed the first COVID-19 sequelae questionnaire (appendix p 6). Participants who did not complete the questionnaire were slightly younger than those who completed the questionnaire (appendix p 5). 77 (21%) of 365 patients with inflammatory rheumatic disease and 23 (13%) of 172 healthy controls had post-COVID condition according to WHO classification criteria. Unadjusted logistic regression analyses showed that having a rheumatic disease was associated with higher odds of post-COVID condition (OR 1·73 [95% CI 1·04– 2·87], p=0·033; table 2 ). After adjusting for confounding, the association no longer reached statistical significance (adjusted [a]OR 1·53 [95% CI 0·90–2·59], p=0·12). Post-hoc evaluation of covariables in the regression model showed that higher BMI (aOR 1·06 [95% CI 1·01–1·12]; p=0·015) and an increased disease severity of the acute infection phase of SARS-CoV-2 (aOR 3·07 [1·81–5·22]; p<0·0001) were associated with higher odds of post-COVID condition (table 2). Mediation analyses showed that the proportion of the association between participant status and post-COVID condition mediated by disease severity of acute infection phase of SARS-CoV-2 was low (17·1% for the crude analysis, 17·6% for the first adjusted analysis, and 11·7% for the second adjusted analysis), which corresponds with the low odds of indirect effects (OR range, 1·05–1·08; table 3 ).Table 2 Logistic regression and E-value analyses OR (95% CI) p value Prospective data: risk of meeting WHO criteria for post-COVID condition Univariable model Healthy controls 1 (ref) .. Patients with rheumatic diseases 1·73 (1·04–2·87) 0·033 E-value* 1·96 (1·16) .. Multivariable model Healthy controls 1 (ref) .. Patients with rheumatic diseases 1·53 (0·90–2·59) 0·12 Age 1·02 (1·00–1·12) 0·14 Sex Male 1 (ref) .. Female 1·08 (0·65–1·79) 0·78 BMI 1·06 (1·01–1·12) 0·015 Cardiovascular disease 0·86 (0·41–1·84) 0·70 Pulmonary disease 1·13 (0·55–2·33) 0·74 History of COVID-19 before Jan 1, 2022 1·35 (0·72–2·52) 0·35 Vaccination status Unvaccinated 1 (ref) .. Vaccinated 0·88 (0·35–2·21) 0·79 Severity of acute infection phase of SARS-CoV-2† WHO severity score <3 1 (ref) .. WHO severity score ≥3 3·07 (1·81–5·22) <0·0001 E-value* 1·78 (1·00) .. Retrospective data: risk of reporting persistent symptoms between Jan 1, 2020, and Jan 1, 2022 All participants‡ Healthy controls 1 (ref) .. Patients with rheumatic diseases 2·11 (1·69 – 2·64) <0·0001 Participants with a history of COVID-19‡ Healthy controls 1 (ref) .. Patients with rheumatic diseases 1·62 (1·08–2·45) 0·021 Participants without a history of COVID-19‡ Healthy controls 1 (ref) .. Patients with rheumatic diseases 2·52 (1·92–3·32) <0·0001 Results of logistic regression analyses and E-value analyses for the unadjusted and adjusted association between participant status and post-COVID condition. Data are OR (95% CI), and point estimates and the lower limit of the 95% CI for E-values, unless otherwise specified. The prospective analyses included all participants who were prospectively monitored after a SARS-CoV-2 omicron infection and completed the first COVID-19 sequelae questionnaire. The retrospective analyses included all participants who completed the questionnaire sent on March 10, 2022. p<0·05 was considered statistically significant. OR=odds ratio. * Data are E-value for point estimate and the lower limit of the confidence interval from left to right. † A WHO score of 1 indicates asymptomatic infection, a score of 2 indicates mild disease without the need for assistance, a score of 3 indicates mild disease with need for assistance (ie, could not care for themselves in daily life due to the severity of their symptoms) but no hospital admission, a score of 4 or higher indicates admission to hospital, intensive care unit, or death. ‡ Models were adjusted for age and sex. Table 3 Mediation analyses Crude Adjusted* Adjusted† OR (95% CI) p value OR (95% CI) p value OR (95% CI) p value Total effects Healthy controls 1 (ref) .. 1 (ref) .. 1 (ref) .. Patients with inflammatory rheumatic disease 1·58 (1·03–2·42) 0·037 1·50 (0·98–2·30) 0·063 1·46 (0·95–2·23) 0·081 Direct effects Healthy controls 1 (ref) .. 1 (ref) .. 1 (ref) .. Patients with inflammatory rheumatic disease 1·46 (0·95–2·24) 0·082 1·40 (0·91–2·13) 0·12 1·40 (0·91–2·13) 0·12 Indirect effects Healthy controls 1 (ref) .. 1 (ref) .. 1 (ref) .. Patients with inflammatory rheumatic disease 1·08 (1·00–1·17) 0·054 1·07 (0·99–1·16) 0·085 1·05 (0·99–1·10) 0·099 Results of regression-based approach for causal mediation analyses using a log-linear model for the outcome variable. p<0·05 was considered statistically significant. * Covariates are conditioned on age, 55 years; sex, female; comorbidities, absent. † Covariates are conditioned on age, 55 years; sex, male; comorbidities, absent. The baseline characteristics were similar for patients with inflammatory rheumatic disease and healthy controls with and without post-COVID condition (appendix p 2). As expected, patients with inflammatory rheumatic disease and healthy controls with post-COVID condition more frequently had WHO COVID-19 severity scores 3 compared to those without post-COVID condition; 26 (34%) of 77 patients with inflammatory rheumatic disease and six (26%) of 23 healthy controls with post-COVID condition compared with 44 (15%) of 288 patients with inflammatory rheumatic disease and 14 (9%) of 149 healthy controls without post-COVID condition (appendix p 2). Fatigue and loss of fitness were the most frequently reported symptoms in both patients with inflammatory rheumatic disease and healthy controls with post-COVID condition (appendix p 7). Clustering of the four most frequently reported symptoms is shown in figure 1 . Time to recovery from post-COVID condition was similar for patients with inflammatory rheumatic disease and healthy controls (figure 2 ; p=0·17).Figure 1 Venn diagram of post-COVID condition symptoms Diagram depicts the relationship between the four most reported symptoms of post-COVID condition in all participants (A), patients with inflammatory rheumatic disease (B), and healthy controls (C). Figure 2 Kaplan-Meier curve of time until recovery from a post-COVID condition Only participants who met WHO criteria for post-COVID condition were included for analyses. More patients with inflammatory rheumatic disease than healthy controls with post-COVID condition reported contacting a health-care professional because of persistent symptoms after a SARS-CoV-2 infection (40 [52%] of 77 patients vs seven [30%] of 23 healthy controls; appendix p 3). Patients with inflammatory rheumatic disease most frequently contacted a general practitioner (29 [73%] of 40), a rheumatologist (14 [35%] of 40), or physiotherapist (14 [35%] of 40); 20 (26%) of 77 patients with inflammatory rheumatic disease had contact with multiple health-care professionals. By contrast, healthy controls exclusively contacted a general practitioner (appendix p 3). 25 (63%) of 40 patients with inflammatory rheumatic disease and four (57%) of seven healthy controls, had a diagnosis of post-COVID condition established by a health-care professional. Disease characteristics for participants who reported persistent symptoms (duration ≥8 weeks) during the first 2 years of the COVID-19 pandemic, stratified for participant status and history of COVID-19, are presented in the appendix (p 4). Persistent symptoms were reported more frequently by participants with a history of COVID-19 than by those without a history of COVID-19, and by patients with inflammatory rheumatic disease compared with healthy controls; 152 (43%) of 351 patients with inflammatory rheumatic disease versus 48 (33%) of 147 healthy controls with a history of COVID-19, and 346 (21%) of 1623 patients with inflammatory rheumatic disease versus 63 (11%) of 586 healthy controls without a history of COVID-19 (figure 3 ; appendix p 4). Notably, patients with inflammatory rheumatic disease who had a history of COVID-19 less frequently attributed persistent symptoms to post-COVID condition than did healthy controls; 50 (33%) of 152 patients with inflammatory rheumatic disease versus 30 (63%) of 48 healthy controls. The distribution of reported symptom types was similar across groups, but insomnia was more frequent in participants without a history of COVID-19 (appendix p 8).Figure 3 Prevalence of persistent symptoms during the first 2 years of the COVID-19 pandemic Graph showing the proportion of participants with persistent symptoms for increasing total symptom duration, stratified by history of COVID-19 before Jan 1, 2022, and participant status. Exploratory analyses to investigate whether unmeasured confounding could explain the observed association between participant status and post-COVID condition showed that the E-value was 1·96 (lower bound confidence interval E-value 1·16) for the crude model and 1·78 (lower bound confidence interval E-value 1·00) for the model adjusted for all potential measured confounders (table 2). Logistic regression analyses confirmed that patients with inflammatory rheumatic disease were more likely to report persistent symptoms that could also be observed in post-COVID condition during the first 2 years of the COVID-19 pandemic, and estimated ORs were similar to the calculated E-values (OR range, 1·62–2·52; table 2). Discussion In this study, we prospectively monitored patients with inflammatory rheumatic disease and healthy controls after a SARS-CoV-2 infection during the omicron-dominant period. We observed that 21% of patients with inflammatory rheumatic disease and 13% of healthy controls had post-COVID condition according to WHO criteria. Type of symptoms and recovery time from post-COVID condition were similar for patients and controls. Notably, participants with inflammatory rheumatic disease without a history of COVID-19 were more likely to report persistent symptoms that are used to identify post-COVID condition during the first 2 years of the pandemic than were healthy controls. Finally, patients with inflammatory rheumatic disease more frequently sought contact with a health-care professional because of persistent symptoms after a SARS-CoV-2 infection than did healthy controls, primarily general practitioners, physiotherapists, and rheumatologists. To our knowledge, this is the first large prospective study comparing the long-term consequences of COVID-19 between patients with inflammatory rheumatic disease and healthy controls. The observed proportion of patients with post-COVID condition after a SARS-CoV-2 infection with the omicron variant was considerably lower than reported in previous studies on post-COVID condition in patients with inflammatory rheumatic disease. For example, two retrospective cohort studies reported persistence of symptoms after a SARS-CoV-2 infection in 57% and 69% of patients with rheumatic diseases,14, 15 and two other cohort studies showed that 45% and 56% had prolonged symptom duration.16, 17 On the basis of these high proportions, it has been concluded that the risk of long-term health problems and the need for health care after a SARS-CoV-2 infection is high for patients with rheumatic diseases”.14, 15, 16, 17 However, we should be careful when interpreting results from these studies. First, the studies of Leon and colleagues,14 Brito-Zeron and colleagues,15 and Di Iorio and colleagues16 included participants after confirmation of a SARS-CoV-2 infection. This could introduce selection bias, as more severe COVID-19 cases are more likely to be PCR-confirmed than are mild or asymptomatic cases, and patients with more severe disease might be more willing to participate in COVID-19 research.18 The proportion of participants admitted to hospital with COVID-19 in these studies is substantially higher than in our own previous observations; 45–69% versus 21%,14, 15 and the study population of Leon and colleagues exclusively consisted of participants in hospital with COVID-19.14 Multiple studies have shown that a higher disease severity during the acute infection phase of SARS-CoV-2 is associated with an increased risk of prolonged symptom duration,19, 20, 21 so it is likely that this overrepresentation of severe COVID-19 cases contributed to raising risk estimates for developing post-COVID condition. Second, Brito-Ziron and colleagues' and Barbhaiya and colleagues' studies were retrospective, and data were retrieved from medical health records or data registries.15, 17 The retrospective design increases the risk of selection bias and data accuracy is limited. Third, Brito-Zeron and colleagues' study found that 57% of patients with primary Sjögren's syndrome remained symptomatic after a median follow-up of 5 months after SARS-CoV-2 infection.15 However, follow-up time ranged from 5 to 388 days, which means that even patients with a very short follow-up time after SARS-CoV-2 infection could contribute to increasing the final risk estimate. Fourth, all four of the aforementioned studies were done before the emergence of the omicron variant of SARS-CoV-2, which might reduce the generalisability to current SARS-CoV-2 infections due to the considerably lower virulence of the omicron variant than previous variants.14, 15, 16, 17 In the present study, participants were included before infection with SARS-CoV-2 and longitudinally evaluated after PCR-confirmed or antigen- confirmed SARS-CoV-2 infections during the omicron-dominant period. This study design minimises selection bias and optimises data validity and reliability, making it more likely that the results more accurately reflect the true risk of developing post-COVID condition in patients with inflammatory rheumatic disease compared with results of the aforementioned studies, at least for infections with the omicron variant. In the current study, we simultaneously investigated patients with inflammatory rheumatic disease and healthy controls, which allowed us to directly assess whether patients with inflammatory rheumatic disease are more susceptible to post-COVID condition than are people from the general population. The inclusion of a control group is highly relevant for research into post-COVID condition, since the large amount of heterogeneity between studies considerably limits between-study comparisons.22 This between-study heterogeneity is mainly caused by the broad and non-specific definition of post-COVID condition, which includes any persistent subjective symptom after a SARS-CoV-2 infection as long as alternative causes cannot be identified.3 Consequently, more than 50 symptoms have already been identified,23 but most studies have reported fatigue, dyspnea, neurocognitive symptoms (eg, brain fog), and musculoskeletal symptoms (eg, myalgia or arthralgia) in people from the general population.24 This symptomology corresponds with results from post-COVID condition studies in patients with rheumatic diseases,14, 15, 16, 17 and is in line with our own observations that both symptomology and recovery time from post-COVID condition were similar between patients with inflammatory rheumatic disease and healthy controls. However, we also observed that the prevalence of post-COVID condition after SARS-CoV-2 omicron infections was 1·7 times higher in patients with inflammatory rheumatic disease than in controls, which contradicts the results of a retrospective study done in participants with COVID-19 in hospital.25 However, the current clinical picture of post-COVID condition bears considerable overlap with disease manifestations of rheumatic diseases,5, 23 which might lead to overestimated risk estimates due to the increased possibility of falsely attributing clinical manifestations of rheumatic diseases to post-COVID condition. In exploratory analyses that investigated unmeasured confounding, we demonstrated that patients with inflammatory rheumatic disease who had not had COVID-19 were more likely to report persistent symptoms that are used to identify post-COVID condition than were healthy controls, and that this association reached the necessary strength to account for the higher proportion of post-COVID condition in patients with inflammatory rheumatic disease than healthy controls. Additionally, patients with inflammatory rheumatic disease reported more severe symptoms during the acute infection phase of SARS-CoV-2 than did controls, but disease severity mediated less than 20% of the association between participant status and post-COVID condition. Collectively, our data highlight the limitations of applying current criteria of post-COVID condition in patients with inflammatory rheumatic disease. We therefore believe that data from studies on post-COVID condition in patients with inflammatory rheumatic disease should be interpreted with caution, and that it might be appropriate for rheumatologists to have a nuanced attitude when communicating the long-term risks of COVID-19 to their patients. Important strengths of our study include the prospective follow-up of a large cohort of patients with various inflammatory rheumatic diseases and the inclusion of a large group of simultaneously enrolled age-matched and sex-matched healthy controls. Additionally, the response rate of COVID-19 sequalae questionnaires used for prospective monitoring after SARS-CoV-2 omicron infections was almost 80%, which is considered to be high.26 Furthermore, participants were included before infection with SARS-CoV-2, which minimises the risk of overrepresentation of severe COVID-19 cases and thus selection bias. Furthermore, we addressed and quantified systematic differences in the reporting of symptoms used to define post-COVID condition between patients with inflammatory rheumatic disease and people from the general population, which had already been hypothesised to be a potentially relevant source of unmeasured confounding by others.5, 23 Finally, we investigated the health-care utilisation of patients with inflammatory rheumatic disease and healthy controls with persistent symptoms after SARS-CoV-2 infections, as it is important to know which health-care professionals are likely to encounter these patients. We observed that patients with inflammatory rheumatic disease more frequently contacted health-care professionals for persistent symptoms than did healthy controls, primarily a general practitioner, physiotherapist, or rheumatologist. However, the threshold for visiting health-care professionals because of persistent symptoms might be lower for patients with inflammatory rheumatic disease than healthy controls, because they require multiple and frequent interactions with physicians because of their underlying condition. This could, in part, explain the difference between patients and controls in the number of times they contacted health-care professionals for persistent symptoms. Our data also imply that health care for patients with inflammatory rheumatic disease in the Netherlands who experience persistent symptoms after SARS-CoV-2 infections does not remain exclusively within the domain of rheumatologists, so knowledge distribution regarding epidemiology and disease presentation and prognosis of post-COVID condition in patients with inflammatory rheumatic disease beyond the borders of rheumatology has important value. Our study also has several limitations. First, data on the occurrence of persistent symptoms during the first 2 years of the COVID-19 pandemic were collected retrospectively, which negatively affects data accuracy. It might also introduce recall bias, as participants with a history of COVID-19 might recall persistent symptoms more accurately than those without a history of COVID-19. Although this does not result in systematic differences between patients and controls, this part of our data should be interpreted with caution. Second, follow-up questionnaires after SARS-CoV-2 infections were sent to participants at fixed timepoints independent of the timing of a SARS-CoV-2 infection, which means that the time between SARS-CoV-2 infection and survey completion differs between individuals. However, the mean time between infection onset and survey completion was similar for patients with inflammatory rheumatic disease and healthy controls, so we do not expect that this negatively affects the validity of our comparisons between the two groups. Third, despite the large number of participants included in our cohort, the number of participants, especially healthy controls, who developed post-COVID condition after an omicron infection were few. This, combined with the diagnostic uncertainty of post-COVID condition that entails a considerable risk of misclassification, limits our ability to draw definitive conclusions regarding the risk of post-COVID condition for patients with inflammatory rheumatic disease compared with people from the general population. Fourth, post-COVID condition cases were identified via a questionnaire and not via individual assessment by health-care physicians. Other potential causes of persistent symptoms could therefore not be ruled out in participants who did not visit a physician, so our findings might be somewhat overestimated. This overestimation is probably more pronounced in patients with inflammatory rheumatic disease than in healthy controls due to similarities in the disease presentation of post-COVID condition and the underlying rheumatic disease. Finally, we collected data up to 26 weeks of follow-up after the onset of infection, but a considerable proportion of patients with inflammatory rheumatic disease and healthy controls with post-COVID condition still had unresolved symptoms at that time. Longer follow-up will therefore be necessary to draw definitive conclusions about recovery time from post-COVID condition. In summary, we found that 21% of patients with inflammatory rheumatic disease and 13% of healthy controls developed post-COVID condition after a SARS-CoV-2 omicron infection based on WHO-criteria. Furthermore, symptomology and recovery time from post-COVID condition were similar between patients and controls. We also found that more patients with inflammatory rheumatic disease than healthy controls without a history of COVID-19 reported symptoms that are also observed in post-COVID condition, and this association reached the strength that is necessary to negate the difference in post-COVID condition between patients and controls. Therefore, it is possible that the observed difference in post-COVID condition between patients and controls could be partly explained by clinical manifestations in the context of underlying rheumatic diseases. Our study highlights the limitations of applying current criteria for post-COVID condition in patients with inflammatory rheumatic disease, and suggests it might be appropriate for physicians to have a nuanced attitude when communicating the long-term consequences of COVID-19 to their patients. Data sharing Aggregated data and code for reproducing the results of this analysis can be shared upon reasonable request to the corresponding author. Declaration of interests We declare no competing interests. Supplementary Material Supplementary appendix Contributors LB wrote the first draft of the manuscript and all other authors revised the manuscript for important intellectual content. LB and FH directly accessed and verified the underlying data and performed the statistical analyses. TR and MS did the serological assays and all other authors contributed to data acquisition. All authors met the criteria for authorship set by the International Committee of Medical Journal Editors. All authors confirm that they had full access to all data in the study, and accept responsibility for the decision to submit for publication. ==== Refs References 1 Fernández-de-Las-Peñas C Palacios-Ceña D Gómez-Mayordomo V Prevalence of post-COVID-19 symptoms in hospitalized and non-hospitalized COVID-19 survivors: a systematic review and meta-analysis Eur J Intern Med 92 2021 55 70 34167876 2 Han Q Zheng B Daines L Sheikh A Long-term sequelae of COVID-19: a systematic review and meta-analysis of one-year follow-up studies on post-COVID symptoms Pathogens 11 2022 269 35215212 3 WHO A clinical case definition of post COVID-19 condition by a Delphi consensus, 6 October 2021 https://www.who.int/publications/i/item/WHO-2019-nCoV-Post_COVID-19_condition-Clinical_case_definition-2021.1 2021 4 Anaya JM Herrán M Beltrán S Rojas M Is post-COVID syndrome an autoimmune disease? Expert Rev Clin Immunol 18 2022 653 666 35658801 5 Sapkota HR Nune A Long COVID from rheumatology perspective—a narrative review Clin Rheumatol 41 2022 337 348 34845562 6 Boekel L Hooijberg F Vogelzang EH Spinning straw into gold: description of a disruptive rheumatology research platform inspired by the COVID-19 pandemic Arthritis Res Ther 23 2021 207 34348783 7 Boekel L Hooijberg F Vogelzang EH Antibody development and disease severity of COVID-19 in non-immunised patients with rheumatic immune-mediated inflammatory diseases: data from a prospective cohort study RMD Open 8 2022 e002035 8 Boekel L Stalman EW Wieske L Breakthrough SARS-CoV-2 infections with the delta (B.1.617·2) variant in vaccinated patients with immune-mediated inflammatory diseases using immunosuppressants: a substudy of two prospective cohort studies Lancet Rheumatol 4 2022 e417 e429 35527808 9 Boekel L Besten YR Hooijberg F SARS-CoV-2 breakthrough infections in patients with immune-mediated inflammatory diseases during the omicron dominant period Lancet Rheumatol 4 2022 e747 e750 36034738 10 Vogelzang EH Loeff FC Derksen NIL Development of a SARS-CoV-2 total antibody assay and the dynamics of antibody response over time in hospitalized and nonhospitalized patients with COVID-19 J Immunol 205 2020 3491 3499 33127820 11 Steenhuis M van Mierlo G Derksen NI Dynamics of antibodies to SARS-CoV-2 in convalescent plasma donors Clin Transl Immunology 10 2021 e1285 12 Selvakumar J Havdal LB Drevvatne M Prevalence and characteristics associated with post-covid-19 condition among nonhospitalized adolescents and young adults JAMA Netw Open 6 2023 e235763 13 Haneuse S VanderWeele TJ Arterburn D Using the E-Value to assess the potential effect of unmeasured confounding in observational studies JAMA 321 2019 602 603 30676631 14 Leon L Perez-Sancristobal I Madrid A Persistent post-discharge symptoms after COVID-19 in rheumatic and musculoskeletal diseases Rheumatol Adv Pract 6 2022 rkac008 15 Brito-Zerón P Acar-Denizli N Romão VC Post-COVID-19 syndrome in patients with primary Sjögren's syndrome after acute SARS-CoV-2 infection Clin Exp Rheumatol 39 suppl 133 2021 57 65 16 Di Iorio M Cook CE Vanni KMM DMARD disruption, rheumatic disease flare, and prolonged COVID-19 symptom duration after acute COVID-19 among patients with rheumatic disease: a prospective study Semin Arthritis Rheum 55 2022 152025 17 Barbhaiya M Jannat-Khah D Levine J Risk Factors for “long haul” COVID-19 in rheumatology outpatients in New York City https://acrabstracts.org/abstract/risk-factors-for-long-haul-covid-19-in-rheumatology-outpatients-in-new-york-city/ 2021 18 Accorsi EK Qiu X Rumpler E How to detect and reduce potential sources of biases in studies of SARS-CoV-2 and COVID-19 Eur J Epidemiol 36 2021 179 196 33634345 19 Daugherty SE Guo Y Heath K Risk of clinical sequelae after the acute phase of SARS-CoV-2 infection: retrospective cohort study BMJ 373 2021 n1098 20 Huang C Huang L Wang Y 6-month consequences of COVID-19 in patients discharged from hospital: a cohort study Lancet 397 2021 220 232 33428867 21 Hirschtick JL Titus AR Slocum E Population-based estimates of post-acute sequelae of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection (PASC) prevalence and characteristics Clin Infect Dis 73 2021 2055 2064 34007978 22 Fernández-de-Las-Peñas C Notarte KI Peligro PJ Post-COVID condition symptoms in individuals infected with different SARS-CoV-2 variants of concern: a systematic review of the literature Viruses 14 2022 2629 23 Calabrese LH Calabrese CM Long COVID: defining the role of rheumatology in care and research Lancet Rheumatol 4 2022 e812 e814 36211989 24 O'Mahoney LL Routen A Gillies C The prevalence and long-term health effects of Long Covid among hospitalised and non-hospitalised populations: a systematic review and meta-analysis EClinicalMed 55 2022 101762 25 Fernández-de-Las-Peñas C Martín-Guerrero JD Navarro-Pardo E The presence of rheumatological conditions is not a risk factor of long-term post-COVID symptoms after SARS-CoV-2 infection: a multicenter study Clin Rheumatol 41 2022 585 586 34561811 26 Holtom B Baruch Y Aguinis H Ballinger GA Survey response rates: trends and a validity assessment framework Hum Relat 75 2022 1560 1584
PMC010xxxxxx/PMC10292828.txt	==== Front Travel Behav Soc Travel Behav Soc Travel Behaviour & Society 2214-367X 2214-3688 Hong Kong Society for Transportation Studies. Published by Elsevier Ltd. S2214-367X(23)00073-X 10.1016/j.tbs.2023.100622 100622 Article Exploring the influences of personal attitudes on the intention of continuing online grocery shopping after the COVID-19 pandemic Asgari Hamidreza a Azimi Ghazaleh a Titiloye Ibukun a Jin Xia b⁎ a Department of Civil and Environmental Engineering, Florida International University, 10555 W. Flagler Street, EC3725, Miami, FL 33174, USA b Department of Civil and Environmental Engineering, Florida International University, 10555 W. Flagler Street, EC 3603, Miami, FL 33174, USA ⁎ Corresponding author. 26 6 2023 10 2023 26 6 2023 33 100622100622 4 3 2022 2 6 2023 19 6 2023 © 2023 Hong Kong Society for Transportation Studies. Published by Elsevier Ltd. All rights reserved. 2023 Hong Kong Society for Transportation Studies Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The unprecedented COVID-19 pandemic has brought drastic changes in our daily activities. One of these essential activities is grocery shopping. In compliance with the recommended social distancing standards, many people have switched to online grocery shopping or curbside pickup to minimize possible contagion. Although the shift to online grocery shopping is substantial, it is not clear whether this change would last in the long term. This study examines the attributes and underlying attitudes that may influence individuals’ future decisions on online grocery shopping. An online survey was conducted in May 2020 in South Florida to collect data for this study. The survey contained a comprehensive set of questions related to respondents’ sociodemographic attributes, shopping and trip patterns, technology use, as well as attitudes toward telecommuting and online shopping. A structural equation model (SEM) was applied to examine the intervening effects of observed as well as latent attitude variables on the likelihood of online grocery shopping after the outbreak. The results indicated that those with more experience in using online grocery shopping platforms were more likely to continue purchasing their groceries online. Individuals with positive attitudes toward technology and online grocery shopping in terms of convenience, efficiency, usefulness, and easiness were more likely to adopt online grocery shopping in the future. On the other hand, pro- driving individuals were less likely to substitute online grocery shopping for in-store shopping. The results suggested that attitudinal factors could have substantial impacts on the propensity toward online grocery shopping. Keywords COVID-19 Pandemic Online Grocery Shopping Online Survey Structural Equation Model ==== Body pmc1 Introduction Starting in early 2020, the COVID-19 pandemic has brought drastic and unprecedented changes in individuals’ travel behavior, routines, and daily activities. Amid this pandemic, there is a significant decrease in in-store sales and a concurrent increase in online shopping due to social distance measures and other mandated preventive regulations (Matson et al., 2021, Richards and Rickard, 2020, Zhao et al., 2020, Hassen et al., 2020, Charlebois, 2020, Donthu and Gustafsson, 2020, Shamshiripour et al., 2020, Chang and Meyerhoefer, 2021). The shift from in-store to online shopping is especially substantial for grocery shopping at the beginning of the pandemic (Shamshiripour et al., 2020). Statistics imply an increasing trend in the online sales during the pandemic. As of August 2019, U.S. online grocery sales totaled $1.2 billion, with an estimated 16.1 million customers who purchased groceries online. By June 2020, the sales total increased to $7.2 billion, and the estimated customers increased to 45.6 million (Forbes, 2020). However, it is unclear whether the shift to online grocery shopping will persist after the pandemic. In particular, both online and in-store shopping alternatives are accompanied by a number of pros and cons, which make it more difficult to predict the future of consumer’s shopping behavior in the post pandemic conditions. In-store shopping offers advantages such as chances for socializing with others, better assessment of the quality of the goods, and no delivery fees. (Raijas, 2002, Hansen, 2006). On the other hand, many people find online grocery shopping a convenient, time-saving, and attractive replacement for in-store shopping (Lee et al., 2017, Raijas, 2002, Huang and Oppewal, 2006, Chakraborty, 2019, Shukla and Sharma, 2018, Forsythe and Shi, 2003, Hansen, 2006). From a planning perspective, holding to a reliable prediction of online shopping behavior and its trend in near future is of the essence for researchers and stakeholders because it has inevitable impacts in a variety of travel-related and urban domains (Pettersson et al., 2018, Le et al., 2022). The literature documents that online shopping can be viewed as a subsequent of information and communication technology (ICT), with similar concepts and complexities involved when addressing the impacts on human travel behavior (Salomon, 1986, Mokhtarian, 1990, Mokhtarian and Meenakshisundaram, 1999, Mokhtarian et al., 2006). In particular, the decision to shop online might affect different aspects of the daily activity-travel plan at individual level, including travel distance (Ferrell, 2004, Ferrell, 2005, Shi et al., 2020a, Shi et al., 2020b, Hiselius et al., 2015), trip modes (Suel and Polak, 2017, Bjerkan et al., 2020, Etminani-Ghasrodashti and Hamidi, 2020, Hjorthol, 2009), activity duration and time use (Ding and Lu, 2017, Lee et al., 2017, Weltevreden, 2007, Lachapelle and Jean-Germain, 2019, Farag et al., 2007), and trip chaining behavior (Ferrell, 2005, Farag et al., 2007, Ding and Lu, 2017). Furthermore, and in long run, the concept of e commerce and online shopping might affect urban sprawl and residential relocation (Yang et al., 2021, Nahiduzzaman et al., 2019, Nahiduzzaman et al., 2021, Nanda et al., 2021, Beckers and Verhetsel, 2021). At aggregate level, the adoption of new logistic and supply chain models in the context of smart cities might result in a higher share of commercial vehicles and heavy trucks on the roads, which in turn can lead to adverse impacts on infrastructure quality (e.g., accelerated wear and tear), traffic congestion, air quality and traffic safety (Singh and Santhakumar, 2021, Pan et al., 2019, Badri, 2020, Schade et al., 2018, Seifert and Markoff, 2017). Given the above motivation, this paper intends to investigate the extent to which people may continue to shop online for groceries after the pandemic. In addition to the pandemic-induced concerns for infection and conforming to the social distancing standards, it is reasonable to expect that individuals’ personal and household attributes as well as their preferences, attitudes, and perceptions play a significant role in how they choose to shop for groceries. Further, the practical experience gained during the pandemic might have reshaped their attitudes and their choices for the post pandemic condition. By exploring these variables that potentially contribute to shoppers’ behavior toward grocery shopping in a multivariate predictive model, one can expect to gain a more transparent picture of the future online shopping market. To accomplish this research objective, a consumer survey in view of shopping activities was conducted in May 2020, during which the pandemic exerted its full effect on the public in terms of voluntary and regulatory travel restrictions. The survey focused on adults in south Florida and gathered information on individuals’ socioeconomic and demographic attributes, online grocery shopping experience (i.e., frequency of online grocery shopping before and during the COVID-19 pandemic), as well as attitudinal statements towards various aspects of online and in-store grocery shopping such as technology use and perceived benefits and concerns for shopping grocery online. A structural equation model (SEM) was then applied to examine the intervening effects of observed as well as latent attitude variables on the likelihood of online grocery shopping after the outbreak. The rest of this paper lays out as follows: Section 2 presents a summary of the current state of the literature in terms of online shopping. In particular, we merely focus on predictive analytics and contributing factors at a personal level. Section 3 elaborates data collection and survey design. Descriptive statistics come next, where we explore consumers’ responses to a handful of online-shopping related questions. Section 4 lays out the theory of structural equation models, where we discuss the suitability of the SEM model to the purpose of this study. Model results and discussion are presented in the next section, followed by potential policy implications of our findings. Finally, we conclude the paper by presenting the major highlights, discussing limitations, as well as providing recommendations on further research avenues. 2 Literature review The review of literature for this study consists of two parts. Findings from the literature on online grocery shopping behavior before the pandemic are first discussed, followed by discussions on studies investigating the impacts of the COVID-19 outbreak on online grocery shopping. Explicitly, we focus on research efforts that used predictive analytical approaches to estimate online shopping behavior at individual level as well as identifying significant contributing factors. 2.1 Online grocery shopping Several studies investigated factors that contribute to the adoption of online grocery shopping (OGS) or electronic grocery stores (EGS). Raijas (2002) explored the characteristics of people who used EGS in Finland by analyzing data from a pilot survey on grocery shopping behavior. They showed that the most significant factors encouraging online grocery shopping included the reluctance to do product selection, time-saving, and easiness of online grocery shopping. On the other hand, challenges in finding the right products, higher prices, and being unsure about the quality were the major disincentives for online grocery shopping. These issues were especially significant for inexperienced online grocery shoppers. Driediger and Bhatiasevi (2019) examined factors that affect online grocery shopping behavior in Thailand using an in-person survey. The results showed that perceived easiness for online grocery, perceived usefulness (i.e., improve their quality of life), and perceived enjoyment during the shopping experience have positive impacts on the adoption of online grocery shopping. Bauerová and Klepek (2018) found similar results from an online survey conducted in the Czech Republic. They showed that the perceived easiness of online grocery shopping has a positive impact on the perceived usefulness, which subsequently influenced the intention of using OGS. Other studies presented similar results (Chakraborty, 2019, Shukla and Sharma, 2018). Hansen (2006) analyzed factors influencing grocery shopping behavior by those who have online grocery shopping experiences based on data from an online survey conducted in Denmark and Sweden. The results showed that avoidance of physical effort associated with in-store shopping encouraged the repurchase of products online. In contrast, the joy of in-store shopping and challenges with online grocery shopping (e.g., electronic grocery shopping is difficult, receiving items delivered at home is not convenient) reduced individuals’ tendency toward frequent online grocery shopping. In another study by Mortimer et al. (2016), the intention to shop groceries online was examined using data from an online survey. They analyzed the effects of perceived risks (e.g., feeling secure purchasing online with credit cards and providing personal information) on their intention to repurchase online. The findings revealed that satisfactory experiences with OGS websites positively influenced these individuals’ trust and intentions to repurchase from the websites. However, perceived risks associated with online grocery shopping mediated the effect of trust on the intention to repurchase for infrequent shoppers. The results indicated that risk perceptions on online grocery shopping platforms intensify irregular shoppers’ concerns for these platforms and discourage them from becoming frequent customers. Similar findings were presented by other studies (Min et al., 2012, Anschuetz, 1997). 2.2 The impact of the COVID-19 pandemic on online grocery shopping behavior There has been a general shift toward online grocery shopping since the beginning of the Covid-19 outbreak in early 2020. And as such, many studies have sought to investigate the changes in consumers’ online grocery shopping behavior due to the Covid-19 pandemic, and the factors affecting the adoption, usage, satisfaction, preference, intention, and stickiness of consumers’ online grocery shopping behavioral changes. It should be noted that previous studies were conducted at different stages of the pandemic and in different countries, with varying degrees of lockdown protocols depending on the seriousness of the health crisis or the vaccination rates of residents in their states or countries. In the initial stages of the pandemic and before the availability of vaccines, consumers chose online grocery shopping primarily to preserve their health or that of their loved ones (Baarsma and Groenewegen, 2021, Chang and Meyerhoefer, 2021, Eriksson and Stenius, 2022, Grashuis et al., 2020, Lo et al., 2021, Mercatus, 2020, Shamshiripour et al., 2020). In Taiwan, for instance, an additional COVID-19 case in early 2020 increased Ubox (an online food shopping platform) sales by 5.7% and weekly customers by 16% (Chang & Meyerhoefer, 2021). In the Netherlands, an additional local hospital admission increased online grocery shopping app traffic by 7.3% in the first eight months of the pandemic (Baarsma & Groenewegen, 2021). In Qatar, Hassen et al. (2020) showed a surge in online grocery shopping frequency amid the pandemic compared to pre-pandemic conditions. The results indicated that young people and those with a university degree have a positive inclination toward online grocery shopping. It was noted that some customers were reluctant to do online grocery shopping because they could not check the freshness of the products. In the US, a choice experiment design showed that increasing Covid-19 cases negatively affected preferences to shop inside the grocery store (Grashuis et al., 2020). Ellison et al. (2021) conducted a panel survey with 1,370 households in the U. S. periodically four times amid the pandemic from March to April of 2020. The results showed about a 50% increase in online grocery shopping adoption from the first wave to the fourth wave of data collection. Similar findings were reported by other studies (Shamim et al., 2021, Li et al., 2020). However, in the middle and later months of 2021 when Covid-19 safety concerns had significantly diminished due to widespread vaccination and lockdown protocols had been lifted in most US states, other reasons for online grocery shopping seemed to have taken priority over health concerns (Mercatus., 2021, Warganegara and Hendijani, 2022). For instance, Warganegara and Hendijani (2022) showed that, in the third quarter of 2021, health risks did not show significant effects on the purchasing intent of 300 Indonesian residents. Mercatus (2021) reported that convenience and time savings were now (in 2021) the top two reasons for the change in consumers’ online shopping preference in the US. The impact of attitudinal factors online grocery shopping behavior has been examined across the Covid-19 timeline. Perceived usefulness and favorable attitude toward online shopping have been shown to be strong drivers of online grocery shopping usage and intention (Bezirgani and Lachapelle, 2021, Li et al., 2020, Qi et al., 2021, Tyrväinen and Karjaluoto, 2022, Warganegara and Hendijani, 2022). Perception of low complexity or ease of online grocery shopping usage encouraged online grocery shopping usage, intention, and satisfaction (Alaimo et al., 2020, Tyrväinen and Karjaluoto, 2022, Warganegara and Hendijani, 2022). Perceived risk or insecurity in using technology or online shopping websites tended to negatively affect online grocery shopping usage and intention (Titiloye et al., 2023, Tyrväinen and Karjaluoto, 2022). Technology savviness, shopping enjoyment, and having a pro-environment attitude were drivers of choosing home delivery and curbside pickup options over the in-store grocery shopping alternative (Titiloye et al., 2023). Social influence or pressure has been found to affect online grocery shopping usage and intention (Tyrväinen and Karjaluoto, 2022, Warganegara and Hendijani, 2022), while subjective norm tended to affect online grocery shopping intention (Bezirgani and Lachapelle, 2021, Tyrväinen and Karjaluoto, 2022). However, Qi et al. (2021) examined shopping intentions among 491 Chinese residents in mid-2021 and found no significant effect between subjective norm and online grocery purchase intentions. Regarding socio-demographic factors, it has been shown that younger individuals have a higher tendency to engage in online grocery shopping than older ones (Eriksson and Stenius, 2022, Hassen et al., 2020, Lo et al., 2021, Titiloye et al., 2023). Since women in general tended to be observant of health safety practices than men (Shamim et al., 2021, Truong and Truong, 2022), women (including women aged 45 and above) tended to be adopters than other groups (Eriksson and Stenius, 2022, Titiloye et al., 2023). Education and income are positively associated with online grocery shopping usage (Eriksson and Stenius, 2022, Hassen et al., 2020, Lo et al., 2021), and those with higher educational levels tended to be more satisfied with their online shopping experience than others (Alaimo et al., 2020). Also, full-time workers and those living in larger household sizes or households with children had a higher tendency to engage in online grocery shopping (Eriksson and Stenius, 2022, Lo et al., 2021, Titiloye et al., 2023). However, results on shopping expenditure seem different for socio-demographic variables, as Truong & Truong (2022) found online grocery shopping expenditure to be positively related with age group, negatively related with education, and not significantly related with income and number of children in the household. The results related to expenditure may be related to the level of utility different groups attached to avoiding in-store shopping depending on the level of their health concerns. For example, the older population were more cautious about their health, and may have been willing to spend more money on online grocery purchases than their younger counterparts. One study has sought to estimate the “stickiness” (that is, the extent to which online grocery shopping usage will stabilize or increase after the pandemic) of online grocery purchase behaviors by asking respondents if they were ordering groceries online more often compared to before the start of the pandemic and expecting to retain or increase the proportion of their grocery shopping done online looking one year to the future (Abou-Zeid et al., 2021). The result of the binary logit analysis indicates that shoppers younger than 65 years, who traveled to the store using multiple transportation modes, and located in places with high online grocery service availability were more likely to hold or increase their already elevated online grocery shopping usage. 3 Survey design and data Data for this study were obtained from an online SP-RP survey targeting adults 18 years and older in South Florida in the United States. This survey was conducted in May 2020 amid the COVID-19 pandemic when the stay-at-home order was in full effect in the study area. The survey was implemented through the Qualtrics platform, and the sampling stratifications were based on the 2018 American Community Survey (ACS) in terms of age, gender, income, ethnicity, and race. A total number of 1,028 complete responses were collected. The survey gathered information on the respondents’ socioeconomic and demographic attributes, work arrangements, daily travel patterns, and preferences for shopping and telecommuting. While telecommuting attitudes might look irrelevant at first sight, the literature suggests that ICT applications (in any form including telecommute, online shopping, autonomous vehicles, ride-share, etc.) tend to impact similar contexts of activity travel behavior (e Silva et al., 2017, Le et al., 2022) and therefore, it is reasonable to assume that any revealed information or attitudes towards on specific ICT domain can contribute to another either as a measure of familiarity or an index of technology embracement. Table 1 illustrates the descriptive statistics of the survey data.Table 1 Descriptive statistics. Attributes Share 2018 ACS Age 18–34 31.0% 27.0% 35–54 37.0% 35.0% 55+ 32.0% 38.0% Household Income Less than $25 K 25.9% 46.4% $25 K - $50 K 24.1% $50 K - $75 K 15.7% 28.6% $75 K - $100 K 12.9% $100 K - $125 K 7.8% 25.0% $125 K - $150 K 6.0% $150 K-200 K 3.8% $200 K or more 3.8% Gender Female 55.8% 52.0% Male 44.2% 48.0% Education Level Less than High School 3.1% – High School Graduate 17.4% – Some College, No Degree 21.1% – Associate Degree 12.3% – Bachelor’s Degree 27.4% – Master’s degree 13.4% – Doctorate Degree 2.7% – Professional Degree 2.5% – Race White 72.4% 72.4% Black or African American 16.6% 19.8% Asian 2.3% 2.6% American Indian American Indian or Alaska Native 0.9% 5.2% Native Hawaiian or Pacific Islander 0.6% Two or more Races 7.2% Employment Status Full-Time 45.9% – Part-Time 16.5% – Self-Employed 6.2% – Unemployed 31.3% – Household Composition Living with significant other 44.6% – Living with children 31.8% – Living with grandparents 2.0% – Livings with parents 19.8% – Living with roommates 6.5% – Other 6.0% – Number of Non-working, Non-student Adults 0 adults 62.2% – 1 to 3 adults 36.6% – More than 3 adults 1.3% – Household Size 1 Person 14.8% – 2 Persons 27.2% – 3 Persons 15.5% – 4 Persons 13.2% – 5 Persons 5.6% – 6 Persons 2.3% – 7 Persons 0.5% – 8 Persons 0.2% – 9 Persons 0.1% – 10 and More Persons 0.3% – Missing 20.2% – Household Structure Living with children only 9.6% Living with children and parents 2.7% Living with roommates only 4.5% Living with significant other only 25.4% Living with significant other and children 15.8% Living with significant other and children and parents 1.3% Livings with parents only 12.5% Other 10.3% missing 17.9% Telecommuting Frequency Do not have the telecommuting option 50.5% – Never 15.2% – A few times per year 1.6% – A few times per month 3.8% – Once a week 3.2% – Twice a week 2.4% – 3 days a week 2.5% – 4 days a week 2.5% – 5 days a week 11.7% – 6 days a week 1.9% – 7 days a week 4.7% – To understand how the pandemic changed survey respondents’ shopping behavior, respondents were asked about their online grocery shopping frequency (for pickup and delivery) before and during the COVID-19 pandemic. The choices were defined as:(1) Never (2) Once every month or less (3) Once every two weeks (4) Once every week (5) A few times a week (6) Every day Fig. 1 shows the comparison between the frequencies of online grocery shopping for delivery before and during COVID-19. As shown, about 65% of the respondents had never experienced online grocery shopping for delivery before the pandemic. This percentage decreased by 12% percentage point during the pandemic to 53%. Before the pandemic, only 15% of the respondents shopped online for groceries at least once a week. This percentage also increased during the pandemic and reached 23%. Fig. 1 illustrates that COVID-19 changed individuals’ shopping behavior due to voluntary or enforced social distancing practices. Similar patterns were reported by other studies, suggesting a substantial increase in the percentage of individuals who adopted online grocery shopping for the first time during the pandemic (Ellison et al., 2021, Hassen et al., 2020, Shamim et al., 2021, Shamshiripour et al., 2020).Fig. 1 Online grocery shopping frequency before and during the COVID-19 (N = 1,028). Based on their online grocery shopping experience, survey respondents were also asked how likely they will use online grocery shopping after COVID-19 is no longer a threat, compared to their experience before the pandemic. The choices were outlined as follows:(1) Much less than before (2) Somewhat less than before (3) About the same (4) Somewhat more than before (5) Substantially more than before Fig. 2 and Fig. 3 summarize the reported intention for increasing or reducing shopping grocery online for delivery after the pandemic (expressed in the above five ordinal scales), plotted against actual online grocery shopping frequencies (in relative scales by time periods) before the pandemic (Fig. 2) and during the pandemic (Fig. 3). Fig. 2 shows that pre-pandemic infrequent online grocery shoppers (once every month or less) tended to remain uninterested in online grocery shopping after the pandemia. On the other hand, frequent or habitual online grocery shoppers (once every week or more) are likely to buy groceries online even more frequently in the future. Similarly, as shown in Fig. 3, infrequent shoppers during the pandemic showed lower interest in online grocery shopping in the future than frequent online grocery shoppers. While viewing Fig. 2, Fig. 3, the correlation between the two should be noted. For example, people who had shopped for groceries online frequently before the pandemic probably did online grocery shopping with the same or higher frequencies during the pandemic.Fig. 2 Reported intention of online grocery shopping after COVID-19 by pre-pandemic online grocery shopping frequency (N = 1,028). Fig. 3 Reported intention of online grocery shopping after COVID-19 by during-pandemic online grocery shopping frequency (N = 1,028). The findings shown in Fig. 2, Fig. 3 are consistent with Shamshiripour et al. (2020), who also investigated the propensity of online grocery shopping after the pandemic in the short- (a few months after the pandemic is over) and long-term. Their results indicated that there could be a more positive inclination in public toward online grocery shopping in the long term compared to before the pandemic. However, they did not consider the role of individuals’ shopping behavior before and during the pandemic in future online grocery shopping preferences. The survey also contains questions to solicit respondents’ attitudes toward online grocery shopping. Results of respondents’ responses to these questions are summarized in Fig. 4 . A high percentage of respondents believed that interaction with online grocery shopping websites is easy; online grocery shopping is convenient and requires minimal physical effort. Interestingly, respondents are less likely to indicate that online grocery shopping requires minimal mental effort. This might indicate that choosing between many products can be exhausting and complicated for some people (Raijas, 2002, Hansen, 2006). Besides, most online grocery shopping websites require registration to complete the purchase, which might be inconvenient for some shoppers. Some people who are less experienced in online grocery shopping may also consider the checkout process lengthy and complicated.Fig. 4 Attitudes toward online grocery shopping (N = 545). Fig. 5 presents the pattern observed in the respondents’ attitudes toward technology adoption. Compared to other attitudinal indicators, a high percentage of respondents liked trying new and different things, and they believed that they knew more than others about the latest technologies on the market. On the other hand, when it comes to purchasing technological products, the majority of the respondents disagreed that they would purchase new products regardless of their prices.Fig. 5 Attitudes toward technology adoption (N = 888). Fig. 6 shows the observed pattern in driving-related attitudes. As expected, most respondents did not enjoy driving in heavy traffic. In contrast, most respondents enjoyed the freedom of private vehicle ownership and considered driving a relaxing way to commute. This group of individuals could be more reluctant than others to do online grocery shopping for delivery, as they may enjoy driving to the stores.Fig. 6 Attitudes toward driving (N = 710). Attitudes toward telecommuting are also explored in this study. The underlying reason for this is two-fold: 1) First, the literature suggests that different manifestations of information and communication technology (ICT) may have similar impacts on activity travel behavior at individual level (Le et al., 2022). This might stem from the fact that different forms of ICT adoption and impacts are rooted in the fundamental technology welcoming attitudes and the perception of benefits at individual level, regardless of the specific form. 2) It might be reasonable to assume that attitudes towards one form of ICT can be used either as a surrogate measure of familiarity or a practical manifestation of tech-welcoming attitude. Authors believe that this provides us with compelling evidence to analyze telecommuting attitudes as they may unveil further information about individual’s overall attitudes toward technology. Fig. 7 shows that the majority of the respondents perceived video calling as a suitable alternative to in-person meetings. On the other hand, fewer respondents would enjoy working from home, which could be because of distractions and lack of the necessary supplies and technology products (e.g., high-speed internet connection, office desks, scanners, printers).Fig. 7 Attitudes toward telecommuting (N = 888). 4 Modeling approach Structural equation modeling (SEM) was employed to analyze the survey data for this study. SEM is a widely used statistical tool in certain cases where several endogenous (i.e., dependent) variables are to be predicted simultaneously, or when the analyst intends to measure latent variables that simultaneously serve as predictors in the model (Bollen, 1989, Bollen and Long, 1993, Golob, 2003, Schumacker and Lomax, 2004, Kaplan, 2008). In the former case, SEM helps account for different dependency structures between endogenous variables including correlation, direct one-way causal effect of one endogenous variable on another, or a direct feedback (mutual two-way) impact. In the latter case, SEM prevents random measurement errors in estimating latent variables, which further results in more consistent and unbiased estimates (Christ et al., 2014). Several successful applications of SEM in different aspects of travel behavior and more specifically online shopping activities have been documented in the literature (Njite and Parsa, 2005, Farag et al., 2007, Cao et al., 2012, Irawan and Wirza, 2015, Rita et al., 2019, Dastane, 2020, Prasetyo and Fuente, 2020, Koch et al., 2020, Tarhini et al., 2021, Sarker et al., 2023). Theoretically speaking, SEM is a multivariate statistical method that uses covariance matrices to simultaneously test the existing causal relationships among a set of variables. In technical terms, variables could be endogenous (i.e., variables to be predicted, but may also be used as predictors for other endogenous variables in the model), exogenous (i.e., observed variables that are only used as predictors), and latent variables (unobserved variables that are regressed on other observed variables in the model and maybe predicted or be used as a predictor). Consequently, SEM consists of two concurrent components, namely the measurement model and the structural model. Measurement models regress latent variables (LVs, also referred to as latent factors) on a set of attitudinal indicators. Structural models estimate relationships between dependent (endogenous) variables and covariates, including observed exogenous variables (SED attributes) and LVs. SEMs are usually accompanied by a schematic view, referred to as the path diagram. The path diagram demonstrates the relationships between the variables (Fig. 8 ). As can be seen, our model uses a simple structure where there is only one observed endogenous variable, “intended frequency of online grocery shopping after the pandemic”. The observed exogenous variables include a variety of socio-economic and demographic attributes, while 5 different latent factors were measured based on likert scale attitudinal statements and simultaneously used as predictors in the structural model.Fig. 8 Model path diagram. It should be noted that the dependent variable in this study (intention for increasing or reducing online grocery shopping after COVID-19) has an ordinal nature. Some research efforts in the literature supported the idea that an ordinal variable maybe treated as a continuous variable should it consist of numerous categories (usually four or more) (Göb et al., 2007, Byrne, 2013, Rhemtulla et al., 2012, Bentler and Chou, 1987). However, this is subject to further limitations such as an approximately normal distribution of categories or the symmetry of thresholds, which do not seem to stand true in our case. On the other hand, there are certain estimation methods for ordinal endogenous variables in SEM such as the Diagonally Weighted Least Square (DWLS) (Muthén and Muthén, 2007, Mîndrilã, 2010, Rhemtulla et al., 2012, Asgari and Jin, 2017). Hence, the dependent variable here is declared as an ordered variable and the coefficients were estimated using a DWLS approach. The measurement model (first step) can be expressed as follows:(1) Y=βy+ω Where, Y: The vector of LVs, β: The matrix of coefficients of attitudinal indicators on LVs, y: The vector of attitudinal indicators, ω: The error terms vector. The mathematical construct of a structural model, which examines the relationship between the dependent variable and explanatory variables (identified latent variables as well as observed exogenous variables), can be formulated as below:(2) U=[03B8]X+τY+∊ Where, U: Dependent variable (Reported online grocery shopping intention after COVID-19), θ: The coefficients matrix of the direct impacts of exogenous variables on the endogenous variable, X: The vector of exogenous variables (socioeconomic and demographic characteristics), τ: The coefficients matrix of the direct impact of LVs on the endogenous variable, Y: The vector of LVs, ∊: the error terms vector. The model was estimated using the latent variable analysis (lavaan) library in R using a Diagonally Weighted Least Square estimation approach for ordinal variables. 5 Model results The model performance results are presented in Table 2 . The Chi-Square test is a traditional performance measure assessing the “differences between the observed and fitted covariance matrices “(Hooper et al., 2008). Small (insignificant) p-values are generally desirable, but the chi-square test is sensitive to sample size. The p-value of the model suggests that there is not enough evidence to reject that the difference between the observed and fitted covariance matrices is insignificant (Shi et al., 2019). Hence, we can conclude that the model had acceptable performance. The Root Mean Square Error of Approximation (RMSEA) specifies the “discrepancy between a hypothesized and a perfect model” (Xia and Yang, 2019). The model showed an acceptable performance regarding this measure, indicating a convergence fit to the data. Standardized Root Mean Square Residual (SRMR) is an “absolute fit index measuring the average of the standardized residuals between the observed and model-implied covariance matrices” (Chen, 2007, Bentler, 1995). Considering the value of SRMR, we can conclude that the average magnitude of the discrepancies between observed and fitted covariance matrices is acceptable. The Comparative Fit Index (CFI) and the Tucker Lewis index (TLI) are “incremental fit indices comparing the fit of the hypothesized model with that of a baseline model (i.e., a model with the worst fit)” (Xia and Yang, 2019). Contrary to the Chi-square test, CFI and TLI are not sensitive to sample size (Xia and Yang, 2019, Fan et al., 1999). As shown in Table 2, the values of CFI and TLI suggest the better performance of the fitted model than the baseline model.Table 2 Summary of model performance measures. Criteria Value Cut-off criterion Additional information Chi-Square 1262.66 – P-value = 0.00 DoF = 317 Root Mean Square Error of Approximation (RMSEA) 0.05 RMSEA < 0.06 90% CI = [0.05, 0.06] Standardized Root Mean Square Residual (SRMR) 0.05 SRMR < 0.08 NA Comparative Fit Index (CFI) 0.94 CFI > 0.90 NA Tucker Lewis index (TLI) 0.96 TLI > 0.95 NA The structure of the SEM developed for this study is presented in Fig. 8. Significant latent factors and their indicators, as well as SED characteristics, are illustrated. 5.1 Measurement model results Table 3 shows the results of the measurement model. In total, five latent attitudes were identified, each with at least three indicators (which were significant at the 99% significance level). The latent attitudes were significant in terms of both the Z-value and the standardized factor loadings (the minimum value considered for the factor loading is 0.5).Table 3 Results of Measurement Model. Attitude Indicators (answer choices to the attitude questions) Factor Loading Z-Value Online Grocery Shopping Experience Frequency of online grocery shopping for delivery-Before COVID-19 1.00 – Frequency of online grocery shopping for pick up-Before COVID-19 1.07* 23.91 Frequency of online grocery shopping for delivery-During COVID-19 1.23* 17.51 Frequency of online grocery shopping for pick up-During COVID-19 1.46* 16.41 Pro-Online Grocery Shopping Online shopping for groceries is more efficient 1.00 – Online shopping for groceries is more convenient 0.96* 27.33 Online grocery shopping helps meeting my daily needs 1.01* 26.85 Ordering groceries online requires a minimal amount of mental effort 0.86* 22.04 Ordering groceries online requires a minimal amount of physical effort 0.91* 20.94 The interaction with the online grocery shopping websites or apps is, overall, easy and understandable 1.02* 23.30 If I have a problem with the grocery shopping platforms, I can easily get the support I need. 0.81* 21.44 Pro-Technology I know more than others about the latest technologies. 1.00 – I often purchase new technology products, even though they are expensive 1.12* 14.66 I like trying things that are new and different 0.84* 12.29 Internet & Communication Technologies can substitute personal need for travel 0.88* 11.33 Pro-Driving I like the freedom of driving my own car 1.00 – Driving a car is a relaxing way to commute 0.78* 10.02 Pro-Telecommuting I like working from home 1.00 Video calling is a good alternative to in-person business meetings 1.29* 11.08 I enjoy spending time with the people I live with 1.00*** 9.57 , , respectively denote significance at 0.1, 0.05, and 0.01 level. The first attitude, labeled as online grocery shopping experience, is related to online grocery shopping frequency (for both delivery and pickup) before and during the COVID-19 pandemic. The second attitude, which is defined as pro-online grocery shopping, reflects the positive attitude toward online grocery shopping’s efficiency, convenience, usefulness, and easiness. The third attitude, labeled as pro-technology, refers to the positive tendency toward technology use. This attitude is related to one’s knowledge about the latest technology, prior purchases of new technological products regardless of their prices, the inclination to try new and different things, and perception of Internet and Communication Technologies (ICT) as a potential substitute for travel. The pro-driving factor reflects the inclination toward private cars due to the convenience and the joy of driving. The pro-telecommuting factor denotes the respondents’ positive attitude toward telecommuting. 5.2 Structural model results 5.2.1 Attitudinal factors Table 4 presents the results of the structural model. In terms of attitudinal factors, the results showed that online grocery shopping experience (i.e., higher frequency of online grocery shopping before and during the COVID-19 pandemic) has a positive impact on more frequent online grocery shopping after the pandemic. This positive impact was expected as having a pleasant online grocery shopping experience can build individuals’ trust and favorable perception, leading to more frequent use of these services. This finding is consistent with previous studies (Hansen, 2006, Min et al., 2012, Anschuetz, 1997). As expected, individuals with a pro-online grocery shopping attitude are more likely to shop groceries online after the COVID-19 pandemic. Similarly, the literature showed that perceived usefulness and easiness of online grocery shopping are the key factors that encourage the shift from in-store to online shopping of groceries (Driediger and Bhatiasevi, 2019, Bauerová and Klepek, 2018, Huang and Oppewal, 2006, Chakraborty, 2019, Shukla and Sharma, 2018).Table 4 Results of Structural Model. Dependent variable Explanatory variables Category Coeff. Z-Value Reported Intention of Online Grocery Shopping After COVID-19 Is No Longer a Threat Attitudes Online Grocery Shopping Experience 0.31* 7.90 Pro-Online Grocery Shopping 0.35* 6.90 Pro-Driving −0.11 −2.46 Education Doctorate Degree 0.76* 4.25 Professional Degree 0.68* 3.69 Employment Status Part-time −0.24 −2.50 Unemployed −0.36* −4.49 Household Composition living with roommate only −0.42*** −2.97 living with significant other only (no kids) −0.16* −1.90 Number of Non-working, Non-student Adults 1 to 3 Adults 0.20* 2.78 Pro-Online Grocery Shopping Attitudes Pro-technology 1.47* 6.59 Pro-Telecommuting −0.47*** −2.56 , , ** respectively denote significance at 0.1, 0.05, and 0.01 level. Interestingly, people who enjoy driving and the freedom associated with private vehicles are less inclined to do online grocery shopping after the pandemic. One potential explanation is that individuals with this attitude see driving as leisure, and they enjoy trips to the stores and being out of home for a change of environment. This attitude could strongly encourage people to make physical trips to grocery stores, which might affect their intentions for future online grocery shopping. This finding is consistent with the findings by Casas et al. (2001). The only two mediation effects observed in our model is the indirect effects associated with pro-technology and pro-telework attitudes. Accordingly, both effects are mediated through the pro-online grocery shopping. The pro-technology attitude has an indirect positive impact on the reported online grocery shopping intention after the pandemic, mediated by the pro-online grocery shopping. A similar finding was presented by previous studies (Park et al., 1996, Etminani-Ghasrodashti and Hamidi, 2020). The relationship between pro-telecommute attitude and online-shopping attitude is a bit complicated though, yet worth noticing. During their model training efforts, authors noticed that in the absence of pro-technology attitude, telework by itself increases online shopping. This positive correlation could stem from the root idea that welcoming new technology and adopting it as a replacement for conventional way of doing things can improve quality of life. Benefits such as avoiding the disutility of out-of-home travel and more flexible and efficient time management could be names as simple examples of such improvements. However, when tech-savviness is accounted for in the structural model, telework impacts turns into negative. In this regard, authors believe that telecommute impacts could be viewed in two separate directions: a technology aspect, and a travel behavior aspect. The technology aspect, which is in common with online shopping attitude, is likely to encourage online shopping; The second aspect, which is related to the relaxed spatial–temporal constraints (as opposed to regular commute to work), has a tendency to increase the chance of other out-of-home non-mandatory activities during the day (Asgari et al., 2016, Asgari and Jin, 2017), which may potentially counteract the first positive effect and increase the likelihood of in-store shopping. Therefore, when the pro-technology attitude is incorporated in the model, the rest of telecommute impacts turn out to address the negative effects of increased out-of-home activities (due to the relaxed constraints) on online behavior. 5.2.2 Socioeconomic and demographic attributes In view of socioeconomic and demographic characteristics, results indicated that individuals with a high education level (i.e., a doctorate or a professional degree) are more likely to shop online for their groceries in the future. People with higher education levels probably are in the higher income brackets and might view the additional fees associated with online grocery shopping services as trivial. Also, people with higher education levels tend to spend more time searching for products online, which might encourage them to have a higher frequency of online grocery shopping (Etminani-Ghasrodashti and Hamidi, 2020, Zhen et al., 2018). Previous studies also showed that the value of time (VOT) might be higher for these individuals; therefore, online grocery shopping might be more attractive to them to save time for other activities (Punj, 2011, Van Droogenbroeck and Van Hove, 2017). In terms of employment status, part-time and unemployed individuals are less likely to do online grocery shopping after the pandemic. Online grocery shopping is usually associated with higher unit costs as well as service and delivery fees, which might not be favorable for low-income individuals. Our model shows that simple household structures (e.g., two roommates living together or a couple with no kids) are less likely to engage in online shopping. Smaller households usually buy lesser amounts of groceries, where the burden to shop and carry is negligible compared to larger households (Bawa and Ghosh, 1999). Moreover, a previous study showed that social interaction might be a strong motivation for people to choose in-store shopping over online grocery shopping. Van Droogenbroeck and Van Hove (2017) suggested that small sized households are more likely to view social interactions as incentives for favoring in-store shopping compared to larger households. Furthermore, with the presence of 1 to 3 adults, who are neither workers nor students, individuals are more likely to shop online for their groceries. This group might represent retired seniors. Online grocery shopping might be desired for seniors requiring mobility assistance because in-store shopping might be difficult and inconvenient (Morganosky and Cude, 2000). 6 Policy implications Practically, one of the main outcomes of a consumer survey analysis is the market segmentation. Market segmentation is the process of dividing a broad range of consumers into sub-classes based on some type of shared characteristics. For instance, our study showed that employment, education, household composition, as well as personal attitudes can help classify current or potential online shoppers, with full time employees, individuals with higher educations, those coming from larger or more complicated households, tech savvy people, and individuals with prior experience to online shopping are more likely to be frequent online shoppers in the post pandemic condition. On the other hand, part-time workers, unemployed individuals, simple households consisting of roommates or couples without children, and pro driving individuals are less likely to favor online shopping. This simple inference can provide valuable insights to businesses on how to classify their target market, which will further help them in a variety of promotional activities such as advertising and marketing strategies. For instance, by detecting tech savvy people, they could be a rewarding target for online advertising methods such as social media platforms, while other groups such as people with lower education or senior families might respond more efficiently to conventional advertising methods such as TV commercials or flyers mailed to their door. Market segmentation also results in a better understanding of the current and potential demand forecasting. For instance, disaggregate models like what was developed in this study could be further aggregated at a local neighborhood level. Clustering neighborhoods in view of their current/potential demand based on the residents’ attributes will provide useful information to a variety of stakeholders. Business owners can use it to expand the size or the number of their stores or delivery hubs in certain areas. It also helps them identify underlying reasons in low demand neighborhoods and rethink incentive strategies such as discounts, free deliveries, or other well-defined amenities to attract latent markets in those neighborhoods. From a transportation planning perspective, expansion of E commerce usually calls for a number of changes in urban logistics (Rutter et al., 2017). In particular, building warehouses and fulfillment centers should meet certain transportation related as well as urban planning standards. Warehouses are usually developed within or in proximity of urban areas and require access to interstate highways to accommodate large trucks and other freight transportation facilities. Fulfillment and distribution centers are also expected to increase the inbound/outbound truck concentration in local roads that provide access to urban areas. Such frequent freight movements should be considered by MPOs and other local planning agencies when it comes to resource allocation in their plans. Local roads might need reconstruction or undergo rehabilitation programs to withstand the additional truck traffic. Increased maintenance activities and relevant costs should also be foreseen in the budget allocation. It is also important that location of such centers meets the required local and federal standards in relation to other adjacent land uses such as schools, residential areas, and other types of industrial and non-industrial uses. In addition, expansion of e commerce in the neighborhood with increase the number of relevant employees which in turn increases number of commute trips that not only increases peak hour traffic congestions but might also call for carpooling services to be considered in the area for those working similar shifts. Groceries are also a good candidate for express or same-day delivery trips. Such deliveries are expected to increase non-peak traffic congestions in the local area. A recent study in Texas showed that many of the express deliveries are accomplished by non-commercial vehicles (Rutter et al., 2017). Such traffic is not usually accounted for in regional transportation planning models and might need to be implemented accordingly. In addition, certain types of standards and certification might be required for drivers and vehicles that deliver perishable goods. From an urban planning standpoint, delivery vehicles require access to some type of drop-off or short-term parking. Such freight delivery needs should be discussed at municipality level. Lack of such amenities can result in increased congestion in denser local areas. In addition, when there is no requirement for face-to-face delivery, there should be additional storage space provided in multi-unit housings to allow package drop off in entry or common areas. This could be a new challenge for multi-family housing developers in dense areas with high demand for online shopping. 7 Summary and conclusion As information and communication technologies continue to reshape human behavior and lifestyles in a variety of perspectives, there is a substantial need to foresee and plan for such technology-induced changes. In particular, the impacts of ICT on activity and travel related decisions of individuals has arisen the interest of researchers and planners. While a variety of different theories have been developed and tested in recent years, the pandemic experience provided a unique empirical opportunity for the researchers to observe and analyze how ICT adoption can substitute the normal way of doing things in practice, what the existing challenges are, and how different segments of a large population of people will react in response to new policies. This study explicitly targets the concept of online shopping. While the literature suggests that online shopping behavior is strongly dependent on the type of goods being discussed, we narrow down our analysis to grocery shopping only. Data from an online survey conducted during the COVID 19 pandemic were analyzed for this study. The survey consists of information on socioeconomic and demographic attributes, online grocery shopping experience, as well as attitudinal questions about technology use, telecommuting, online grocery shopping, and driving. A structural equation model was applied to explore the impacts of the contributing factors on the reported online grocery shopping for delivery after the pandemic. The descriptive analysis of the data showed a substantial increase in the frequency of online grocery shopping as well as the number of first-time online grocery shoppers during the pandemic. Besides, it was observed that the online grocery shopping frequencies before and during the pandemic are associated with the intention for using these services after the pandemic. In view of attitudinal questions, most respondents believed in the convenience of online grocery shopping; however, the mental effort required in choosing products from pictures on the web pages might still present barriers for some shoppers. The model results indicated that those who showed a higher frequency of online grocery shopping during the pandemic would also be more likely to shop more frequently for groceries online after the pandemic. This positive correlation might emphasize the positive role of prior experience or familiarity, where consumers with higher familiarity are more comfortable with the service and continue using it. Moreover, consistent with findings in the literature, positive attitudes toward online grocery shopping which stem from higher levels of efficiency, convenience, usefulness, and easiness would lead to more frequent online grocery shopping in the long term. On the other hand, those who enjoy driving or the freedom associated with private vehicles are less likely to adopt online grocery shopping for delivery after the pandemic. In view of the pro-telecommuting attitude, authors believe that extra attention should be paid. In general, and in the absence of tech-savviness information, the pro-telecommuting attitude has a positive correlation with online shopping attitude. This is mainly rooted in the common values that pro-technology people believe in, that ICT applications regardless of form, can improve the quality of life both at individual and society level (Shabanpour et al., 2018, Asgari et al., 2019). However, when tech-savviness is directly and separately addressed in the model (just as we see in our final model here), telework tends to moderate the positive technology-induced impact by exerting a negative effect, probably associated with the higher temporal/spatial freedom that allows a teleworker to pursue non-mandatory out-of-home activities, such as in-store shopping. In terms of socioeconomic and demographic characteristics, the findings indicated that unemployed individuals or part time workers have a negative tendency toward online grocery shopping after the pandemic, probably associated with their income level or lifestyle preferences. On the other hand, full-time employed individuals, and those with higher levels of education are more likely to shop groceries online, which could be attributed to their higher income levels, familiarity with technologies, and higher value of time. As expected, larger households with more complicated structures are associated with a higher reported online grocery shopping intention after COVID-19 after the pandemic, perhaps because of higher demands for groceries and a lower need for social interaction. COVID-19 has changed individuals’ shopping behavior and habits, especially when it comes to shopping for groceries. While many online grocery shoppers could go back to traditional in-store shopping after the pandemic, some others might continue online grocery shopping. Although literature predicted that the pandemic has brought about fundamental changes in individuals’ shopping behavior, and people probably will be more interested in online grocery shopping after the pandemic, they did not mention the attributes that increase the likelihood of future online grocery shopping. With a focus on attitudinal factors, this study fills this gap and explores the attributes that contribute to online grocery shopping after the pandemic. This paper provides a valuable perspective on how to improve our ability to predict the potential changes in the pattern of online grocery shopping activities after the pandemic. The findings of this study suggest that people’s grocery shopping behavior after the pandemic might be subject to change, at least based on their preference. In this regard, the results indicated that the number of people who shop their groceries online considerably increased during the pandemic. It was also found that people with a higher frequency of online grocery shopping during the pandemic are more likely to continue buying their groceries online in the future. This finding suggests that although there might be a decrease in the number of people who do online grocery shopping after the pandemic, the post-pandemic number is likely to be higher than that before the lock-down period. The switch to online grocery shopping is especially substantial for those with positive attitudes toward online grocery shopping as well as people with a technology-based lifestyle. As evidenced by our experience throughout the COVID-19 pandemic, emerging trends toward the adoption of new technologies and services have significant impacts on the transportation systems. However, the potential effects of these trends on travel demand cannot be reasonably predicted by conventional models that rely on demographic and socioeconomic variables alone. In the future, agencies conducting regional household travel surveys can consider designing questions to collect respondents’ attitudes toward technology adoption (e.g., online grocery shopping and other online services; electric and automated vehicles) and other emerging socioeconomic trends (e.g., lower birth rate). Exploring individuals’ attitudes toward these trends offers additional information for modelers to build forecasting models that consider how attitudes intermediate between travelers’ characteristics and their travel behavior. This study is subject to a number limitations. Speaking of methodology, since the dependent variable (the reported intention of online grocery shopping expressed in five ordinal scales) reports the preferred (intended) frequency and therefore is more of a qualitative and relative nature, the results of the study cannot be interpreted as a prediction of future online grocery shopping frequency. Future studies based on quantitative dependent variables (e.g., revealed frequency of online grocery shopping) can be used to develop predictive models for online grocery shopping frequency. In addition, taking into account that estimating attitudes at personal level is not feasible, one might consider predicting them based on socio-economic and demographic attributes, which could be considered as a further enhancement of the model structure proposed in this study. One might also simultaneously model the pre and post pandemic frequency of online shopping frequency and test the causal relationship between the two. This may also help to compare the mechanism of the decisions in the two timelines and probably observe the evolution of attitudes during the pandemic condition. From a data perspective, it should be noted that not all respondents in our survey are employed or could telecommute, hence the impact of telecommuting on the dependent variable is subject to further scrutinization. Future studies can verify this with data from only employed individuals and/or those with telecommuting opportunities. In addition, because the dataset for this study is limited to South Florida, replication of this research with state- or national-level data can help verify and expand the findings of this study. CRediT authorship contribution statement Hamidreza Asgari: Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. Ghazaleh Azimi: Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. Ibukun Titiloye: Investigation, Writing – original draft, Writing – review & editing. Xia Jin: Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Writing – original draft, Writing – review & editing. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References Abou-Zeid, G., Clifton, K.J., Unnikrishnan, A., Anderson, J., & Yang, D.Y. (2021). Adoption and Use of E-Grocery Shopping in the Context of the COVID-19 Pandemic: Implications for Transport Systems and Beyond. Alaimo L.S. Fiore M. Galati A. How the covid-19 pandemic is changing online food shopping human behaviour in Italy Sustainability 12 22 2020 9594 10.3390/SU12229594 Anschuetz N. Profiting from the 80–20 rule of thumb J. Advert. Res. 37 6 1997 51 57 Asgari H. Jin X. Impacts of telecommuting on nonmandatory activity participation: role of endogeneity Transp. Res. Rec. 2666 1 2017 47 57 10.3141/2666-06 Asgari H. Jin X. Du Y. Examination of the impacts of telecommuting on the time use of nonmandatory activities Transp. Res. Rec. 2566 1 2016 83 92 10.3141/2566-09 Asgari H. Jin X. Rojas M.B. IV Time geography of daily activities: a closer look into telecommute impacts Travel Behav. Soc. 16 2019 99 107 Baarsma B. Groenewegen J. COVID-19 and the demand for online grocery shopping: empirical evidence from the Netherlands Economist (Netherlands) 169 4 2021 407 421 10.1007/S10645-021-09389-Y/TABLES/3 Badri M. Adoption of Online Shopping Apps Innovation on Digital Natives Generation In 2020 International Conference on ICT for Smart Society (ICISS) 2020 IEEE 1 5 Bauerová R. Klepek M. Technology acceptance as a determinant of online grocery shopping adoption Acta Universitatis Agriculturae et Silviculturae Mendelianae Brunensis 66 3 2018 737 746 Bawa K. Ghosh A. A model of household grocery shopping behavior Mark. Lett. 10 2 1999 149 160 Beckers J. Verhetsel A. The sustainability of the urban layer of e-commerce deliveries: the Belgian collection and delivery point networks Eur. Plan. Stud. 29 12 2021 2300 2319 Bentler P.M. EQS Structural Equations Program Manual Vol. 6 1995 Multivariate Software Encino, CA Bentler P.M. Chou C.P. Practical issues in structural modeling Sociol. Methods Res. 16 1 1987 78 117 Bezirgani A. Lachapelle U. Online grocery shopping for the elderly in Quebec, Canada: the role of mobility impediments and past online shopping experience Travel Behav. Soc. 25 2021 133 143 10.1016/J.TBS.2021.07.001 Bjerkan K.Y. Bjørgen A. Hjelkrem O.A. E-Commerce and prevalence of Last mile practices Transp. Res. Procedia 46 2020 293 300 10.1016/j.trpro.2020.03.193 Bollen K.A. Structural Equations With Latent Variables Vol. 210 1989 John Wiley & Sons Bollen, K.A., & Long, J.S. (Eds.). (1993). Testing structural equation models (Vol. 154). Sage. Byrne, B.M., 2013. Structural equation modeling with EQS: Basic concepts, applications, and programming. Routledge. Cao X.J. Xu Z. Douma F. The interactions between e-shopping and traditional in-store shopping: an application of structural equations model Transportation 39 5 2012 957 974 Casas, J., Zmud, J., Bricka, S., 2001, January. Impact of shopping via Internet on travel for shopping purposes. In 80th Annual Meeting of the Transportation Research Board, Washington, DC. Chakraborty D. Indian shoppers’ attitude towards grocery shopping apps: a survey conducted on smartphone users Metamorphosis 18 2 2019 83 91 Chang H.H. Meyerhoefer C.D. COVID-19 and the demand for online food shopping services: Empirical Evidence from Taiwan Am. J. Agric. Econ. 103 2 2021 448 465 Charlebois, S., 2020. Why COVID-19 will change Canadian grocery industry forever: Expert. Retail Insider. Chen F.F. Sensitivity of goodness of fit indexes to lack of measurement invariance Struct. Equ. Model. Multidiscip. J. 14 3 2007 464 504 Christ S.L. Lee D.J. Lam B.L. Zheng D.D. Structural equation modeling: a framework for ocular and other medical sciences research Ophthalmic Epidemiol. 21 1 2014 1 13 24467557 Dastane O. Impact of digital marketing on online purchase intention: mediation effect of customer relationship management J. Asian Bus. Strategy 10 1 2020 142 158 Ding Y. Lu H. The interactions between online shopping and personal activity travel behavior: an analysis with a GPS-based Activity Travel diary Transportation 44 2 2017 311 324 10.1007/s11116-015-9639-5 Donthu N. Gustafsson A. Effects of COVID-19 on business and research J. Bus. Res. 117 2020 284 289 32536736 Driediger F. Bhatiasevi V. Online grocery shopping in Thailand: consumer acceptance and usage behavior J. Retail. Consum. Serv. 48 2019 224 237 e Silva J.D.A. de Oña J. Gasparovic S. The relation between travel behaviour, ICT usage and social networks. The design of a web based survey Transp. Res. Procedia 24 2017 515 522 Ellison B. McFadden B. Rickard B.J. Wilson N.L. Examining food purchase behavior and food values during the COVID-19 pandemic Appl. Econ. Perspect. Policy 43 1 2021 58 72 Eriksson N. Stenius M. Online grocery shoppers due to the Covid-19 pandemic - An analysis of demographic and household characteristics Procedia Comput. Sci. 196 2022 93 100 10.1016/J.PROCS.2021.11.077 Etminani-Ghasrodashti R. Hamidi S. Online shopping as a substitute or complement to in-store shopping trips in Iran? Cities 103 2020 102768 Fan X. Thompson B. Wang L. Effects of sample size, estimation methods, and model specification on structural equation modeling fit indexes Struct. Equ. Model. Multidiscip. J. 6 1 1999 56 83 Farag S. Schwanen T. Dijst M. Faber J. Shopping online and/or in-store? A Structural Equation Model of the Relationships between e-shopping and in-store shopping Transp. Res. A Policy Pract. 41 2 2007 125 141 10.1016/j.tra.2006.02.003 Ferrell C.E. Home-Based Teleshoppers and shopping travel: Do Teleshoppers travel less? Transp. Res. Rec. 1894 1 2004 241 248 10.3141/1894-25 Ferrell C.E. Home-based teleshopping and shopping travel: where do people find the time? Transp. Res. Rec. 1926 1 2005 212 223 10.1177/0361198105192600125 Forbes, (2020). Lasting changes to grocery shopping after covid-19. Forsythe S.M. Shi B. Consumer patronage and risk perceptions in Internet shopping J. Bus. Res. 56 11 2003 867 875 Göb R. McCollin C. Ramalhoto M.F. Ordinal methodology in the analysis of Likert scales Qual. Quant. 41 5 2007 601 626 Golob T.F. Structural equation modeling for travel behavior research Transp. Res. B Methodol. 37 1 2003 1 25 Grashuis J. Skevas T. Segovia M.S. Grocery shopping preferences during the COVID-19 pandemic Sustainability 12 13 2020 5369 Hansen T. Determinants of consumers' repeat online buying of groceries Int. Rev. Retail Distrib. Consum. Res. 16 1 2006 93 114 Hassen T. el Bilali H. Allahyari M.S. Impact of COVID-19 on food behavior and consumption in Qatar Sustainability 12 17 2020 6973 10.3390/SU12176973 Hiselius L.W. Rosqvist L.S. Adell E. Travel behaviour of online shoppers in Sweden Transp. Telecommun. 16 1 2015 21 30 10.1515/ttj-2015-0003 Hjorthol R.J. Information searching and buying on the internet: travel-related activities ? Environ. Plann. B. Plann. Des. 36 2 2009 229 244 Hooper D. Coughlan J. Mullen M. Evaluating model fit: a synthesis of the structural equation modelling literature In 7th European Conference on Research Methodology for Business And Management Studies 2008 195 200 Huang Y. Oppewal H. Why consumers hesitate to shop online: an experimental choice analysis of grocery shopping and the role of delivery fees Int. J. Retail Distrib. Manag. 34 4/5 2006 334 353 Irawan M.Z. Wirza E. Understanding the effect of online shopping behavior on shopping travel demand through structural equation modeling J. East. Asia Soc. Transp. Stud. 11 2015 614 625 Kaplan D. Structural Equation Modeling: Foundations and Extensions Vol. 10 2008 SAGE Publications Koch J. Frommeyer B. Schewe G. Online shopping motives during the COVID-19 pandemic—lessons from the crisis Sustainability 12 24 2020 10247 Lachapelle U. Jean-Germain F. personal use of the internet and travel: evidence from the Canadian General Social survey’s 2010 time use module Travel Behav. Soc. 14 2019 81 91 10.1016/j.tbs.2018.10.002 Le H.T. Carrel A.L. Shah H. Impacts of online shopping on travel demand: a systematic review Transp. Rev. 42 3 2022 273 295 Lee R.J. Sener I.N. Mokhtarian P.L. Handy S.L. Relationships between the online and in-store shopping frequency of Davis, California residents Transp. Res. A Policy Pract. 100 2017 40 52 Li J. Hallsworth A.G. Coca-Stefaniak J.A. Changing grocery shopping behaviours among Chinese consumers at the outset of the COVID-19 outbreak Tijdchr. Econ. Soc. Geogr. 111 3 2020 574 583 Lo A. Duffy E. Ng S.W. Who’s grocery shopping online and why: cross-sectional analysis of a nationally-representative sample since the pandemic Curr. Dev. Nutr. 5 Supplement_2 2021 231 10.1093/CDN/NZAB029_032 Matson G. McElroy S. Lee Y. Circella G. Longitudinal analysis of COVID-19 impacts on mobility: an early snapshot of the emerging changes in travel behavior Transp. Res. Rec:. J. Transp. Res. Board 2677 4 2021 298 312 Mercatus. (2020). eGrocery’s New Reality: The Pandemic’s Lasting Impact on U.S. Grocery Shopping Behavior. https://www.mercatus.com/newsroom/online-grocery-sales-projected-to-reach-250b-by-2025-according-to-new-research-from-mercatus-and-incisiv/. Mercatus. (2021). Market projections and insight into online grocery’s elevated future eGrocery Transformed in collaboration with. https://info.mercatus.com/online-grocery-shopper-consumer-behavior. Min S. Overby J.W. Shin Im K. Relationships between desired attributes, consequences and purchase frequency J. Consum. Mark. 29 6 2012 423 435 Mîndrilã D. Maximum likelihood (ML) and diagonally weighted least squares (DWLS) estimation procedures: a comparison of estimation bias with ordinal and multivariate non-normal data Int. J. Digital Soc. 1 1 2010 60 66 Mokhtarian P.L. A typology of relationships between telecommunications and transportation Transp. Res. Part A: General 24 3 1990 231 242 10.1016/0191-2607 (90)90060-J Mokhtarian P.L. Meenakshisundaram R. Beyond tele-substitution: disaggregate longitudinal structural equations modeling of communication impacts Transp. Res. Part C: Emerg. Technol. 7 1 1999 33 52 10.1016/S0968-090X(99)00010-8 Mokhtarian P.L. Salomon I. Handy S.L. The impacts of Ict on leisure activities and travel: a conceptual exploration Transportation 33 3 2006 263 289 10.1007/s11116-005-2305-6 Morganosky M.A. Cude B.J. Consumer response to online grocery shopping Int. J. Retail Distrib. Manag. 28 1 2000 17 26 Mortimer G. Fazal e Hasan S. Andrews L. Martin J. Online grocery shopping: the impact of shopping frequency on perceived risk Int. Rev. Retail Distrib. Consum. Res. 26 2 2016 202 223 Muthén L.K. Muthén B.O. Mplus User’s Guide 2007 Muthén & Muthén Los Angeles, CA Nahiduzzaman K.M. Aldosary A.S. Mohammed I. Framework analysis of E-commerce induced shift in the spatial structure of a city J. Urban Plann. Dev. 145 3 2019 04019006 Nahiduzzaman K.M. Holland M. Sikder S.K. Shaw P. Hewage K. Sadiq R. Urban transformation toward a smart city: an e-commerce–induced path-dependent analysis J. Urban Plann. Dev. 147 1 2021 04020060 Nanda A. Xu Y. Zhang F. How would the COVID-19 pandemic reshape retail real estate and high streets through acceleration of E-commerce and digitalization? J. Urban Manage. 10 2 2021 110 124 Njite D. Parsa H.G. Structural equation modeling of factors that influence consumer internet purchase intentions of services J. Serv. Res. 5 1 2005 43 Pan S. Roy A. Choi Y. Sun S.Q. Gao H.O. The air quality and health impacts of projected long-haul truck and rail freight transportation in the United States in 2050 Environ. Int. 130 2019 104922 Park, K., Perosio, D., German, G.A., McLaughlin, E.W., 1996. What’s in store for home shopping?. Pettersson F. Winslott Hiselius L. Koglin T. E-commerce and urban planning–comparing knowledge claims in research and planning practice Urban Plann. Transp. Res. 6 1 2018 1 21 Prasetyo, Y.T., & Fuente, D.G.D.D. (2020, September). Determinant factors affecting customer satisfaction among Filipinos in Lazada online shopping during COVID-19 pandemic: A structural equation modeling approach. In 2020 7th International Conference on Frontiers of Industrial Engineering (ICFIE) (pp. 48-52). IEEE. Punj G. Effect of consumer beliefs on online purchase behavior: the influence of demographic characteristics and consumption values J. Interact. Mark. 25 3 2011 134 144 Qi X. Tian X. Ploeger A. Exploring Chinese consumers’ online purchase intentions toward certified food products during the COVID-19 pandemic Foods 10 11 2021 2729 10.3390/FOODS10112729 34829010 Raijas A. The consumer benefits and problems in the electronic grocery store J. Retail. Consum. Serv. 9 2 2002 107 113 Rhemtulla M. Brosseau-Liard P.É. Savalei V. When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions Psychol. Methods 17 3 2012 354 373 22799625 Richards T.J. Rickard B. COVID-19 impact on fruit and vegetable markets Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 68 2 2020 189 194 Rita P. Oliveira T. Farisa A. The impact of e-service quality and customer satisfaction on customer behavior in online shopping Heliyon 5 10 2019 e02690 31720459 Rutter, A., Bierling, D., Lee, D., Morgan, C., Warner, J. (2017). How Will E-commerce Growth Impact Our Transportation Network? Final Report, Transportation Policy Research Center, Texas A&M Transportation Institute, PRC 17-79 F. Salomon I. Telecommunications and travel relationships: a review Transp. Res. Part A: General 20 3 1986 223 238 10.1016/0191-2607(86)90096-8 Schade, K., Hübscher, M., & Korzer, T. (2018, July). Smart Retail in Smart Cities: Best Practice Analysis of Local Online Platforms. In ICETE (1) (pp. 313-323). Sarker A.M.A. Asgari H. Jin X. Aging Population and Automated Mobility: Exploring the Impacts of Land Use Patterns and Attitudes Trans. Res. Rec. 2023 1 17 10.1177/03611981231160548 Schumacker R.E. Lomax R.G. A Beginner’s Guide to Structural Equation Modeling 2004 Psychology Press Seifert R.W. Markoff R. Amazon Fresh and the disruption of the supply chain: Disruption is not always what it seems Int. Inst. Manage. Dev. 2017 https://www.imd.org/research-knowledge/articles/amazon-fresh-and-the-disruption-of-the-supply-chain/ Shabanpour R. Golshani N. Tayarani M. Auld J. Mohammadian A.K. Analysis of telecommuting behavior and impacts on travel demand and the environment Transp. Res. Part D: Transp. Environ. 62 2018 563 576 Shamim K. Ahmad S. Alam M.A. COVID-19 health safety practices: Influence on grocery shopping behavior J. Public Aff. 2021 e2624 33786020 Shamshiripour A. Rahimi E. Shabanpour R. Mohammadian A. How is COVID-19 reshaping activity-travel behavior? Evidence from a comprehensive survey in Chicago Transp. Res. Interdiscip. Perspect. 7 2020 100216 Shi K. De Vos J. Yang Y. Li E. Witlox F. Does E-shopping for intangible services attenuate the effect of spatial attributes on travel distance and duration? Transp. Res. A Policy Pract. 141 2020 86 97 10.1016/j.tra.2020.09.004 Shi K. Cheng L. De Vos J. Yang Y. Cao W. Witlox F. How does purchasing intangible services online Influence the travel to consume these services? A focus on a Chinese context Transportation 48 5 2020 2605 2625 10.1007/s11116-020-10141-9 Shi D. Lee T. Maydeu-Olivares A. Understanding the model size effect on SEM fit indices Educ. Psychol. Meas. 79 2 2019 310 334 30911195 Shukla A. Sharma S.K. Evaluating consumers’ adoption of mobile technology for grocery shopping: an application of technology acceptance model Vision 22 2 2018 185 198 Singh S. Santhakumar S.M. Empirical analysis of impact of multi-class commercial vehicles on multi-lane highway traffic characteristics under mixed traffic conditions Int. J. Transp. Sci. Technol. 2021 Suel E. Polak J.W. Development of Joint models for channel, store, and travel mode choice: grocery shopping in London Transp. Res. A Policy Pract. 99 2017 147 162 10.1016/j.tra.2017.03.009 Tarhini, A., Alalwan, A.A., Al-Qirim, N., Algharabat, R. (2021). An analysis of the factors influencing the adoption of online shopping. In Research Anthology on E-Commerce Adoption, Models, and Applications for Modern Business (pp. 363-384). IGI Global. Titiloye I. Sarker M.A.A. Jin X. Watts B. Examining channel choice preferences for grocery shopping during the Covid-19 pandemic Int. J. Transp. Sci. Technol. 2023 10.1016/J.IJTST.2023.03.006 Truong D. Truong M.D. How do customers change their purchasing behaviors during the COVID-19 pandemic? J. Retail. Consum. Serv. 67 2022 102963 10.1016/J.JRETCONSER.2022.102963 Tyrväinen O. Karjaluoto H. Online grocery shopping before and during the COVID-19 pandemic: a meta-analytical review Telematics Inform. 71 2022 101839 10.1016/J.TELE.2022.101839 Van Droogenbroeck E. Van Hove L. Adoption of online grocery shopping: personal or household characteristics? J. Internet Commer. 16 3 2017 255 286 Warganegara D.L. Hendijani R.B. Factors that drive actual purchasing of groceries through E-Commerce platforms during COVID-19 in Indonesia Sustainability 14 6 2022 3235 10.3390/SU14063235 Weltevreden J.W.J. Substitution or complementarity? How the Internet changes City centre shopping J. Retail. Consum. Serv. 14 3 2007 192 207 10.1016/j.jretconser. 2006.09.001 Xia Y. Yang Y. RMSEA, CFI, and TLI in structural equation modeling with ordered categorical data: the story they tell depends on the estimation methods Behav. Res. Methods 51 1 2019 409 428 29869222 Yang Z.Z. Yu S. Lian F. Online shopping versus in-store shopping and its implications for urbanization in China: based on the shopping behaviors of students relocated to a remote campus Environ. Dev. Sustain. 23 2 2021 2846 2866 Zhao A. Li Z. Ke Y. Huo S. Ma Y. Zhang Y. Zhang J. Ren Z. Dietary diversity among Chinese residents during the COVID-19 outbreak and its associated factors Nutrients 12 6 2020 1699 32517210 Zhen F. Du X. Cao J. Mokhtarian P.L. The association between spatial attributes and e-shopping in the shopping process for search goods and experience goods: evidence from Nanjing J. Transp. Geogr. 66 2018 291 292

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.