ijerph8

Med-MDPI/ijerph_8/ijerph-18-21-11018.txt

@@ -0,0 +1 @@
+Fuse–UKCRC Centre for Translational Research in Public Health, A virtual centre, operating across five universities in north-east England: Durham, Newcastle, Northumbria, Sunderland and Teesside, UK.This paper provides a longitudinal examination of local inequalities in health behaviours during a period of austerity, exploring the role of ‘place’ in explaining these inequalities. Data from the Stockton-on-Tees prospective cohort study of 836 individuals were analysed and followed over 18 months (37% follow-up). Generalised estimating equation models estimated the deprivation gap in health behaviours (smoking status, alcohol use, fruit and vegetable consumption and physical activity practices) between the 20% most- and least-deprived neighborhoods (LSOAs), explored any temporal changes during austerity, and examined the underpinning role of compositional and contextual determinants. All health behaviours, except for frequent physical activity, varied significantly by deprivation (p ≤ 0.001). Smoking was lower in the least-deprived areas (OR 0.21, CI 0.14 to 0.30), while alcohol use (OR 2.75, CI 1.98 to 3.82) and fruit and vegetable consumption (OR 2.55, CI 1.80 to 3.62) were higher in the least-deprived areas. The inequalities were relatively stable throughout the study period. Material factors (such as employment, education and housing tenure) were the most-important and environmental factors the least-important explanatory factors. This study suggests that material factors are the most important ‘place’ determinants of health behaviours. Health promotion activities should better reflect these drivers.Geographical inequalities in health behaviours (smoking status, alcohol use, fruit and vegetable consumption and physical activity levels) are present at all spatial scales—between neighbourhoods, local authorities, regions and countries [1]. For example, a number of studies have found that there are inequalities in smoking prevalence, with higher rates in places with higher levels of deprivation [2,3,4]. The socio-spatial distribution of alcohol consumption is less clear; whilst some studies have reported that it is associated with area-level deprivation [5], other literature reports an absence of association [6]. A more clear finding is that the most severe levels of alcohol use–and alcohol-related harm—are more prevalent in more-deprived areas. In terms of physical activity, a number of studies have found an inverse association with deprivation [2,5]—people living in more deprived areas have lower levels of physical activity. In terms of diet it is generally recognised that those in more deprived areas consume less fruit and vegetables [6]. Geographical research has highlighted the important role that ‘place’ has in shaping these socio-spatial inequalities in health behaviours. In terms of smoking, research has found that areas with a higher density of tobacco retailers have higher smoking rates [7]. Further, the social and spatial stigmatisation of smoking may create areas with higher concentrations of smokers—smoking islands [8]. Associations between higher alcohol consumption and the density of alcohol retail outlets have also been found [9]. Associations with the neighbourhood environment have also been found with regard to physical activity. For example, a study conducted in the USA found that those in less-deprived areas are able to be more physically active as the areas they live in have more facilities available to allow for such activities [10]. Further, residents of areas with a higher density of fast-food outlets have lower rates of fruit and vegetable consumption [11] and food deserts and obesogenic environments increase excessive food intake [12].These studies draw on the idea that ‘place’ matters for the socio-spatial distribution of health and health behaviour as the characteristics of places can promote salutogenic or pathogenic health behaviors [1]. The definition of place is often contentious [13] but it can be considered as a specific geographical location or area, requiring shared experience and bounded membership [14]. Variations between the characteristics of individuals and their interaction with the social, economic and physical environment thereby combine to shape the nature of a specific place [15]. Much of the literature addressing the relationship between health and place conceptualises these mediating factors into two main categories: compositional and contextual. The compositional explanation argues that geographical inequalities in health and health behaviours arise from the individual characteristics of the people that live in the areas, most notably in terms of individual material (e.g., income, housing and employment) and psychosocial (e.g., control and self-worth) circumstances [13,15,16]. On the other hand, the contextual explanation argues that geographical inequalities in health are explained by the characteristics of the local place in which individuals live. This can include factors such as area-level unemployment and access to social and physical resources such as food, health care and green space [2,17,18]. A more recent framing of health and place is the political economy one [1,15,19]. It ‘scales-up’ the contextual explanation by highlighting the importance of macro-economics, political choices and public policies—beyond the characteristics of individuals and locales—in shaping both place and health [19,20,21]. A key example of the influences of political economy on health inequalities comes from studies of austerity (a policy response to the 2007/8 financial crisis whereby the UK government—amongst others—reduced public spending and welfare benefits) [22]. The studies found that inequalities in health increased [23,24,25,26]. Of course, compositional, contextual and macro political economy factors cannot be viewed independently—they are not mutually exclusive but are relational processes that interact with each other [27]. The characteristics of individuals will often be influenced by the characteristics of their surroundings and societal structures and vice versa. This study adds to this important international body of literature by examining the relational nature of contextual, compositional and political economy influences on geographical inequalities in health behaviours through data from the Stockton-on-Tees cohort study. This study examines three main research questions:Are there inequalities in health behaviours between the most- and least-deprived neighbourhoods of Stockton-on-Tees?Which place-based factors are associated with health behaviours?Do any inequalities in health behaviours change temporally during austerity?Are there inequalities in health behaviours between the most- and least-deprived neighbourhoods of Stockton-on-Tees?Which place-based factors are associated with health behaviours?Do any inequalities in health behaviours change temporally during austerity?This paper assessed inequalities in health behaviours, factors associated with health behaviours at baseline, and changes in inequality gap over the study period using data from a prospective 18-month household cohort survey of health and the social and behavioural determinants of health. The study was conducted in the most- and least-deprived areas of Stockton-on-Tees, a local authority in the North East of England (Figure 1). Stockton-on-Tees was chosen as the study site because, at baseline, it had the highest neighborhood (lower super output area level (LSOA)) health inequalities in England both for men (a 17-year difference in life expectancy at birth) and for women (11-year gap) [28]. This makes it a particularly important case study for the analysis of health inequalities during austerity. This research is relevant to other local authority areas with similar levels of deprivation and inequality, particularly those in North East England, such as Middlesbrough, Redcar and Cleveland, Gateshead, North Tyneside and Newcastle upon Tyne [29]. Stockton-on-Tees has a population of 191,600 residents [30]. The population is overwhelmingly white (93.4%) [30], and there are high levels of social and economic inequality. Full details of the sampling technique are contained in Mattheys et al., (2016) [31] and Bhandari et al., (2017) [32]. The sample size was estimated based on a conservative power calculation to detect a 5% difference in health outcomes between the least- and most-deprived areas in Stockton-on-Tees, as measured by validated indicators (EQ5D, SF8 PCS and SF8 MCS) [32]. Allowing for a 20% attrition rate between baseline and first follow-up and an additional 5% attrition at all other follow-ups, an estimated sample size of 800 (400 in each group) was estimated to be required to detect the difference in health between areas. Given the attrition expected, it was assumed that a sample of 800 at baseline would ensure that there would be sufficient participants available in the follow-up period to undertake statistical analysis [31,32]. In summary, the survey used a random baseline sample of adults aged over 18, split between participants from the 20 most- and 20 least-deprived LSOAs of Stockton-on-Tees (derived using 2010 Index of Multiple Deprivation (IMD) scores for England) (Figure 1). Multistage sampling was used whereby the Stockton LSOAs were first grouped into the 20 most- and 20 least-deprived (IMD range 1.54–74.5). Within each group, a random sample of households (addresses) were selected and a single participant per household was determined using a household selection grid to ensure even distribution of age and gender [33]. To meet the targeted number of 800 participants, 200 target households were randomly sampled in each of the 40 LSOAs assuming a 10% enrolment rate (because the survey used a postal recruitment approach and so response was expected to be lower than for other recruitment methods) [34,35]. A total of 8000 households (4000 each from the most- and least-deprived LSOAs) were sent study invitation letters to obtain consent to participate in the study based on an opt-in consenting approach. Participants were then surveyed four times over 18 months between April 2014 (baseline, wave 1 and face-to-face) and October 2015 (wave 4 and telephone). Details are presented in Figure 2. This was a baseline response of 10% or 36% of contacted households [31]. Attrition reduced the final wave 4 sample size to 310, a 37% follow-up rate but it fell within our conservative power calculation [31].The questionnaires included questions on mental and physical health, demographics, health behaviours and the social determinants of health. The main outcomes in this analysis are behavioural factors: smoking (yes/no), alcohol consumption (yes/no), fruit and vegetable consumption (five portions per day, yes/no), and physical activity (often, i.e., a couple of times/week, more frequent/not often, or once a week or less). The outcome variables were calculated based on participants’ responses: if they smoke; drink alcohol; how many portions of fruit and vegetables they eat on a usual day; and how often they have physical activity/exercise (everyday, most days, a couple of times a week, once a week, less than once a week, never). Supplementary Table S2 includes the questions asked to collect data on outcome variables.The primary explanatory variable was area-level deprivation, i.e., whether the participants lived in the ‘least-deprived’ (the 20 LSOAs with lowest IMD scores) or the ‘most-deprived’ (the 20 LSOAs with highest IMD scores) areas within Stockton-on-Tees. Age and gender were used as controlled variables in the models. Compositional factors (material) included educational status (highest level of educational qualification achieved), housing tenure (owned outright, rented, mortgaged, rent-free or others), household receipt of benefits (derived from responses on receiving specific listed benefit schemes), receipt of housing benefit (yes, no), employment (currently employed, unemployed), workless household (no adult household members currently in work), and household annual income (recorded via a range of income bands). Psychosocial factors included participants’ perceptions of neighbourhood safety (if they feel safe to walk after dark), lack of companionship (whether lacked companionship: hardly ever, sometimes, often), feeling left out (whether felt left out: hardly ever, sometimes, often), feeling isolated (whether felt isolated: hardly ever, sometimes, often), frequency of social meetings (never, less than once a month, once a month, several times a month, once a week, several times a week, every day), and happiness scale score (range 0–10; 0 = not happy at all, 10 = very happy). Contextual factors included whether the respondents reported that the accommodation had problems with damp (e.g., leaking roof, damp wall, rotten wall or floorboard), was dark (at least one room too dark or do not have enough light), was not warm enough (in winter months), or had problems of neighbourhood (within 15-min walk) noise (yes, no), pollution (yes, no) and crime (yes, no). The political economy of place (austerity) was assessed using time. For easier interpretation, categorical variables, such as lack of companionship, feeling left out, or feeling isolated were recoded into binary variables (often = 1, others = 0). Table 1 shows the response categories of the predictor variables.After data cleaning 736 baseline participants were included within the full analyses: 357 participants from the most-deprived LSOAs and 379 from the least-deprived. At wave 4 follow-up there were 310 participants: 176 from most- and 134 from least-deprived LSOAs. Descriptive analysis of the baseline data was conducted using summary statistics (frequencies, percentages, mean, standard deviation). Generalised estimating equations (GEE) accounting for clustering at LSOA level was applied to the baseline data to quantify the gap in health behaviours and to assess the associations between behaviours and the explanatory factors. To examine inequality in health behaviours, base models were fitted for the behavioural outcomes with only the deprivation indicator as the predictor variable. Thereafter, models including age and sex were fitted to test for associations between behavioural outcomes and deprivation by including explanatory compositional and contextual covariates to obtain a parsimonious model. Since health behaviours tend to vary by age and sex, these variables were included in the models so that the results could be adjusted for presence of these factors. Finally, age- and sex-adjusted models were also used to assess changes in the inequality gap over time (austerity). All statistical analyses were completed on SAS 9.4 version (SAS Institute Inc., Cary, NC, USA).Table 1 provides a descriptive analysis of the baseline sample. It incorporates demographics, behavioural outcomes, and compositional (material and psychosocial) and contextual variables, stratified by deprivation level. For both most-deprived and least-deprived areas, more women than men participated in the study. At baseline, 27.5% of the participants in the most-deprived areas were aged 65 years or over, whilst 32.8% aged 65 years or over were in the least-deprived areas. In addition, a higher percentage of participants living in the most-deprived areas were below 25 years of age (10% vs. 3%). A lower percentage of participants in the most-deprived areas had a degree or higher level of education (5% vs. 27%) than in the least-deprived areas. At least twice as many participants in the most-deprived areas reported feeling isolated, left-out or lacking companionship. However, average happiness score was quite similar in both areas.In the most-deprived areas, 37.0% of participants smoked, compared to 10.3% in least-deprived areas; about 59.1% of those in the most-deprived areas consumed alcohol, while 78.9% did so in the least-deprived areas. Only 21.6% of participants consumed five or more portions of fruit and vegetables a day in the most-deprived areas, compared to 41.7% in the least-deprived areas; about 67.8% engaged in regular physical exercise in the most-deprived areas, while 60.2% did so in the least-deprived areas. Most health behaviours varied significantly by deprivation level (Table 2). Smoking behaviour was significantly lower in least-deprived areas (OR 0.21, CI 0.14 to 0.30), whereas, alcohol use (OR 2.75, CI 1.98 to 3.82), and eating five portions of fruit and vegetables a day (OR 2.55, CI 1.80 to 3.62) were significantly higher in the least-deprived areas. However, frequent exercise behavior did not vary significantly by deprivation levels. Table 3 contains the final model results about associations of health behaviours with material, psychosocial and environmental factors and Supplementary Table S1 shows the initial bivariate associations tables for the health behaviours. Both tables utilized baseline data only.The bivariate analysis results found that in terms of material factors, those who were employed, educated and lived in an owned or mortgaged house were significantly less likely to smoke; smoking was significantly higher among those who received housing benefit or any other benefits, or belonged to a workless household. However, alcohol use was significantly more likely among those who were employed, educated (degree or higher) and lived in an owned or mortgaged house. Unlike smoking, alcohol use was significantly lower among those who were in receipt of benefits or belonged to a workless household. Intake of five fruits or vegetables a day and frequent exercise were both significantly higher among those who were educated (degree or higher) or owned their accommodation. However, benefit recipients were significantly less likely to have frequent exercise.In terms of environmental factors, those who lived in areas with higher levels of noise and crime were significantly more likely to smoke and those living in houses with a dark room were less likely to drink alcohol or consume five portions of fruit and vegetables a day. There were no significant associations between environmental factors and frequent exercise behavior.In terms of psychosocial factors, those who often lacked companionship, often felt isolated, or were less happy were significantly more likely to smoke. On the contrary, often feeling isolated or left out were negatively associated with alcohol use. Those who socialised more frequently or were happier were also significantly more likely to have five fruit or vegetables a day. Similar associations were seen for frequent exercise.Overall, more of the material factors than the physical environmental or psychosocial factors were associated with health behaviours.The final model explaining smoking behavior (Table 3) found that smoking was significantly less in the least-deprived areas, and among those who lived in an owned or mortgaged house compared to those who rented or lived rent-free. However, those who lived in a workless household were significantly more likely to smoke, so were less happy people. On the other hand, the odds of alcohol drinking were significantly higher in the least-deprived areas, among those employed or educated at a degree level (twice as likely for both), but was lower among women. The likelihood of having five portions of fruit and vegetables a day was significantly higher (OR = 2.01; 95% CI = 1.38, 2.95) among those living in the least-deprived areas, having degree level education, or among happier people. Women were less likely to exercise frequently (Table 3), and so were those who are employed, received a benefit or belonged to a workless household. The likelihood of doing frequent exercise was significantly higher (OR = 2.65; 95% CI = 1.40, 5.00) among those with a degree level education compared with those having no formal education. Those living in noisy areas had significantly higher likelihood of frequent exercise than those lived in areas that were not noisy. Happiness score was positively associated with frequent exercise.Figure 3 shows the percentage of people practising different health behaviours in the least-deprived and most-deprived areas of Stockton-on-Tees, over the four study waves (18 months). Throughout the period, except for smoking, all health behaviours were more prevalent in the least-deprived areas. Compared to wave 1 (April 2014), the prevalence of smoking was somewhat lower in wave 4 (October 2015) in both types of areas of Stockton on Tees. However, the difference was much smaller (5%) in the least-deprived areas than in the most-deprived areas (13%). The change in prevalence of smoking (wave 4–wave 1) did not vary much between areas (8% and 9% lower prevalence in the least- and most-deprived areas, respectively). Throughout the period, eating five portions of fruit and vegetables a day remained almost two times higher in the least-deprived areas than in the most-deprived areas. The change in prevalence (wave 4–wave 1) was 4% vs. 1% in least-deprived and most-deprived areas, respectively. Contrary to other behaviours, higher percentages of respondents in the least-deprived (4% higher) and the most-deprived (7% higher) areas were having frequent exercise in wave 4.Figure 4 shows the trends in the inequalities in health behaviours between most- and least-deprived areas of Stockton-on-Tees, over the four study waves (18 months). A considerable inequalities gap (% in least deprived–% in most deprived) persisted across time for each of the health behaviors (Figure 4). Over the study period, the inequality remained much larger for smoking (range: −47% to −19%), alcohol drinking (range: 20–23%), and eating five portions of fruit and vegetables a day (range: 16–20%) than for frequent exercise (6–9%). In terms of variation in inequality between waves, the inequality in smoking became much smaller (−19%) at wave 4 than at wave 1 (−47%). Overall, the variability in inequality between waves was 5%, 4% and 3%, for drinking alcohol, eating five portions of fruit and vegetables a day, and frequent exercise, respectively. It is to be noted that while large inequalities existed in terms of alcohol consumption and fruit and vegetable consumption, between-wave variation for these inequalities was relatively small (Figure 4).Table 4 shows the results for the GEE model that statistically examined the difference in health behaviours between the least- and the most-deprived areas, and over time (between wave 1 and wave 4). Furthermore, the addition of an interaction term (deprivation status × time) in the model provided the information as to whether the inequality gap had any statistically significant change over the 18 month period. As was observed from the results in Table 2, the model results further confirmed the inequality pattern (less smoking, higher alcohol drinking, and higher consumption of five portions of fruit and vegetables a day in the least-deprived areas than in the most-deprived areas). There was no significant difference in exercise behavior between areas (95% CI: 0.84, 2.87). Although the odds of smoking were significantly lower in wave 4 than wave 1, the odds of smoking remained 80% lower among participants in the least-deprived areas than those from most-deprived areas (OR: 0.20; 95% CI: 0.13, 0.32). However, as the interaction term (deprivation × time) was non-significant, it shows that inequality in smoking behaviour remained stable over the study period. Similarly, the odds of alcohol drinking were nearly three times higher in least-deprived areas (95% CI: 1.84, 3.58), and there was no statistically significant interaction between time and area deprivation. Similar to alcohol drinking, the odds of eating fruit and vegetables (five a day) was about three times higher among those from the least-deprived areas than those in the most-deprived areas. However, there was no statistically significant change in inequality in eating fruits and vegetable over the period (wave 4 95% CI: 0.59, 1.47) than what was observed in wave 1. Unlike other health behaviours, exercise behavior did not vary between the least- and the most-deprived areas and this pattern remained stable over the study period (Table 4).This study found that there are inequalities in health behaviours in Stockton-on-Tees. Smoking status, alcohol use and fruit and vegetable consumption all varied significantly by deprivation level. Smoking was more prevalent in the most-deprived areas, while alcohol use, and consuming five portions of fruit and vegetables a day were more prevalent in the least-deprived areas. Our findings about inequality in smoking echo the study by Duncan and colleagues (1999), who used the British Health and Lifestyle survey data to examine inequality in smoking by area and individual factors [36]. They concluded that area-level deprivation has an independent effect on smoking. Similarly, another study from the USA also observed that smoking was highly prevalent among men in more disadvantaged neighbourhoods [4]. A qualitative study to understand area effects on health behaviours in Glasgow, UK also observed associations between smoking and community disadvantage. Living in a disadvantaged area can be more stressful and smoking could be seen as a coping mechanism. Smoking can also represent a cultural norm within low income communities [3]. Unhealthy behaviours tend to cluster in disadvantaged areas—with an increase in area-level disadvantage there is also a tendency for higher rates of multiple unhealthy behaviours [37]. Our findings–that the proportion of participants smoking at wave 4 compared to that of wave 1 dropped largely in the most-deprived areas—could be a result of the disproportionate financial impact of austerity and previous research has found that in times of economic crisis, smoking behavior drops amongst those on low incomes [38,39,40].In terms of explaining these inequalities, material (compositional) factors were the most- and environmental (contextual) factors the least-important mediators of inequalities in health behaviours, particularly for smoking status and alcohol consumption [3,37,41]. It is also not surprising that large and persistent inequalities existed among our study participants for fruit and vegetable consumption (five portions a day). One study examining pathways of inequality in fruit and vegetable intake in Europe suggested that it can be constrained by financial capability, thus reiterating the association between material factors and health behaviours [42]. Low availability of fruit and vegetables was also found in a USA study after economic change in a neighbourhood following the Great Recession [43]. Social researchers often use the political ecology framework to explain neighbourhood effects on health, which explains how poor political decisions together with ecological factors are associated with persistent structural inequalities and poor health [41]. Material factors were associated with all health behaviours, whereas environmental factors were only associated with frequent exercise. Inequalities in all health behaviours were relatively stable throughout the study period against a backdrop of austerity. Recent research, though, has highlighted that regional inequalities in life expectancy between the North East of England and the South-East, more commonly known as the north–south health divide, is worsening [44]. In this context, our findings of health inequalities in Stockton-on-Tees, an area with long-term exposure to health inequalities calls for priority attention [44,45]. There is extensive international research into the association between deprivation and patterns of health behaviours. Previous research has found that smoking is higher in places with higher levels of deprivation and alcohol consumption can be a means of socialisation and a choice for those able to maintain responsibilities [46] and therefore higher in least-deprived areas, however, severe levels of alcohol use—and alcohol-related harm—is more prevalent in more-deprived areas. There is an inverse association between deprivation and physical activity and diets contain less fruit and vegetables in more deprived areas [2,5,6]. This body of work has also highlighted the important role that ‘place’ has in shaping these socio-spatial inequalities in health behaviours—such as higher access to unhealthy goods and commodities in more deprived communities [7,9,11,12] as well as a lack of health-promoting services and infrastructure such as green spaces for physical activity [10]. Some research has suggested that contemporary austerity has exacerbated health inequalities in England and internationally [23,47], and that people living in the most-deprived areas of England have seen the largest increases in poor mental health [7] and self-harm [8,9,24,25]. Using a case-study approach, our study has added to this important international body of literature by providing a detailed examination of: (1) geographical inequalities in a range of health behaviours; (2) the relational nature of contextual and compositional factors; and (3) the influence of political economy. It found high inequalities in health behaviours; that material factors were the most- and environmental factors the least-important influences; and that despite austerity, inequalities in all health behaviours were relatively stable throughout the study period. The latter finding is in contrast to previous research into the health impacts of austerity (although in keeping with our own research into physical and mental health) [15,22] and might reflect issues with our sample (older people were largely protected from austerity), the follow-up length (an 18-month follow-up might not have been long enough to detect changes) and the timing of our study (the baseline survey in 2014 was in a period after the economic recession and after some austerity measures had already been implemented) [22]. The study is subject to a number of important limitations. The baseline sample size was moderate (although within power calculations) and the response rate was low with only 36% of contacted households (and only 10% of all of our 8000 sampling frame) participating in the survey. The survey also experienced high attrition with only 37% in the final wave [22], some of which could be associated with use of a telephone survey in wave 4. This may undermine the representativeness of the cohort sample and indeed, older people and women were over-represented compared to the general population. Whilst models were adjusted, these factors may still effect the generalisability of the findings. The survey also relied on self-reported health measures, which may have limited precision and reliability. Finally, this study relates only to Stockton-on-Tees. This local authority has the highest gap in life expectancy between people in the most- and least-deprived areas in the whole of England and the results may not be generalisable to other places especially outside of the North East region of England [22].This study used a household survey to examine inequalities in health behaviours during a time of austerity. It found clear and stable associations between deprivation and health behaviours. The exploration of risk factors suggests that material compositional factors are the most common determinants of geographical inequalities in health behaviours and that tackling these could be an important approach to health promotion. The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111018/s1, Table S1: Associations between health behaviours (smoking, alcohol, 5 portions of fruit and vegetables/day, frequent exercise) and material socio-economic, material physical environmental, and psychosocial explanatory variables; Table S2: Description of outcome variables.C.B. conceptualized the study; N.A. and A.K. developed the study methods; J.W. designed and oversaw data collection; N.A. prepared the database and conducted the analysis; R.S.F. conducted initial exploratory analysis; N.A., C.B., M.P. and R.S.F. drafted the paper with input from all authors. All authors reviewed and contributed to revisions of the paper. All authors have read and agreed to the published version of the manuscript.This study was funded by a Leverhulme Trust Research Leadership Award (reference RL-2012-006). C.B. and N.A. are members of Fuse: UKCRC Centre for Translational Research in Public Health. Funding for Fuse comes from the British Heart Foundation, Cancer Research UK, Economic and Social Research Council, Medical Research Council, and the National Institute for Health Research, under the auspices of the UK Clinical Research Collaboration, and is gratefully acknowledged (MR/K02325X/1). The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Durham University Department of Geography (16 April 2012). Informed consent (written/verbal), as appropriate during in-person and telephone interviews, was obtained prior to data collection, allowing publication of anonymised results.Data available on request.The authors declare no conflict of interest regarding the publication of this article. 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.Maps of Stockton-on-Tees showing most- and least-deprived neighbourhoods.Sampling strategy for the survey [28].Percentage of respondents smoking, drinking alcohol, eating fruit and vegetables (five portions/day), and exercising frequently in the most- and least-deprived areas of Stockton-on-Tees, over four waves (18 months).Trends in the inequalities in health behaviours (difference: % least deprived–% most deprived) between most- and least-deprived areas of Stockton-on-Tees, over four waves (18 months).Baseline descriptive analysis of demographic, material, psychosocial, contextual and behavioural variables in the analysis cohort, stratified by level of deprivation.Generalised estimating equation analyses adjusted for age and gender, showing odds ratios, 95% confidence intervals, and p values for the association between deprivation and heath behaviours.Factors associated with smoking, alcohol use, consumption of five fruits and vegetables/day, frequent exercise, adjusted for clusters, age, gender and deprivation.Analysis of behaviour outcomes by time and deprivation.† Time and Deprivation interaction term.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11019.txt

@@ -0,0 +1 @@
+The COVID-19 outbreak is a global health concern. Understanding the transmission modes of the SARS-CoV-2 virus is key to limit the spread of the pandemic. A lack of knowledge about the possibility of SARS-CoV-2 transmission and infection through contaminated surfaces is noticeable and recent studies have stated conflicting findings. This scoping review aims to understand the risks of contaminations via fomites better. Relevant publications were selected through Google Scholar, Web of Science, PubMed, Embase, Medline, and Cochrane Library, with related keywords. PRISMA-ScR guidelines were followed. Out of the 565 articles found, exclusion criteria were applied, duplicates removed, and a total of 25 articles were finally included in the study. The included documents were assessed by the contamination risk: “low” (37.5%), “high” (16.7%), “plausible” (8.3%), “unlikely” (8.3%) risk, and “insufficient evidence” (29.2%). Research in hospital settings was found as the main setting in the reviewed papers, which precisely indicated the risk of contaminated surfaces. This scoping review underscores the risk of SARS-CoV-2 infection via contaminated surfaces assessed as low in the majority of the reviewed articles. Further evaluation of the risk of the virus transmission by fomites and providing adequate information on its infectivity via contaminated surfaces in real-life conditions is essential.The coronavirus disease 2019 (COVID-19) outbreak is a global health concern [1], with an undoubtedly serious impact affecting people’s lives in various aspects, including healthcare, economic, and social factors [2,3]. The novel coronavirus was firstly detected in Wuhan in December 2019 [4]. On 12 March 2020, due to the high infectivity of SARS-CoV-2 [5] and numerous reported cases and deaths, the World Health Organization (WHO) declared the COVID-19 as a pandemic [6]. The virus SARS-CoV-2 belongs to the Coronaviridae family, some of whose viruses are associated with respiratory tract infections [7].Understanding of the transmission modes of SARS-CoV-2 is a crucial matter. From a general perspective, the transmission of the respiratory virus has a few known routes, which are mainly (a) direct contact (person-to-person), (b) indirect contact by contaminated surfaces (fomites), and (c) airborne transmission by droplets or aerosols [8]. Based on the evidence to date, the transmission of SARS-CoV-2 is stated to be mostly airborne via droplets or aerosols through close contact with infected individuals [9,10]. However, other previous studies have suggested the possibility of hand to face transmission of pathogens by touching contaminated objects and surfaces, otherwise known as fomite transmission [11,12,13,14,15,16]. Moreover, face-touching behavior has been understood as a self-inoculation for microbial transmission [11,12,17,18]. In addition, some other studies underscore that contaminated surfaces could play a key role in the transmission of pathogens [13,14,15,16].However, in the specific case of the SARS-CoV-2 virus, recent studies on the risk of fomites transmission have stated various findings. Most authors have demonstrated a low risk of contamination [11,19,20,21], while fewer authors suggested a higher risk [22,23].For instance, as Kraay et al. stated, “while direct transmission is important, our model suggests that fomites can also transmit, which is important for exposures that are not in-person. Therefore, fomites transmission may be an important source of risk [22]”. Another study emphasized a key role of fomites transmission, as “the results showed that moderate protein concentration in droplets markedly increased the infectivity of SARS-CoV-2, suggesting that a protein-rich medium like airway secretions could protect the virus when it is expelled and may enhance its persistence and transmission by contaminated fomites. Accordingly, it is plausible that fomites infected with SARS-CoV-2 play a key role in the indirect transmission of COVID-19 [24].”However, Goldman believes that “the chance of transmission through inanimate surfaces is very small and only in instances where an infected person coughs or sneezes on the surface, and someone else touches that surface soon after they cough or sneeze (within 1–2 h) [19]. Another study has found that “despite prolonged viability of SARS-CoV-2 under laboratory-controlled conditions, uncultivable viral contamination of inanimate surfaces might suggest low feasibility for indirect fomite transmission [25].” In addition, findings from Moore et al. demonstrated that “the concentration of viral RNA was low and ranged from <10 to 460 genomic copies/m3 air. Infectious virus was not recovered from any of the PCR-positive samples analyzed.” [26].Therefore, in this present scoping review, we aim to provide a comprehensive overview of the literature about the risk of SARS-CoV-2 transmission via fomites in order to participate in a better understanding of relevant approaches to mitigate the propagation of the COVID-19 pandemic.A scoping review of peer-reviewed and grey literature was conducted to identify the risks of infection with SARS-CoV-2 by contaminated surfaces.A general literature search was first done on the SARS-CoV-2 virus to have a preliminary overview of the concepts and previous studies on this specific topic. Several studies on the SARS-CoV-2 virus were related to the risk of infection, such as contact and droplet transmission, airborne transmission, fomite transmission, and the possibility of transmission by animals. Importantly, this scoping review solely focused on resources related to the risks of infection with the SARS-CoV-2 due to contaminated surfaces. This review follows the framework of the Preferred Reporting Items for Systematic Review and Meta-analysis for Scoping Reviews (PRISMA-ScR) guideline.In order to broadly capture existing research on the defined topic, relevant publications were gathered using the search engines of Google Scholar, Web of Science, PubMed, Embase, Medline, and Cochrane Library.Firstly, the search for the relevant literature was conducted using the following keywords in the titles of the articles: (SARS-CoV-2) OR (Novel Coronavirus) OR (COVID-19) AND (Fomites) OR (surface) OR (Inanimate surfaces) OR (Contaminated surfaces) OR (Environmental contamination) AND (Transmission) OR (Pathogen transmission) OR (Disease Transmission) AND (Viability) OR (Stability) OR (Survival) OR (Persistence).The search was then broadened by including the ‘Abstracts’ in the search field. The methodology process used is summarized in Figure 1 (PRISMA flowchart).Additionally, the search was also conducted by Mesh terms in PubMed as follows: (SARS-CoV-2) AND (Fomites) AND (Disease Transmission, Infectious) AND (Microbial viability) OR (Survival). The World Health Organization (WHO) and Centers for Disease Control and Prevention (CDC) websites were analyzed for relevant publications, which were considered as additional publications from other sources.In order to find potential additional relevant sources, identification and screening of further articles from references were undertaken. After completing the search with all relevant publications, duplicates were omitted.The inclusion criteria were: (i) English-language publications related to the potential risks of infection by contaminated surfaces with the SARS-CoV-2 virus; (ii) all types of scientific publications such as editorials, viewpoints, articles, guidelines, etc., were included; (iii) any relevant settings such as fomites in hospitals, public places or in-house, for instance.Exclusion criteria included publications related to other findings on COVID-19 disease and its clinical aspects or other routes of transmission of the virus, for instance, as examples. In addition, publications that did not imply a correlation between variables, such as the viral stability or virus load with the risk of infection, for example, were excluded.The relevant dataset was extracted and imported into an Excel extraction table by the main researcher (M.M.) This table includes the title, author(s), date of publication, study location focus, type of publication, the main topic, key findings, contamination risk assessment, and limits. The other co-authors screened publications to ensure their relevance and eligibility. The selected articles were imported by EndNote X7. The process of selection is reported in Figure 1.The search performed in the different databases underscores various interesting findings related to the risk of infection with SARS-CoV-2 from contaminated surfaces. The key results can be summarized as follow:From the 25 documents selected for this review, 24 were retrieved through the database searches and 1 from the website of the CDC.As mentioned in Figure 1, of a total of 565 publications found in the databases and SARS-CoV-2 source-related websites, 116 publications remained after removing the duplicates and then were screened by titles and abstracts. At this stage, 59 of the reviewed articles were classified as out of the scope of our research question and, therefore, were excluded. The full text of the remaining 57 publications was assessed, and 24 articles finally met the inclusion criteria.Importantly, some excluded articles did not exactly state the risk of infection by fomites but had investigated some features of the virus that might lead to infectivity, but the relationship between those features and infectivity was not precisely discussed.In addition, several excluded articles discussed the stability of the virus in different environmental conditions, but a lack of evidence on the relationship between the stability features and the risk of infection by contaminated surfaces was observed. Contrastingly, 17 articles from this category discussed a potential correlation and were included in our scoping review. Five out of the 24 selected articles were surveyed in Europe, namely France, Switzerland, Germany, Italy, and England. Two of them were from the USA and one article, respectively, from Israel, China, and Singapore.We could not identify a specific study setting in the remaining articles classified as Not Applicable, most of which were review articles and editorial letters. Articles since the beginning of the COVID-19 pandemic in early 2020 were assessed, and the last search was performed on 10 February 2021.Amongst the included articles, ten out of 24 were review articles, which stated a clear conclusion about the risk of infection by fomites contaminated with SARS-CoV-2. Amongst the articles which have precisely indicated the risk of contaminated surfaces with SARS-CoV-2, five of them performed the research in hospital settings, three of them in community settings including schools, offices and high-touch surfaces in a city, one in a laboratory environment, and finally one in a quarantine household.Appendix Table A1, Table A2 and Table A3 summarize the main features of the included publications.Based on our review, the main topics of the relevant publications found were: (I) persistence of the SARS-CoV-2 on fomites or inanimate surfaces, (II) risk of transmission of SARS-CoV-2 from fomites, (III) infectivity of SARS-CoV-2 on surfaces, (IV) modes of transmission, (V) nosocomial transmission of SARS-CoV-2 from surface environments. Moreover, the articles were assessed by the contamination risk and were classified by “low”, “high”, “plausible”, and “unlikely” risk. Furthermore, a category was defined as “insufficient evidence”. Nine articles out of 24 (37.5%) assessed a low probability of transmission, and four articles out of 24 (16.7%) stated a high transmission probability.Importantly, seven out of the 24 articles (29.2%) indicated not having enough evidence to determine the risk of SARS-CoV-2 infection by contaminated surfaces. These findings are summarized in Figure 2, and a detailed review of the risk assessment can be found in Table 1.This scoping review was performed to evaluate the available literature related to the risks of infection transmission via contaminated surfaces with SARS-CoV-2.Our research highlights a noticeable variability in the findings of articles assessing the risk of transmission via fomites classified as follows: low or high possibility, unlikeliness, plausibility, or lack of adequate evidence to identify the risks of infection transmitted by fomites.While previous studies on different microorganisms have demonstrated the existing risk of transmission by contaminated surfaces [13,14,15,16,17,18], surprisingly, only six out of 24 of the articles had indicated the “plausibility” or high risk of infection via fomites in the case of SARS-CoV-2 contamination.We found that 29.2% of the reviewed articles stated an absence of enough evidence, and most of the articles, i.e., 45.8%, concluded a low probability or unlikeliness. Interestingly, those articles were mostly focused on the following conclusions: (i) the existence of the virus on surfaces does not prevail the risks of infection by the virus, and (ii) there is a lack of information about the infectious dose of SARS-CoV-2 on the surfaces to be transmitted in order to cause infection [32,35,38,39]. For instance, “The infectious dose of SARS-CoV-2, namely the average number of viral particles required to establish an infection for COVID-19 is unknown” as stated by Xue et al. 2020 [36]. Another rationale observed to support the low probability of transmission of the SARS-CoV-2 via contaminated surfaces is the demonstration that viruses cannot reproduce outside the host, and the observation of the rare occurrence of touching surfaces contaminated with viral loads high enough to be infective and the subsequently touch face membranes such as eyes, or nose [11,39].In addition, as mentioned by the World Health Organization (WHO), “it is difficult to disentangle the relative contributions of inhaled droplets and contaminated surfaces because people who have come into contact with potentially infectious surfaces have generally also been in close contact with infected individuals”.Moreover, there are differences between real-life and laboratory conditions, which lead to lower risks in the real living environment, especially when hygiene protocols and cleaning procedures are followed [11,29,33].From another standpoint, the long persistence of the virus on surfaces might cause a high possibility of infection via contaminated surfaces [24,28]. In one reviewed study focused on hospital settings, with the assumption made that the concentration of the virus might be more important, fomites were identified as potential sources of the virus spread [25]. On the contrary, three other studies performed in the same settings concluded that the risk of transmission and infection is not high [21,26,31].Interestingly although this scoping review aimed to identify specifically the risk of infection via fomites, we could observe debates about the duration of the virus presence on different surfaces and the environmental effects on it. For instance, “in laboratory-controlled conditions and at the ambient temperature, SARS-CoV-2 lost its infectivity completely by day 4” [31].We also consider it crucial to mention that studies demonstrated that higher temperature, sunlight, and UV radiation highly lead to the inactivation of SARS-CoV-2 [23,37]. Those are important factors to be understood in the real-life context of the SARS-CoV-2 life cycle. In addition, studies demonstrated that the RNA of common viruses, such as SARS-CoV-2, Influenza, and MERS-CoV, can persist on surfaces for days after they have lost their infectivity [23], therefore, detecting viral RNA on surfaces cannot solely provide information about the infectivity of the virus [35]. Thus, such information has to be taken with the necessary scrutiny to avoid potential incomplete conclusions about the probability of getting infected via fomites in real-life conditions.Given the date of the last search performed, articles published after 10 February 2021, were not integrated in our review. Some recent publications on that topic were found, confirming the relevance of this topic.Conclusively, the main outcome of this scoping review is that the risk of SARS-CoV-2 infection via contaminated surfaces was assessed as low in the majority of the reviewed articles. Further evaluation of the risk of the virus transmission by fomites and adequate information on its infectivity via contaminated surfaces in real-life conditions are essential. Those investigations would participate in setting more efficient guidelines to limit the spread of the SARS-CoV-2. Until more evidence on the risk of the virus transmission by fomites in real-life situations can be gathered, it remains important to follow disinfection guidelines and, most importantly, respect physical distancing and the use of masks to limit the propagation of the COVID-19 pandemic. Finally, the authors acknowledge that scientific data and related issues regarding the SARS-CoV-2 virus and potential variants evolve rapidly, but at the time of the research, no sufficient data was available mentioning possible variants and fomites transmission. Therefore, it appears key to take into consideration those elements in future research.M.M. contributed to the implementation of the research, analysis of the results, and writing of the manuscript. A.B.-A. contributed to the validation, review of the overall manuscript. A.L. reviewed the content of the manuscript. A.F. designed, directed, and supervised the project. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors stated that they had no interests, which might be perceived as posing a conflict of interest, or bias.Detailed information of review articles.Key findings of review articles.Recommendations and limitations of review articles.PRISMA flowchart.Contamination risk assessment of SARS-CoV-2 with contaminated surfaces (%).Comments from the reviewed articles regarding the level of risk of SARS-CoV-2 infection through contact with contaminated surfaces.“While direct transmission is important, our model suggests fomites can also transmit, which is important for exposures that are not in-person. Therefore, fomite transmission may be an important source of risk.” [22]“After reviewing ‘Similarities between SARS-CoV-2 and Other Coronaviruses’, ‘Effect of Media, Temperature, Relative Humidity, UV Irradiation, and Material-Type on SARS-CoV-2 Persistence’, the researchers concluded “the virus will persist on high-touch surfaces long enough to spread to new individuals.” [23]“Interpretation of findings for SARS-CoV-2 strongly suggest that the environment can serve as a medium of transmission of SARS-CoV-2, through touch contamination and subsequent self-inoculation of mucous membranes by a non-infected individual coming into contact with a contaminated environmental surface or fomite.” [27]“Our data showed that SARS-CoV-2 infectivity was remarkably preserved in the presence of proteins, regardless of the type of surface.”; “The results showed that moderate protein concentration in droplets markedly increased the infectivity of SARS-CoV-2, suggesting that a pro-tein-rich medium like airway secretions could protect the virus when it is expelled and may enhance its persistence and transmission by contaminated fomites. Accordingly, it is plausible that fomites infected with SARS-CoV-2 play a key role in the indirect transmission of COVID-19.” [24]“These findings suggest that the hospital environment could potentially be a source of virus spread, including among HCWs, patients, and visitors.” [25]“Our results indicate fomite transmission of SARS-CoV-2 is plausible since the virus can remain viable and infectious on surfaces up to days.” [28]“The work supports the current perception that contaminated surfaces are not a primary mode of transmission of SARS-CoV-2.”; “The risks posed by contact surfaces in 30 communities are low for community infection prevalence rates ranging from 0.2–5%.” [11]“The chance of transmission through inanimate surfaces is very small, and only in instances where an infected person coughs or sneezes on the surface, and someone else touches that surface soon after the cough or sneeze (within 1–2 h).” [19]“Findings suggest that environmental contamination leading to SARS-CoV-2 transmission is unlikely to occur in real-life conditions, provided that standard cleaning procedures and precautions are enforced.”; “transmission is unlikely to occur in real-life conditions, provided that standard cleaning procedures and precautions are enforced.” [29]“The risk of transmission via touching contaminated paper is low.” [20]“The results indicate that at that early time of SARS-CoV-2 outbreak research in Germany the contamination of the domestic environment is negligible during quarantine measured with the current state of the art methods. We could not detect any viral RNA in air samples and only 3.36% of all fomite samples.”; “This study supports the hypothesis that indirect environmental transmission may only play a minor role, which needs clarifications in further studies.” [21]“Toilet bowl and sink samples were positive, suggesting that viral shedding in stool could be a potential route of transmission.” [30]“The estimated risk of infection from touching a contaminated surface was low (less than five in 10,000) by quantitative microbial risk assessment, suggesting fomites play a minimal role in SARS-CoV-2 community transmission.”; “our results are consistent with fomite-mediated transmission of COVID-19 being possible but likely a secondary pathway.” [31]“Despite prolonged viability of SARS-CoV-2 under laboratory-controlled conditions, uncultivable viral contamination of inanimate surfaces might suggest low feasibility for indirect fomite transmission.”; “Aerosol or indirect transmission from inanimate surfaces around hospitalized or quarantined COVID-19 patients is not supported by the data presented in this study” [25].“The concentration of viral RNA was low and ranged from <10 to 460 genomic copies/m3 air. Infectious virus was not recovered from any of the PCR-positive samples analyzed.” [26]“Data on the transmissibility of coronaviruses from contaminated surfaces to hands were not found”; “The viral load of coronaviruses on inanimate surfaces is not known.” [32]“Our data suggest that although environmental contamination may occur in real-life conditions, it might be less extensive than hitherto recognized. Moreover, the inability of the SARS-CoV-2 RNA collected from the CPAP helmet to infect susceptible cell monolayers suggests that recent contamination of plastic surfaces, which apparently maintain SARS-CoV-2 infectivity for several hours, is unlikely to contribute to nosocomial spread.” [33]“Fomite transmission, i.e., viral dissemination via a material, including a door handgrip, door-bell, or inhalator, also has a critical contribution to the virus spread.”; “Survival duration of the COVID-19 causing virus on surfaces is not certainly known” [34]“Indirect transmission of COVID-19 has been assumed to be possible via fomites although direct evidence is currently not available.”; “The virus has been detected in hospital and household settings but detection of viral RNA on surfaces does not provide any information about viral infectivity or viability.” [35]“The infectious dose of SARS-CoV-2, namely the average number of viral particles required to establish an infection for COVID-19 is unknown.”; “It is currently unclear what role the surface chemistry plays in viral survival, infectivity, and denaturation, and the role of the local environment is unclear.” [36]“Fomite transmission would depend on the surface characteristics, which can affect virus survival and can help determine the extent of spread of the disease.” [37]“Overall, there was an inability to align SARS-CoV-2 contaminated surfaces with survivability data; and also a knowledge gap on fomite contribution to SARS-CoV-2 transmission.” [38]“Virus detection does not necessarily represent an infectious dose of SARS-CoV-2. Although SARS-CoV-2 may be transmitted via direct and indirect contact by touching contaminated sur-faces or medical equipment, followed by touching mouth, nose, or eyes, it remains unknown what portion of the transmission is attributable to a fomite.” [39]“The important ways of transmitting the virus are through Droplets, infected hands, and skin-to-skin contact, as well as inanimate surface contact.” [40]Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11020.txt

@@ -0,0 +1 @@
+These authors contributed equally to this work.These authors contributed equally to this work.Circadian pattern influence on the incidence of out-of-hospital cardiac arrest (OHCA) has been demonstrated. However, the effect of temporal difference on the clinical outcomes of OHCA remains inconclusive. Therefore, we conducted a retrospective study in an urban city of Taiwan between January 2018 and December 2020 in order to investigate the relationship between temporal differences and the return of spontaneous circulation (ROSC), sustained (≥24 h) ROSC, and survival to discharge in patients with OHCA. Of the 842 patients with OHCA, 371 occurred in the daytime, 250 in the evening, and 221 at night. During nighttime, there was a decreased incidence of OHCA, but the outcomes of OHCA were significant poor compared to the incidents during the daytime and evening. After multivariate adjustment for influencing factors, OHCAs occurring at night were independently associated with lower probabilities of achieving sustained ROSC (aOR = 0.489, 95% CI: 0.285–0.840, p = 0.009) and survival to discharge (aOR = 0.147, 95% CI: 0.03–0.714, p = 0.017). Subgroup analyses revealed significant temporal differences in male patients, older adult patients, those with longer response times (≥5 min), and witnessed OHCA. The effects of temporal difference on the outcome of OHCA may be a result of physiological factors, underlying etiology of arrest, resuscitative efforts in prehospital and in-hospital stages, or a combination of factors.Out-of-hospital cardiac arrest (OHCA) is a universal public health problem that claims nearly 3.7 million lives annually [1]. Incidence and outcomes of OHCAs vary greatly around the world [2,3]. In a recent report from the International Liaison Committee on Resuscitation (ILCOR), the annual incidence of emergency medical services (EMS)-treated OHCA in a global population ranged from 30.0 to 97.1 per 100,000 individuals. Despite the advances in the treatment of OHCA, the rate of survival to discharge or 30-day survival remains poor at 3.1–20.4% across different regions of the world [2]. Increased understanding of the variables that affect OHCA clinical outcomes is important for developing preventative strategies and optimizing care for OHCA.The Utstein Style guidelines consist of elements relevant to the clinical outcomes of OHCA. These include systemic factors (characteristics of the EMS system and the population served), dispatch factors (identified OHCA and dispatcher-assisted cardiopulmonary resuscitation (CPR)), patient factors (demographics, initial cardiac rhythm, witnessed status, arrest location, and bystander response), prehospital resuscitation process (response time, number of defibrillation shocks, quality of CPR, airway control, vascular access, delivery of epinephrine, etc.), and post-resuscitation processes (target temperature management, reperfusion therapy, etc.) [4]. It is important to understand how these core elements may be influenced by temporal differences.There is significant circadian variation in the incidence of OHCA. The occurrence of OHCA is highest in the morning and lowest at night [5,6]. However, the association between temporal differences and clinical outcomes in patients with OHCA remains controversial [5,7,8,9,10,11]. Two studies conducted in Japan suggested that OHCA occurring at night had significantly lower rates of 30-day survival and survival with favorable neurological outcomes [7,8]. A similar finding from a USA study showed that the rate of survival to discharge was significantly lower for OHCA occurring at night than during the daytime [5]. On the other hand, two Austrian studies and one US/Canadian study observed no significant difference between OHCA occurring at daytime and nighttime with regard to the sustained return of spontaneous circulation (ROSC), survival to discharge, 30-day survival with favorable neurological outcome, and 1-year survival [9,10,11].The cause of the relationship between temporal differences and varying clinical outcomes of OHCA remains unknown. In addition, there are no data concerning temporal differences and the clinical outcomes of OHCA in Taiwan. In this study, we aimed to evaluate the association between temporal differences and clinical outcomes of OHCA in an urban city in Taiwan; factors attributed to this observation were also evaluated in different OHCA statuses.A 3-year retrospective cohort study was conducted between 1 January 2018 and 31 December 2020 with the aim of evaluating temporal variations in clinical outcomes of OHCA in Chiayi City, Taiwan. Chiayi City extends across an area of 60.02 km2 and has 266,000 residents: It is the second-most densely populated city in Taiwan (4431.53 people per square kilometer) [12]. Residents aged over 65 years account for 16.2% of the Chiayi City population [12]. There are two primary hospitals, two secondary hospitals, and one tertiary referral hospital in the city. Patients with OHCA who require EMS are transferred to one of the five hospitals of the city. They are included in the OHCA registry of Chiayi City, a prospectively collected multi-center Utstein-style registry. Data were collected from the registry during the study period. Cardiac arrest, defined as the absence of signs of circulation, was verified by emergency medical technicians (EMTs) at the scene. Patients with traumatic cardiac arrest, including drowning and hanging, aged younger than 20 years, obvious death without being transferred to hospital, and those with valid do-not-resuscitate (DNR) orders were excluded from this study. The Institutional Review Board of the Ditmanson Medical Foundation Chia-Yi Christian Hospital approved this study (CYCH-IRB 2021057).The EMS systems of Chiayi City have been described previously [12]. The fire-based EMS system of Chiayi City consists of seven ambulance stations and one dispatch center operated 24/7 by experienced EMTs. A total of 243 EMTs serve in the fire bureau, and the bureau is composed of 2 EMT-1 (0.82%), 212 (87.24%) EMT-2, and 29 (11.93%) EMT Paramedics (EMTP). Upon receiving a call for medical assistance, the dispatcher will ask key questions to identify patients with OHCA. EMTs are then dispatched from the nearest EMT stations as concurrent dispatcher-assisted CPR (DACPR) is conducted. At the rescue scene, EMTs provide basic life support, including CPR, ventilation by bag-valve-mask, insertion of a laryngeal mask airway, and automated external defibrillators (AED). If EMTPs are dispatched, advanced life support, including endotracheal intubation and epinephrine injection, can be provided. Unless ROSC is achieved, continued CPR, ventilation, and use of AEDs during transportation are mandated for all OHCA cases. All EMTs must receive regular CPR training according to national guidelines based on the American Heart Association, ILCOR, and European Resuscitation Council Guidelines. Quality assessments and controls are conducted monthly in order to ensure resuscitation quality.Data were collected from the Chiayi City OHCA Registry. We obtained patient demographic information and recorded the time the EMS call was received (which was defined as the time of cardiac arrest); EMS response time (time from the call to the ambulance arriving at the scene); scene time (the time from the ambulance arriving at the scene to leaving the scene); transport time (the time from the ambulance leaving the scene to arriving at the hospital); total EMS time of identifying OHCA (defined as the sum of response time, scene time, and transport time); start time of DACPR or bystander CPR (BSCPR) from the time the call was received; number of dispatched EMTs; whether EMTPs were dispatched or not; location of cardiac arrest (public area, home, medical institution (local clinic and nursing home), ambulance transport, etc.); witness status of cardiac arrest; initial cardiac rhythm according to the AED record; prehospital management by EMTs; and the level of transferred hospital.In order to evaluate the association between circadian variation and clinical outcomes of OHCA, we divided the patients into three groups according to the time the EMS emergency call was received: daytime (08:01–16:00), evening (16:01–24:00), and night (00:01–08:00). The outcome measurement included achievement of ROSC at any time, sustained (≥24 h) ROSC, and survival to discharge (Figure 1).We hypothesized that 30% of OHCA occurred at nighttime, with an odds ratio of 0.6 in achieving ROSC in patients with an OHCA at nighttime compared to daytime, and set a ROSC rate of 30% as the baseline [5,7,13]. A two-tailed test size of 5% and a power of 80% was applied. We calculated that a total of 784 patients with OHCA would be required to reach statistical power. Therefore, we included three years of data to meet this requirement.We analyzed and compared data of patients with an OHCA between the three groups: EMS call received during daytime (08:01–16:00), evening (16:01–24:00), and at night (00:01–08:00). For categorical variables, the chi-squared test was used to evaluate differences between groups. For continuous variables, analysis of variance (presented as mean ± standard deviation) or the Kruskal–Wallis analysis (presented as medians (interquartile range)) was used appropriately after evaluating the data distribution. In order to evaluate the net effect of temporal differences (daytime vs. evening vs. nighttime) on patient outcomes, logistic regression with forward selection Wald test was performed with adjustments for the associated predictors related to outcomes of OHCA and variables with a p-value < 0.1, as derived from the univariate analysis. The adjusted variables included age [14,15], sex [14,16], EMS time interval [17,18], dispatch of EMTP [19,20], DACPR or BSCPR [21,22,23,24], witnessed cardiac arrest [12], shockable initial rhythm [12,21], location of cardiac arrest [25], use of public AED before EMT arrival [16,26], pre-hospital injection of epinephrine [27,28], prehospital use of mechanical CPR devices [12,29], and level of transferred hospital [30]. Moreover, in order to evaluate the temporal differences in different types of OHCA, including different sex and age groups (younger or older than 65 years), different EMS response times (less or longer than 5 min), witnessed status, and shockable or non-shockable cardiac rhythm, subgroup analyses were also performed using logistic regression with adjustments for influencing factors. Statistical significance was set at p < 0.05. Statistical analysis was performed using IBM SPSS version 22 (IBM Corp., Armonk, NY, USA) statistical software packages.During the study period, 1295 OHCA patients who had activated the EMS system were identified. Seven patients younger than 20 years old and 88 patients with traumatic cardiac arrest were excluded from the study, as well as 358 patients with DNRs or obvious death at the scene who were not transferred to the hospital. A total of 842 patients with OHCA were included in this study. There were 371 incidences of OHCA during the day (08:01–16:00), 250 OHCAs in the evening (16:01–24:00), and 221 OHCAs at night (00:01–08:00) (Figure 1). Significant differences in clinical outcomes were observed between the daytime, evening, and night temporal groups (Table 1). OHCAs that occurred at night had a significantly lower rate of achieving ROSC (19% vs. 28.57% vs. 26%, p = 0.033), sustained (≥24 h) ROSC (9.96% vs. 21.02% vs. 18.40%, p = 0.002), and survival to discharge (0.91% vs. 6.74% vs. 5.6%, p = 0.005) compared to OHCAs occurring in the daytime and evening, respectively (Table 1). For the OHCAs at night, EMS response time (p < 0.001), scene time (p = 0.001) and total EMS time (p < 0.001) were longer than those in OHCAs during daytime and evening. Additionally, the identification time of the OHCA by dispatcher was shorter (p = 0.001) and witnessed cardiac arrest was lower (p < 0.001) for the OHCAs at night. Moreover, the location of cardiac arrest was significantly different (p = 0.003). The majority of OHCAs occurred at home for all temporal groups; however, a higher percentage of daytime OHCAs occurred in public areas (10.24%) than nighttime, while nighttime OHCAs occurred more often in medical institutions (16.52%) (Table 1) than daytime OHCAs.Figure 2 shows how the frequency of OHCA incidents varies per 2-hourly intervals over 24 h as well as the percentage of achieving ROSC, sustained (≥24 h) ROSC, and survival to discharge. The rate of incidence of OHCA was higher during the daytime and peaked between 08:00 and 10:00. The rate of incidence decreased from daytime to evening, and a small peak was noted between 18:00 and 20:00. The lowest rate of incidence occurred at night between 00:00 and 04:00. The percentage per 2-hourly interval of achieving any ROSC, sustained (≥24 h) ROSC, and survival to discharge also showed temporal differences and were lower during the nighttime (00:00–08:00) (Figure 2A). In addition, the percentage of witnessed cardiac arrest per 2-hourly interval was lower between 02:00 and 06:00. However, no obvious temporal differences were found in the percentage of shockable rhythm and receiving DACPR or BSCPR (Figure 2B).In order to assess the association of the occurrence time of OHCA with patient clinical outcomes and achievement of ROSC, multivariate analysis was performed. The variables with p values < 0.1 obtained from the univariate analysis (Table 1) and the associated factors related to outcomes of OHCA (described in the statistical analysis) were adjusted using logistic regression with forward selection analysis. The results showed that there was no significant difference between achieving ROSC in OHCA that occurred at night compared to OHCA that occurred during the daytime (adjusted odds ratio (aOR) = 0.74, 95% confidence interval (CI): 0.477–1.148, p = 0.18) (Table 2). The independent factors associated with achievement of ROSC were witnessed cardiac arrest (aOR = 2.362, 95% CI: 1.658–3.364, p < 0.001), occurrence of cardiac arrest in public areas (aOR = 3.666, 95% CI: 2.033–6.612, p < 0.001), ambulance transport (aOR = 3.944, 95% CI: 1.459–10.664, p = 0.007), use of public AED before EMT arrival (aOR = 3.421, 95% CI: 1.197–9.776, p = 0.022), and prehospital use of mechanical CPR device (aOR = 1.588, 95% CI: 1.122–2.248, p = 0.009) (Table 2).The associations between the occurrence time of OHCA and achievement of sustained (≥24 h) ROSC and survival to discharge are shown in Table 3 and Table 4, respectively. In contrast to achieving ROSC, in the multivariate analysis, OHCA occurring at night had a significantly decreased probability of achieving sustained (≥24 h) ROSC (aOR = 0.489, 95% CI: 0.285–0.840, p = 0.009) compared to OHCA that occurred during the daytime (Table 3). The other independent factors associated with sustained (≥24 h) ROSC were similar to those of any ROSC, including witnessed cardiac arrest (aOR = 1.749, 95% CI: 1.159–2.64, p = 0.008), cardiac arrest occurring in the public area (aOR = 3.906, 95% CI: 2.096–7.28, p < 0.001), ambulance transport (aOR = 4.729, 95% CI: 1.654–13.52, p = 0.004), use of public AED before EMT arrival (aOR = 3.46, 95% CI: 1.218–9.826, p = 0.02), and prehospital use of mechanical CPR device (aOR = 1.761, 95% CI: 1.174–2.64, p = 0.006). Here, shockable rhythm was significant in this outcome (aOR = 1.557, 95% CI: 0.994–2.438, p = 0.053).Consistent with sustained (≥24 h) ROSC, OHCA occurring at night also had a significantly decreased probability of achieving survival to discharge (aOR = 0.147, 95% CI: 0.03–0.714, p = 0.017) compared to OHCA occurring in the daytime (Table 4). Witnessed cardiac arrest (aOR = 2.85, 95% CI: 1.172–6.932, p = 0.021), shockable rhythm (aOR = 5.98, 95% CI: 2.868–12.469, p < 0.001), cardiac arrest in public areas (aOR = 3.523, 95% CI: 1.435–8.647, p = 0.006), and use of public AED before EMT arrival (aOR = 7.811, 95% CI: 1.873–32.581, p = 0.005) were independently associated with survival to discharge (Table 4).In order to analyze the association between the occurrence time of OHCA and clinical outcomes of different variables associated with OHCA, we conducted subgroup analyses according to sex, age group (≥65 years and <65 years), EMS response times (<5 min and ≥5 min), witness status, and initial cardiac rhythm. The confounding factors described above were also adjusted for. The results (Figure 3) revealed that male patients had a significantly decreased probability of achieving sustained ROSC (aOR = 0.367, 95% CI: 0.176–0.766, p = 0.008) and survival to discharge (aOR = 0.11, 95% CI: 0.014–0.881, p = 0.038) when OHCAs occurred at night compared to OHCAs that occurred in the daytime. However, this trend of temporal differences was not significant in female patients. Older adult patients (age ≥ 65 years) had a significantly decreased probability of achieving sustained ROSC (aOR = 0.418, 95% CI: 0.218–0.803, p = 0.009) when OHCA occurred at night, but this was not significant in younger patients (age < 65 years).Significantly decreased probabilities of achieving sustained ROSC (aOR = 0.347, 95% CI: 0.151–0.8, p = 0.013) and survival to discharge (aOR = 0.073, 95% CI: 0.007–0.725, p = 0.026) were found at night, especially in OHCA patients with ≥ 5 min of EMS response time. However, there was no significant temporal difference in patients with < 5 min of EMS response time. With regard to witnessed cardiac arrest, decreased probabilities of achieving sustained ROSC (aOR = 0.435, 95% CI: 0.210–0.903, p = 0.026) and survival to discharge (aOR = 0.103, 95% CI: 0.013–0.807, p = 0.03) were also found at night, but these significances were not observed in non-witnessed OHCA. Regarding initial rhythm, both shockable rhythm (aOR = 0.282, 95% CI: 0.091–0.867, p = 0.027) and non-shockable rhythm (aOR = 0.533, 95% CI: 0.296–0.961, p = 0.036) showed significantly decreased probability of achieving sustained ROSC at nighttime than during daytime.In this study, we observed that the circadian pattern of the incidence of OHCA concurs with existing literature; that is, the incidence of OHCA is lower at night and peaks during the morning [6,31]. In addition, a temporal difference was found in the witness status of OHCA and the clinical outcomes of OHCA. After adjusting for the factors influencing survival, we found that there was no significant temporal difference in achieving ROSC (Table 2). However, the achievements of sustained (≥24 h) ROSC and survival to discharge were significantly poorer for incidences of OHCA at night (Table 3 and Table 4). Further subgroup analysis was performed, which found that temporal differences that influenced clinical outcomes were more significant, especially in male patients with older adult OHCA, OHCA with longer response time (≥5 min), and witnessed OHCA (Figure 3).Core elements described in the Utstein-style guidelines are associated with clinical outcomes of OHCA [4]. However, temporal variations in these factors have not been thoroughly investigated. Regarding patient factors, our study showed a temporal difference in the witness status of OHCA and location of arrest. A lower percentage of witnessed cardiac arrest was found in nighttime incidences of OHCA compared to the daytime or evening (Table 1 and Figure 2). This indicates that a significant number of patients may experience an OHCA at night and die without being witnessed. Subsequently, the identification time of an OHCA by the dispatcher was shorter at night (Table 1). Additionally, a higher percentage of nighttime OHCA incidents occur in medical institutions and less in public areas than daytime or evening OHCAs. Notably, OHCAs occurring in public areas are known to have a better survival rate than in OHCAs occurring at home because of the increased probability of being witnessed and receiving BSCPR and AED [32,33]. Patients living in medical institutions (such as nursing homes) may have more comorbidities than those who do not live in medical institutions, which could contribute to the poor clinical outcomes of OHCA incidents at night. In order to control for these factors, a multivariate analysis was performed (Table 2, Table 3 and Table 4), which revealed that witnessed OHCAs and OHCA incidents in a public area were independently associated with clinical outcomes of OHCA. After adjusting for these confounding factors, achievements of sustained ROSC and survival to discharge were still significantly poor at night. The subgroup analysis specifically for witnessed arrest also showed consistent findings (Figure 3).The association between patient demographics (age and sex) and temporal differences in the clinical outcomes of OHCA was examined, although no obvious differences were found in patient age and sex subgroups between daytime, evening, and nighttime (Table 1). We found that the temporal differences in clinical outcomes were more significant in male and older adult (≥65 years) patients in the subgroup analysis (Figure 3). The interactions between age and sex and the temporal difference may be explained by the underlying etiologies related to cardiac arrest. OHCAs of cardiac origin, such as acute myocardial infarction and subsequent ventricular arrhythmia, have a higher rate of incidence during the day compared to nighttime [34,35]. This implies that OHCA of non-cardiac origin may be more prevalent at nighttime compared to daytime. For males, the probability for one-month survival is lower among OHCA of non-cardiac origin than that of cardiac origin. This discrepancy was more significant in older adult men [14]. This observation may explain why the temporal difference in outcomes of OHCA was more significant in male and older adult patients.Successful prehospital resuscitation can also be influenced by temporal differences, which further affects the outcomes of OHCA. According to previous literature, there are fewer personnel on duty during the night, which can disrupt the circadian rhythm of EMTs and potentially affect performance, motivation, and decision-making ability [36]. This study identified longer EMS response times and longer time spent at the rescue scene at night compared to in the daytime or evening, which may be due to personnel arrangement after midnight and that two EMTs were responsible for the medical calls from 00:00 to 08:00 a.m. Potential factors influencing response time include disruption of sleep and time required to get dressed. However, the number of dispatched EMTs, dispatch of EMTPs, and management during the prehospital phase (insertion of laryngeal mask airway, intravenous epinephrine injection, and use of mechanical CPR device) did not vary with time. Although slightly longer response times (median 5 vs. 4 min) and scene times (median 10 vs. 9 min) were found at night compared to the daytime and evening, there was no significant impact of these EMS time intervals on the clinical outcomes of OHCA in multivariate analysis (Table 2, Table 3 and Table 4). However, subgroup analysis (Figure 3) for OHCAs with a response time ≥ 5 min demonstrated that a poor outcome was more likely during the nighttime. Hence, the temporal group of EMS activity influences the clinical outcome of OHCA. [17,18,37] Conversely, this study has shown that OHCA incidents with relatively short response times and scene times did not have an impact on patient outcomes. Based on our finding, more resuscitation efforts may be needed for patients with prolonged response times, especially at night.Previous studies investigating the association between temporal differences and clinical outcomes of patients with OHCA have shown inconsistent findings [5,7,8,9,10,11,38]. Studies by Matsumura et al. and Ho et al. both showed that 30-day survival was worse in patients with OHCAs occurring at night than during the day [7,38]. These two studies conducted in Asia both observed significantly lower applications of BSCPR during nighttime compared to daytime. Another study investigating the temporal variation in dispatcher-assisted and bystander-initiated resuscitation efforts also showed that the performance of DACPR or BSCPR was less prevalent at night compared to in the day [39]. The above studies attributed poor outcomes at night to lower dispatcher-assisted or bystander resuscitation efforts. Moreover, a recent study conducted in Vienna demonstrated no significant difference between day and night in sustained ROSC rates and survival with favorable neurological outcomes after OHCA, as nearly identical rates of BSCPR were found [9]. However, the relationship between DACPR and BSCPR, temporal differences, and the clinical outcomes of OHCA is still unclear. In contrast to previous studies, this study found no significant difference in the rates of DACPR/BSCPR and start time of DACPR (Table 1 and Figure 2). However, a significant association was found between temporal differences and clinical outcomes of OHCA. Thus, in addition to dispatcher-assisted or bystander resuscitation efforts, other factors contribute to poor outcomes at night.Resuscitation efforts at the in-hospital stage for patients with OHCA have not been extensively studied. For in-hospital cardiac arrest (IHCA), similar temporal differences were observed. Incidences of IHCA during off-hours (at night and during the weekend) experience lower survival rates than incidences of IHCA during the daytime on weekdays [40]. Fewer in-hospital personnel resources during off-hours compared to on weekdays in the daytime may contribute to limited in-hospital resuscitation efforts and influence the patient clinical outcome [41,42]. A study by Matsumura et al. in Japan showed that resuscitation-associated procedures, such as endotracheal intubation and blood gas analysis, were performed less frequently at night compared to the day, despite no difference in the use of epinephrine and defibrillation during the nighttime [7]. One possible reason for this is that in-hospital care providers rarely work 24 h shifts in Japan [7,43]. A study conducted in Vienna investigated patients post-OHCA admitted to a specialized resuscitation center in which percutaneous coronary intervention is available 24/7, with a 30 min call-in-time at night. No differences in survival and neurologic outcomes between day and night admission were observed [11]. Hence, the temporal differences in the outcomes of OHCA may be affected by variation in the quality of post-resuscitation care. This study found no significant association between the temporal group and the initial achievement of ROSC (aOR = 0.74, p = 0.18) (Table 2). However, significance was exhibited in achieving sustained (≥24 h) ROSC (aOR = 0.489, p = 0.009) and even worse in achieving discharge survival (aOR = 0.147, p = 0.017). This may imply that resuscitation efforts from prehospital resuscitation to in-hospital resuscitation cannot be retained, especially at night. Although we adjusted the hospital factors according to the level of the transferred hospital, these findings still suggest that the quality of in-hospital resuscitation efforts may be inadequate at night. The concept of specialized post-arrest care centers for OHCA is needed.This study has several limitations. First, as in other OHCA registry studies, the data regarding management during the in-hospital stage, such as the availability of post-resuscitation care, could not be sufficiently acquired and analyzed. Post-resuscitation care, such as revascularization therapy and therapeutic hypothermia, can influence the patient’s clinical outcome and may vary with the time of day. In order to account for this, adjustments were performed according to the level of the hospital where the patient with OHCA was sent in the study. Second, information on patients’ underlying diseases could not be obtained. Pre-existing conditions may have influenced the clinical outcomes of patients. However, this factor is unlikely to vary with time. Third, the time of EMS call received was used as a substitute for the time of cardiac arrest. This is not a precise measurement, especially for patients with unwitnessed cardiac arrest. However, it is a common problem in several studies [5,7,9,10,38].Temporal differences influence the incidence and clinical outcomes of OHCA in Taiwan. Patients with OHCA at nighttime had a significantly decreased probability of achieving sustained ROSC and survival to discharge compared to patients with OHCA during the daytime and evening. This temporal difference was more significant in male patients, older adult patients, those with longer response times (≥5 min), and witness status. The temporal effects on the clinical outcome of OHCA may be a result of physiological factors, underlying etiology of arrest, resuscitative efforts in prehospital and in-hospital stages, or a combination of these factors. Further research into these findings may help to develop preventative strategies, improve ambulance deployment and hospital interventions, adjust resource allocation of EMS staff, and optimize care for OHCA.M.-J.T. and C.-F.H. conceived the study and developed the study protocols. H.-C.H., T.-Y.L., C.-H.T., Y.-S.S., Y.-R.C., Y.-N.Y. and C.-F.H. provided administrative support and acquisition of data. M.-J.T. performed statistical analysis. H.-C.H., T.-Y.L. and M.-J.T. interpreted the study results and drafted the manuscript. C.-F.H. and M.-J.T. are the corresponding authors who take responsibility for this paper. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of the Ditmanson Medical Foundation Chia-Yi Christian Hospital (CYCH-IRB 2021057).The Institutional Review Board of the Ditmanson Medical Foundation Chia-Yi Christian Hospital (CYCH-IRB 2021057) has approved this study to be exempt from informed consent.The data that support the findings of this paper are available from the corresponding author, M.-J.T., upon reasonable request.We appreciate the dedication of the Chiayi City fire bureau.The authors declare no conflict of interest.Flowchart of the patients included in the study. DNR: Do-Not-Resuscitate order; OHCA: out-of-hospital cardiac arrest.(A)Temporal variation of the frequency and percentage of OHCA, clinical outcomes, (B) DA/BSCPR, witnessed status, and initial cardiac arrest by 2 h intervals of a day. OHCA: out-of-hospital cardiac arrest; DA/BSCPR: dispatcher-assisted or bystander cardiopulmonary resuscitation.Subgroup analyses of nightmare (00:01–08:00) on the outcomes of OHCA patients. aOR: adjusted odds ratio; CI: confidence interval; OHCA: out-of-hospital cardiac arrest; ROSC: return to spontaneous circulation.Baseline characteristics and clinical outcomes by time of day of cardiac arrest occurrence.Values shown are n (%), mean (±SD), or median (interquartile range). AED: automated external defibrillators; BSCPR: bystander cardiopulmonary resuscitation (CPR); DACPR: dispatcher-assisted CPR; EMS: emergency medical services; EMT: emergency medical technician; EMTP: emergency medical technician paramedics; OHCA: out-of-hospital cardiac arrest; ROSC: return to spontaneous circulation.Predictors of achievement of any return of spontaneous circulation.OR: odds ratio; aOR: adjusted OR; BSCPR: bystander cardiopulmonary resuscitation (CPR); CI: confidence interval; DACPR: dispatcher-assisted CPR; EMS: emergency medical services; EMT: emergency medical technician; EMTP: emergency medical technician paramedics; OHCA: out-of-hospital cardiac arrest; ROSC: return to spontaneous circulation.Predictors of achievement of a sustained (≥24 h) return of spontaneous circulation.OR: odds ratio; aOR: adjusted OR; BSCPR: bystander cardiopulmonary resuscitation (CPR); CI: confidence interval; DACPR: dispatcher-assisted CPR; EMS: emergency medical services; EMT: emergency medical technician; EMTP: emergency medical technician paramedics; OHCA: out-of-hospital cardiac arrest; ROSC: return to spontaneous circulation.Predictors of achievement of survival to discharge.OR: odds ratio; aOR: adjusted OR; BSCPR: bystander cardiopulmonary resuscitation (CPR); CI: confidence interval; DACPR: dispatcher-assisted CPR; EMS: emergency medical services; EMT: emergency medical technician; EMTP: emergency medical technician paramedics; OHCA: out-of-hospital cardiac arrest; ROSC: return to spontaneous circulation.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11021.txt

@@ -0,0 +1 @@
+Refugees are at great risk of developing mental health problems. Yet, little is known about how to optimally help this vulnerable group as there is a lack of evaluated refugee mental health interventions. The current article presents the results of a literature review which investigates the importance of place attachment for the promotion of refugees’ well-being in the resettlement process. This review concentrated on the most recent and current literature regarding the potential role, importance, and relevance of people–place bonds in the dynamic process of refugee resettlement. It examines literature from the field of positive and environmental psychology, highlighting key theoretical concepts and research findings as well as gaps in research. The review revealed that little is known about the dynamics of place bonding, while the debate rages on about the geometry of the psychological constructs of person–place relationships. Yet, knowing more about which needs should be satisfied for easing place bonding could be of crucial importance for facilitating refugee well-being. Ultimately, improving the knowledge and understanding of the phases of this dynamic process could be useful for a more successful implementation of refugee resettlement practices and activities.Good health and well-being, decent work and economic growth, and reduced inequalities are three of the UN sustainable development goals for 2030 [1]. Although many efforts have been made, this goal is still out of reach, especially for refugees. It is therefore relevant to examine the current protocols for and future perspectives on promoting refugees’ well-being and positive integration into society. We will give a summary of the present knowledge and relevant literature and the potential role of people–place bonding on the promotion of refugee well-being. Implications for practice and future research are described in order to foster knowledge and understanding of this global issue.The available literature on the role of people–place bonds in the promotion of refugees’ well-being is scarce, fragmented, and incomplete. This paper aims to fill this gap by identifying and examining the most recent and relevant literature which deals with the role refugees’ place-bonding dynamics have for their wellbeing and how these bonds develop over time in new settlements. The following question guided this review: “What is the potential relevance of people–place bonds for the promotion of refugee well-being?”.A literature review synthesizes published material from diverse sources covering a wide range of subjects [2]. Grant and Booth [2] note that a literature review usually includes material with a degree of permanence and possibly has been peer-reviewed. This type of review takes stock of the current body of work and provides a synthesized textual analysis of its contribution and value.For the current study, the databases of PsycINFO, Web of Science and Scopus were used to search for literature. Papers included in this review were published in international peer-reviewed journals. The search terms used were refugees/migrants/immigrants/migration, combined with well-being/positive psychology interventions, positive mental health, place attachment/place identity place meaning/sense of place/place bonding/place making/self-determination theory/continuity or integration. The search terms and criteria were deliberately kept broad to cast the net as wide as possible since there is a limited body of research available on the topic. For similar reasons, no restriction on publication date was imposed. As the literature on refugees is sparse, papers related to migrants were also included in this review. Finally, 79 papers were included in the paper (studies used for the review on refugees and migrants are marked with an asterisk (*) in the reference list), of which 17 specifically related to refugees and 15 to migrants.The literature from, e.g., positive psychology, environmental psychology, etc., was synthesized and is presented as a comprehensive narrative. Gaps in research are identified, and the relevance of the results for the practice of the promotion of refugees’ well-being is discussed.The results of the review are presented in three sections. First, insight gained from the reviewed literature about refugee mental ill-health, mental well-being, and the relevance of mental health well-being interventions is presented. Second, Self-Determination Theory as a strategy/framework for positive integration is reviewed. Third, the role of people–place bonding in the promotion of refugee well-being is discussed.Refugees are one of the most vulnerable groups in our society [3]. They have had to leave their countries, homes, social networks and work due to life-threatening circumstances, such as the fear of being persecuted or threatened by other people and/or environmental hazards. Many refugees go through a perilous journey in search of safety, leaving many traumatized and suffering from mental health problems [4]. A systematic review estimated the chance that refugees in Western countries suffer from post-traumatic stress disorder (PTSD) as ten times more likely than that of the local age-matched population. They also found that the prevalence of serious mental disorders was as high as 9% for PTSD and 5% for major depression, with high co-morbidity [5]. Refugees also have a high risk for suicide and social exclusion. Due to these risk factors, there is a large need for interventions that improve refugee mental health [4].Post-migration experiences have even more impact in terms of undermining refugees’ well-being than pre-migration experiences or trauma [6]. Refugees have to rebuild their lives, often in an environment where they do not feel welcome [7]. During resettlement, refugees are continuously exposed to stressors, such as social, economic and social insecurity, which negatively affect their psychological functioning [8]. Refugees are therefore at great risk of developing mental health problems during the time of resettlement. Poor social integration partly explains the high rates of long-term mental disorders [9] observed in resettled refugees.To obtain a complete picture of refugee’s mental health, in addition to mental illness, positive mental health also has to be considered. Positive mental health, or well-being, is more than the absence of ill-being. Well-being and psychopathology are two related but distinct dimensions of mental health [10]. Well-being, which is studied in the field of positive psychology [11], is beneficial for several outcomes such as coping during difficult times, problem solving, performance in complex mental tasks, and various health outcomes [12].There are three dimensions of well-being. First, emotional or subjective well-being refers to experiencing positive emotions in relation to negative emotions and satisfaction with life [13]. Second, psychological well-being is defined as positive growth, environmental mastery, and positive relationships with others and self-acceptance [14]. Third, social well-being is about social functioning and encompasses the dimensions of social coherence, acceptance, actualization, integration and actualization [15]. In order to flourish, all three of these types of well-being are important [10].When examining determinants of well-being for refugees, coping with adversity is especially important. Resilience (the ability to bounce back after adversity) plays an important role in well-being [16]. On a personal level, self-esteem and positive adaptability to stress are key factors to developing resilience. On a social level, support from families, family cohesion, and peer support facilitate resilience. According to Fielding and Anderson [17], collective resilience facilitates resettlement of refugees by building communities that support refugees in recovering from trauma. It has been shown that resilience can be improved with specific interventions [18].Traumatic experiences can also lead to post-traumatic growth [19]. It can develop as a result of the cognitive processes that are initiated in order to cope with traumatic events, leading to thriving and personal growth post-trauma, resulting in new self-insights, a new sense of meaning and purpose in life, changes in perception, relationships and life priorities [16]. In order to develop post-traumatic growth, hope is vital. A hopeful disposition despite challenging circumstances is a protective factor and helps refugees to acculturate and feel empowered during resettlement in the post-migration phase [20].Several meta-analyses indicate that positive psychology interventions (PPIs) can enhance well-being in the general population and in vulnerable groups, such as those with psychiatric disorders [21,22,23]. Improving well-being is another approach to protecting people from developing mental disorders and increasing the likelihood of recovering from mental illness [24,25,26]. Therefore, addressing well-being in refugees is a promising approach [27], especially because, in many cultures, receiving help for psychological problems is stigmatized and therefore, might be rejected by refugees. Still, PPIs for refugees are rare. One PPI developed for refugee children in Greek refugee camps showed improvements in their well-being, self-esteem, and optimism and depressive symptoms, with the children especially valuing building their strengths and the new sense of togetherness [28].In conclusion, in order to obtain well-being, integration into society is vital, for example via work. While European countries seek ways to integrate refugees into society, little is known on how to optimally help this vulnerable group. There is a lack of both a theoretical framework and consequently a lack of evaluated refugee mental health interventions. Yet, in order to be able to find viable solutions to promote well-being of refugees and their integration into society, we need such a theoretical framework. Only when we can base our practical efforts on sound theory can we develop interventions, projects, and initiatives that will work optimally which can be further improved in a systematic way in the future.A framework explaining the needs and determinants for refugee well-being is Self-Determination Theory (SDT) [29]. SDT is a well-researched theory of motivation and offers a theoretical explanation for how to improve well-being. SDT postulates that every human being has three basic psychological needs: autonomy (the feeling of being the director of one’s own life), competence (the feeling of being good at something) and relatedness (the feeling of being connected with others). All three are required in order to reach and maintain a good level of well-being [30].These basic needs are likely to be unsatisfied in the lives of many refugees, as they have lost control of their lives, are unable to use their skills in a job or hobby, and have lost their social networks. Integration into society fulfills the needs of refugees. For example, integration into a social group, such as one’s neighborhood or a religious community, will create new social bonds and can thus fulfill the need for relatedness. Being an active member of a club, for example, the local football club or a choir, can promote relatedness via shared interests. Moreover, participating actively in a hobby can support the feeling of being good at something, or of having the capacity of becoming better at a skill, and thus fulfills the need for competence. Organizations that support integration should leave decisions up to the refugee to support his or her autonomy.Work is another highly successful way to achieve integration. Finding employment can satisfy basic psychological needs as it allows one to regain autonomy, use one’s competencies, and feel connected to colleagues at work [31]. Finding a job may not only help refugees to provide for themselves but could also improve their well-being and lay the foundation for integration. As an approach to helping people to fulfill their needs, interventions, including job-seeker programs that support refugees in findings employment, should focus on improving competencies (specific to the job they want to apply for), fostering autonomy (regaining a feeling of being in control by actively searching for a job) and relatedness (by building a support network).Satisfaction of the basic psychological needs and well-being can also be improved by place attachment. Visualization of a meaningful place, a place where one feels emotionally connected to, has been found to improve self-esteem, belonging, and meaning, including for people who experience need threats [32]. Place attachment is thus a relevant concept for understanding refugees’ resettlement experience, for whom the satisfaction of their needs is not a given.Place attachment has been defined as a multidimensional affective bond between people and places, involving a symbolic relationship with the place and the willingness to maintaining proximity with the place [33,34,35,36]. The traditional conceptualization of place attachment focuses on residential stability and the creation of person–place bonds through time, habits, and place-related life experiences [37,38]. However, this conceptualization is thought to be simplistic, since place attachment can be ambivalent [39], constitutes various types (traditional and active), and is non-attached (alienation, place relativity, and placelessness) [40]. A growing body of research in recent years has focused on how people develop affective bonds with new places. This includes a focus on contemporary mobility patterns that are characterized not only by migrations but also by a general reduced residential stability [40,41,42].Most of the literature on bonding to places of resettlement has focused on the roles of self-continuity and place making. The concept of self-continuity was first introduced by Hallowell [43] and later incorporated into the identity process theory by Breakwell [44] as the need for the self to be organized through a coherent “story” that links past and present behaviors. Based on Breakwell’s model, Twigger-Ross and Uzzell [45] include self-continuity among the functions of the person–place relationship. For instance, people are known to bond more easily to places with a climate similar to the place they come from [46].Consistently, several studies conducted by Shampa Mazumdar, Sanjoy Mazumdar and colleagues show that self-continuity is a relevant factor to create a bond with new places of settlement for migrants. In a study on the Vietnamese enclave of Little Saigon in Westminster, California, the architectural environment and the immigrants’ social, commercial, and ritual activities were found to contribute to the sense of place of the local Vietnamese community, thus enabling it to simultaneously remain connected to the places of origin while developing significant new place ties [47].The relevance of continuity is also highlighted in within-country migration processes, usually from rural to urban areas. For instance, Becerra, Merino, Webb, and Larrañaga [48] showed how indigenous Mapuche, moving from rural areas to the megalopolis of Santiago de Chile, strengthen their relationship with the new place by recreating and locally translating traditional Mapuche practices. Similar studies have been conducted in China, by Liu, Fu, Van den Bosch, and colleagues [49], showing that integrating landscape elements which are familiar to newcomers could ease the development of place attachment to new urban places of residence. The importance of the role of familiar elements in easing bonding to new places has also been shown by Cheng and Kuo [50] with an experimental design and participants from Macao and Taiwan.Continuity is often achieved through place making, i.e., the possibility of migrants changing private and public spaces to make them look more familiar and similar to places from their country of origin. Place making can aim to make the place look more familiar in terms of design, materials, and practices, with a particular emphasis on the creation of spaces for cultural and religious practices, via place planning and organization, place design, and place rituals [51,52,53,54].Public green areas, such as parks, can help recall the rural landscapes of the home country and thus can be a privileged setting for place bonding, place making and cultural practices among migrants, as shown in studies conducted both in Chile [55,56] and in the USA [57]. Place-making can also be the result of participation in local initiatives, such as gardening, an activity that is practiced by many migrants coming from agrarian backgrounds. This can take place within a diverse community, promoting interactions between migrants and locals, with material and emotional benefits for the newcomers [58,59] as well as the locals.Although most of the literature on place attachment and newcomers focuses on the role of self-continuity, a few very innovative studies focus on the role of basic need satisfaction in the development of place attachment, giving a new perspective that can be directly linked to well-being.Even though need satisfaction has mainly been examined in different contexts such as well-being and secure interpersonal attachment (e.g., La Guardia et al. [60]), some new studies have started to show the role of need satisfaction for the development of place attachment (e.g., Van Riper, Yoon, Kyle, Wallen, Landon, and Raymond [61]). Landon et al. [62] found a correlation between place attachment and basic need satisfaction (as proposed by SDT [60]). Among visitors of wild areas in the Southern Appalachian Region, perceiving a landscape as supporting autonomy, relatedness, and competence was associated with identification, dependence, and emotional connection with that landscape [62]. Similarly, a recent study on place attachment in the context of environmental risk showed that manipulating an evacuation site to show it as more fit to satisfy individuals’ basic psychological needs made participants more likely to declare greater attachment towards it, and more likely to evacuate in case of environmental emergency [63]. These findings do suggest that people develop affective bonds with places that satisfy one or more of their basic needs. It would not be unusual then to observe that it is likely that refugees are able to develop an affective bond with places that satisfy their basic psychological needs.Other results, even if not directly referencing SDT and basic need satisfaction, suggest a link between need satisfaction and the development of a person–place bond [32,64]. Some studies on the topic of place bonding in the context of resettlement also give some hints in this direction. Sampson and Gifford [65], for instance, in a study on young refugees in Australia, found that four kinds of places (places of opportunity, places of restoration, places of sociability, and places of safety) are particularly important to young refugees, as they have a therapeutic and restorative role for newcomers and contribute to their well-being. Among these places, places of opportunities (often schools, but also places for leisure time such as parks) seem to be able to support autonomy and competence, allowing youths to conduct the activities they want to feel competent at. Places of sociability instead refer to the importance of relatedness. Places of safety and places of restoration, however, make reference to needs—i.e., restorativeness, relaxation, safety—that are not part of SDT, but that are commonly known in the place attachment literature.Similarly, a study on refugee women from Myanmar in New Zealand that was part of a multisensory research project found that refugee participation in local initiatives and familiarization with local places of need and of pleasure reduced their stress and anxiety. Additionally, it increased their feelings of safety, autonomy and belonging, ultimately contributing to building place attachment [66]. The relevance of relatedness is reported by many studies, such as a study about Ukrainian immigrants in Poland, where it was found that migrants who had strong ties with the Polish people were more strongly attached to Warsaw, which in turn increased their willingness to stay in Poland [67]. A Dutch study on Syrian refugees reported that many of those who had been assigned a home in small Dutch communities often moved to bigger urban areas where more Syrians could be found [68]. Finally, a study on Afghan refugees in Finland highlighted how different individuals from the same community have different resettlement experiences and need different adapters to establish a successful relationship with the new place [69]. This suggests that the UN Refugee Agency, UNHCR, and institutions working in refugees resettlement processes should take these different place attachment strategies into account for easing refugees’ well-being and successful resettlement [69]. More broadly, in the general population, correlation studies show that carrying out personally significant and involving activities in a given place, is associated with developing and strengthening the psychological meaning of that place for the actor (e.g., in terms of place identity [70]).These few examples seem to draw a connecting line between need satisfaction and place attachment, often by means of specifically directed social-psychologically meaningful activities, thus making a stronger and more relevant link among the environment, the person’s place attachment and their resulting well-being. This highlights the importance of successful place bonding for refugees’ well-being. However, more research is needed to investigate how to integrate the different approaches and results. Many questions remain to be answered such as what is the role of self-continuity in a need satisfaction framework? Is it just another need? How many other needs are there as well as the well-known basic psychological ones (see Ariccio, Lema-Blanco and Bonaiuto) [63]?The process of creating a successful person–place bond in the resettlement location is complex, varied, and dynamic. Some of the literature has recently focused on this last feature, i.e., on what necessary steps or phases help to develop a successful person–place bond in a new place. Even though there is an agreement that this process goes through several steps, no consensus has yet been reached on which steps these are. This is also consistent with the ongoing debate about the geometry and mutual relationships between the different person–place constructs [42,71,72]. For instance, the concept of place identity is considered as the cognitive facet of the person–place relationship, contributing therefore together with personal and social identity to the definition of the individual’s identity. Knez [46] showed in a study about residents’ relationship with the Swedish town of Gothenburg that five dimensions of place identity (place-related distinctiveness, place-referent continuity, place-congruent continuity, place-related self-esteem, place-related self-efficacy) are all predicted by place attachment which in itself is predicted by length of residence.Two studies on place attachment and place identity in natives and non-natives conducted in Spain suggest that place attachment develops before place identity, at least in the case of the non-natives [73]. A contrasting result is the one proposed by Kyle, Jun, and Absher [74] in a study about residents living in the wildland–urban interface outside of San Diego and Los Angeles. They found that the tripartite organization of human–place bonding includes different steps, with place identity working as an antecedent to place attachment and place dependence. This is consistent with the identity theory [75].Similarly, Trąbka [76] conducted a qualitative study on Polish migrants living in London and Oslo and found four phases of person–place bonding in the migration context: place dependence, place discovered, place identity, and place inherited. Place dependence is the functional bond to place that accounts for the relationship people establish with a place due to the activities or tasks that the place enables (see also [77]). Place discovered is defined by Trąbka [76] as encompassing “both behavioral and emotional aspect, and it is best observed among those participants who gain enjoyment, aesthetic pleasure and a sense of mastery from an intentional familiarizing with the city as such” (p. 70). Place identity is considered as a cognitive effect of place dependence that makes the place become part of the self. Place inherited is defined as “a strong and taken for granted bond, prevalent among people who have spent their whole life in one place, who have strong family connections in it and who cannot imagine leaving” [76] (p. 71). According to Trąbka [76], it is strongly associated with length of residence and it can be found in refugees who have been in London or Oslo for more than ten years and often arrived there at a young age. These phases of person–place bonding often coexist and may emerge gradually in the process of adaptation to a new place, contributing to the overarching dimension of place attachment, with a person–place relationship that evolves from functional bonds to feelings of belonging. Another study about people living abroad for long-term periods in a variety of countries highlights the coexistence of different kinds of person–place relationships between the individual and the host country [78]. Individuals seem to have multi-local identities, feeling like tourists, immigrants, and locals in their new settlement. Even though a temporal continuum can be found in the emergence of these identities, each can coexist within the same individual, triggered by everyday micro-moments and by the relationship with their countries of origin and their country of resettlement [78].This paper presented a literature review on the role of place attachment for refugee well-being, focusing particularly on how this bond develops both in term of antecedents and as a dynamic process. The literature on this topic is fragmented. We focused primarily on refugees, who come from contexts of war and lack of safety and are thus particularly fragile from a psychological point of view. This makes it especially relevant to investigate how their well-being can be improved in resettlement contexts. We also discussed studies on people that do not strictly fit into the definition of “refugees” (i.e., migrants, intra-state migrants), as it is plausible that the psychological processes of bonding to a new place are similar for non-refugees that move to a different country. It should also be noted that when dealing with people who move to a different country for a long-term stay, different terminologies are employed depending on the legal status of these individuals and their backgrounds and motivations for travelling, e.g., “refugee”, “migrant” or “asylum seeker” [79], although their resettlement experiences may be similar.The literature summarized in this paper shows that a promising approach to refugee resettlement is focusing on positive aspects, such as hope and support of need fulfillment to support refugees instead of the traditional focus on (solving) problems. SDT is a practical approach to helping refugees (who often suffer from mental health problems) by promoting their positive mental health. Supporting autonomy, relatedness, and competence can be attempted in various ways that can be adapted to each specific context. Professionals working with refugees, who are often used to a problem-based approach, would likely need additional training to implement this very different approach.For well-being interventions to be effective, it is fundamental that all three basic psychological needs are addressed. For example, merely focusing on fostering competence development (e.g., interventions aiming at developing job applying skills) will not be as effective on refugees’ well-being if autonomy and relatedness are not also taken into account. In general, interventions should be based on the actual needs of refugees. Refugees should be involved in the development of interventions as much as possible. This will increase their sense of autonomy and relatedness in their integration process. Unfortunately, few evaluated well-being interventions for refugees are available that have an emphasis on people–place bonding processes.Migration processes are one of the contemporary phenomena that question traditional environmental psychology constructs such as the idea of a singular, meaningful place for the individual. Research is growing on how people create new person–place bonds while maintaining the old ones and simultaneously adapting to changing places. This highlights the role of continuity, empowerment, and place making as ways to maintain and create new bonds (e.g., Di Masso et al. [42]). However, studies on migration show that bonding with a place can be a long and articulated process, whose different steps and dynamics are still to be defined and investigated [41,76]. In this sense, migration and refugees are contemporary social phenomenon that provide an interesting angle on the constructs and theories usually studied by person–place relationship studies.On the other hand, understanding more of how refugees develop a bond with their new places of settlement would be of fundamental importance to ease refugees’ integration and promote their well-being. For instance, it would be fruitful to allow refugees to bond with a place that has familiarity with their original place of attachment. This is consistent with the importance of continuity. In this sense, enabling place-making would also be an efficient strategy, since refugees would be able to actively improve the place, adding familiarity and thus easing the bonding process. Similarly, consistent with SDT, it should be possible to facilitate person–place bonds in a migration context by making places of settlement able to satisfy refugees’ psychological basic needs for autonomy, competence and connectedness.The role of place attachment within the SDT framework is especially promising for the role of people–place bonds in the process of well-being promotion. The development of place attachment seems to be affected by self-continuity and by the satisfaction of psychological needs. However, it is not clear how these are inter-related. Is self-continuity a need specific to place attachment in relocation contexts? Are there other needs to be satisfied (e.g., safety, relaxation)? Several studies on environmental psychology try to list the needs places can satisfy and their role in the development of place-bonding [32,63,64]. As there is still a lack of systematization of these needs and especially of integration with existing theories such as SDT, more research is needed in order to optimize the effect of well-being interventions for refugees.Refugees are a vulnerable group at risk for mental health problems and lesser well-being, with their potential often being undervalued or neglected in their new resettlement environment. To understand their situation and to support positive integration, Self-Determination Theory is a useful framework. If the need for autonomy, competence and relatedness is supported, the circumstances for positive mental health are shaped. Little is still known about the steps involved in place bonding which is consistent with the ongoing debate about the geometry of the psychological constructs of person–place relationship. Yet, knowing more about needs to be satisfied for easing place bonding could be of capital importance for facilitating refugee well-being. Ultimately, improving knowledge and understandings of the phases of this dynamic process could be useful for a more successful implementation of refugee resettlement practices and activities.T.A., S.A. and L.A.W. were involved in the literature review process and writing of the review. T.A., S.A., L.A.W., F.D. and M.B. took part in the manuscript revision. All authors have read and agreed to the published version of the manuscript.This research received no external funding.Not applicable.Not applicable.The authors are grateful to the Department of Psychology of Developmental and Socialization Processes at Sapienza University of Rome, Italy, for support regarding the editorial costs. The authors also thank Laura Loritz for proof-reading the manuscript.The authors declare that the literature review was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11022.txt

@@ -0,0 +1,218 @@
+Restraint use in Australian residential aged care has been highlighted by the media, and investigated by researchers, government and advocacy bodies. In 2018, the Royal Commission into Aged Care selected ‘Restraint’ as a key focus of inquiry. Subsequently, Federal legislation was passed to ensure restraint is only used in residential aged care services as the ‘last resort’. To inform and develop Government educational resources, we conducted qualitative research to gain greater understanding of the experiences and attitudes of aged care stakeholders around restraint practice. Semi-structured interviews were held with 28 participants, comprising nurses, care staff, physicians, physiotherapists, pharmacists and relatives. Two focus groups were also conducted to ascertain the views of residential and community aged care senior management staff. Data were thematically analyzed using a pragmatic approach of inductive and deductive coding and theme development. Five themes were identified during the study: 1. Understanding of restraint; 2. Support for legislation; 3. Restraint-free environments are not possible; 4. Low-level restraint; 5. Restraint in the community is uncharted. Although most staff, health practitioners and relatives have a basic understanding of restraint, more education is needed at a conceptual level to enable them to identify and avoid restraint practice, particularly ‘low-level’ forms and chemical restraint. There was strong support for the new restraint regulations, but most interviewees admitted they were unsure what the legislation entailed. With regards to resources, stakeholders wanted recognition that there were times when restraint was necessary and advice on what to do in these situations, as opposed to unrealistic aspirations for restraint-free care. Stakeholders reported greater oversight of restraint in residential aged care but specified that community restraint use was largely unknown. Research is needed to investigate the extent and types of restraint practice in community aged care.Over the past decade, there has been increased scrutiny on high rates of psychotropic use and restraint practice in Australian residential aged care from researchers [1,2,3,4], the media [5], policymakers [6,7], human rights [8] and advocacy groups [9]. This attention led to restraint being highlighted as a key area of focus for the Royal Commission into Aged Care Quality and Safety which commenced in 2018 [10]. Further, the Aged Care Act was amended in 2019 to include Australia’s first legislation regulating the use of restraint in residential aged care. From 1 July 2019, aged care providers have explicit obligations in relation to restraint use [11]. The use of restraint must be the strategy of last resort after rigorous assessment and other non-restraint approaches have been trialed. When judged as appropriate, the use of restraint must be the least restrictive form only after informed consent is gained. Moreover, all use must be monitored and reviewed on a regular basis [11].‘Restraint’ is defined by the Australian Medical Association (AMA) as ‘a device or medication that is used for the purpose of restricting the movement and/or behavior of a person’ [12]. The use of restraint in people receiving aged care is often justified based on reducing risk or preventing harm to the person or others [3]. Yet, restraint use is associated with detrimental consequences, including cognitive decline, increased falls, pressure injuries, lessened activities of daily living (ADLs) and death [3,4,13,14,15]. Despite these adverse effects, restraint is used commonly in residential aged care. A recent systematic review and meta-analysis cited the average prevalence over the last two decades of physical restraint as a third (33%) of all residents and chemical restraint use as 32% of residents [16].Ideally, aged care providers, wherever situated, should strive for a restraint-free environment. However, in practice, it is often difficult to balance risk management with the promotion of autonomy for older people needing care. Similarly, it can be challenging to provide a safe environment but at the same time enhance a person’s quality of life [17]. There will be situations when difficult decisions regarding restraint need to be made. Recognizing this, in 2012, the Australian Department of Health developed a set of resources, the ‘Decision-Making Tools’ (DMT), to guide providers, nursing and care staff, care-givers and, wherever possible, relatives and residents, to make informed decisions about restraint [18,19]. At the start of 2020, our research group was commissioned to update these resources to align with legislative changes and contemporary best practice.To inform and develop clear, practical and influential resources we conducted qualitative research aimed to explore the attitudes, beliefs and experiences of a diverse group of interdisciplinary stakeholders towards restraint use in aged care. The current interpretations of what constitutes ‘restraint’ were scoped, along with views on amendments to the Australian Aged Care legislation relating to restraint [9].Qualitative research can be defined as the study of the nature of phenomena and strives to understand why something is, or is not, observed [20]. To achieve our research aim we conducted interviews with a variety of care providers and relatives and two dedicated focus groups with management staff. The triangulation of interview and focus group qualitative data was intended to achieve an in-depth understanding of restraint practice, incorporating both individual perspectives and views of homogeneous groups with relevant expertise and experience [20].Participants of the interviews and focus groups were purposely selected to represent the key stakeholders involved when restraint is proposed and used in aged care. The semi-structured interviews were held with health practitioners working within, or those with relatives living in aged care settings, including residential, community and day care. The first focus group included senior nurse managers and clinical directors based in residential aged care; the second focus group was comprised of community care managers.For the semi-structured interviews, we recruited participants working in various roles and aged care settings to capture a wide range of interdisciplinary health practitioner perspectives, including registered nurses (RNs), enrolled nurses (ENs), personal care assistants (PCAs), physicians, physiotherapists and pharmacists. We also sought to obtain viewpoints from relatives of people receiving aged care. Most of the interviews were held in Hobart, Tasmania, where the research was based, but participants from other Australian States were also sought. Potential participants were identified by all members of the interdisciplinary research team across their professional networks, e-mailed an information sheet and invited to take part. Those who responded were phoned by J.B. or C.S. who outlined the study and arranged an interview after gaining verbal consent. Written consent was obtained before each interview and focus group.The focus group participants were recruited by J.B. who sent emails to potential candidate aged care home provider groups and community care providers inviting them to be part of the study. The focus groups were initially intended to be held before the interviews; however, due to workload and uncertainty associated with COVID-19 they were conducted after the majority of the semi-structured interviews were completed. Both focus groups were moderated by J.B. and conducted remotely via video internet platforms.Approval for this research was obtained from the Tasmanian Health and Medical Human Research Ethics Committee (ID: 20044). As part of the ethics approval, all identifying information was removed from interview data and details of participants were kept confidential. All participants were assured that they could withdraw from the interview or the study at any time. The interviews were conducted between 23 April 2020 and 1 June 2020 and the two focus groups were held in the final week of May 2020. After their interview or focus group, each participant was offered a $100 book voucher to compensate them for the prereading and their interview time. A third of interviewees declined this incentive.Prior to the semi-structured interviews, all participants were sent a link to the 2012 DMT resource and told they would be asked for their opinions on this resource during their interview [18,19]. Originally, we planned to conduct most of the semi-structured interviews face-to-face; however, due to COVID-19 restrictions, all interviews were conducted by J.B. and C.S. remotely through internet video-meeting platforms or by phone.The interviews were conducted using an interview guide (see Table 1) developed by all members of the research team which consisted of three pharmacists, two nurses, a physiotherapist, physician and a speech pathologist. Standard demographic questions were followed by a series of closed and open-ended questions, which were adapted as the study progressed and new areas of enquiry emerged [20]. Participants were free to express their views and experiences and diverge from the interview guide. Likewise, the interviewers were free to ask additional questions or omit questions when not considered relevant. During the interview, participants were asked to recall a case where restraints had been proposed or used, so their real-life experiences were described.As with the individual interviews, each focus group participant was given a link to the 2012 DMT resource [18,19] to enable content review and completed an on-line demographic questionnaire. A topic guide based on the semi-structured interview guide (see Table 1) was customized for residential aged care or community home care for the first and second focus group, respectively. The Focus Group moderator (J.B.) began each session with an introduction to encourage an open environment for participants to share their opinions and experiences [20]. The moderator then initiated group discussion using open-ended questions from the topic guide and ensured that each participant was given several opportunities to speak during the session.The interviews and focus groups were recorded, stored confidentially and transcribed verbatim by a professional transcription company. Returned transcripts were anonymized and missing/unclear words clarified by C.S. and J.B. by listening to the original recording. Both focus group transcripts and a sample of semi-structured interview transcripts were checked by J.B. and C.S. to verify the accuracy of the transcription. Interview participants were also offered the opportunity to check and amend their transcribed interviews. Nineteen participants were sent transcripts and three made slight amendments. All interview data files were uploaded onto NVivo 12 software for analyses [21].Data were thematically analyzed using a pragmatic approach in which the most appropriate research methods were chosen to investigate the topic as opposed to a single paradigm based on a philosophical doctrine [22]. The ‘Framework Method’ [23], often used in applied health care research, was used to answer the research question about what constitutes ‘restraint’. A bubble plot was used to visually represent the frequency of each theme, along with their conceptualized interconnections, creating a framework of these elements grouping codes into sub-themes [23]. Qualitative analysis was undertaken using Braun and Clarke’s six-step process of thematic analysis which involved data familiarization; interim code generation; seeking themes; reviewing themes; defining and naming themes; write-up [24].To start, three authors (J.B., L.G. and B.W.) familiarized themselves with the semi-structured interviews, made notes and discussed findings. Likewise, J.B. and B.W. read through the focus group transcripts several times and met to review findings. Then, data were independently coded by J.B. and B.W. using a hybrid inductive and deductive approach, with emerging themes hierarchically coded utilizing the NVivo 12 platform [21]. The data were organized into themes and sub-themes, mapped, and interpreted. As themes were identified, they were cross-checked and debated. Any differences in interpretation were resolved by discussion and adjusted until consensus was reached. Exemplar quotes supporting each theme were captured by J.B and B.W.Participant demographics for the semi-structured interviews are listed in Table 2. A total of 28 participants were interviewed over a six-week period from April to June 2020. Most were females (n = 22) and 6 were males, with ages ranging between 28–68 years. Ten interviewees were RNs (4 were also clinical care managers), 5 were PCAs and 1 EN. The other participants included 3 physicians, 3 physiotherapists, 3 pharmacists and 3 relatives. The majority (n = 16) of health practitioners interviewed worked in residential aged care, 4 worked exclusively within community care and 2 in day-care centers. Three of the participants, all physicians, worked in both residential and community aged care settings. The majority of participants (n = 15) had worked in aged care for ten years or longer. Of the three relatives, one had a sibling living in residential care, one had a parent receiving community care and the final relative’s parent attended a day-care and respite center. Sixteen of the participants lived in Tasmania and 12 were based elsewhere in Australia. The semi-structured interviews lasted between 35 to 85 min.Demographic data for each focus group are as follows: Six residential care managers, all RNs, participated in the first group (FG Residential). Four were based in Tasmania and two in Victoria, all were female and aged between 48 and 64 years. For the second focus group (FG Community), three community managers participated (two occupational therapists and an RN). All were female and aged between 42 to 53 years. Two other community care managers from Tasmania had verbally agreed to participate in the second focus group but withdrew without explanation on the day of the meeting. The residential care focus group ran for 87 min and the community focus group for 65 min.Following analysis, five themes relating to restraint practice in aged care settings were identified. They were: “understanding of restraint”, “support for legislation”, “a restraint-free environment is not achievable”, “low-level restraint” and “community restraint use is uncharted”. These themes are described further below.All the participants, except two, were able to define what restraint meant to them. Several participants provided more than one interpretation. One of the physicians refused to give a definition, claiming that to do so was a “meaningless circular pursuit”. Over half of those interviewed defined restraint under the sub-theme of ‘limiting what people do’ stating that restraint involved stopping people doing things, restricted their movement or impeded their freedom.
+            
+              
+                “It is the act of stopping someone from doing something they want to do. Whatever they want to do, be it a decision or action.”
+              
+              Physio 1
+            
+          
+                “It is the act of stopping someone from doing something they want to do. Whatever they want to do, be it a decision or action.”
+              The next most common sub-theme is related to ensuring the safety of the older person or others. Notably, those defining restraint as needed for safety or to reduce harm were predominantly RNs working in residential care. A few participants, all pharmacists, defined restraint as a means to control a person’s behavior. Several interviewees defined restraint more broadly as also impeding the ability to make choices.
+            
+              
+                “It’s about restricting movement, restricting rights, anything about the person’s ability to retain their independence or choice.”
+              
+              FG Residential
+            
+          
+                “It’s about restricting movement, restricting rights, anything about the person’s ability to retain their independence or choice.”
+              Three participants stated that restraint altered the mind of a person or the way they thought. Finally, all the PCAs and one of the relatives defined the term in literal terms as either ‘physical’ or ‘chemical’ forms of restraint.The codes and sub-themes relating to the overriding theme of ‘understanding of restraint’ are presented as a bubble plot below (Figure 1). A summative approach was used to calculate the total number of definitions including a coded element. The size of the bubble is proportional to the number of definitions coded to each sub-theme [23].Many participants felt that definitions of restraint varied widely depending on the aged care organization, the aged care setting and between staff working within an organization.
+            
+              
+                “I think a lot of it comes down to sometimes how people define it, and obviously that changes hugely. Even in one facility, you talk to maybe the manager and they say one thing, and then the RN thinks something different, and then someone else thinks something different; so it can be quite confusing.”
+              
+              Pharm 2
+            
+          
+                “I think a lot of it comes down to sometimes how people define it, and obviously that changes hugely. Even in one facility, you talk to maybe the manager and they say one thing, and then the RN thinks something different, and then someone else thinks something different; so it can be quite confusing.”
+              Several participants mentioned inconsistencies in restraint definitions used in different government publications, the aged care sector and the National Disability Insurance Scheme (NDIS). This lack of clarity and consistency about what restraint means confuses staff and other health practitioners working within aged care.
+            
+              
+                “In the residential context it’s even more complicated, because the national quality indicator program defines ‘restraint’ differently to the Legislation. And in providers where there’s mixed circumstances like ours, where potentially you could have staff providing support to community and residential aged care, it’s diabolical….. we’ve also got NDIS consumers in the community where the restrictive practices obligations are different again.”
+              
+              FG Community
+            
+          
+                “In the residential context it’s even more complicated, because the national quality indicator program defines ‘restraint’ differently to the Legislation. And in providers where there’s mixed circumstances like ours, where potentially you could have staff providing support to community and residential aged care, it’s diabolical….. we’ve also got NDIS consumers in the community where the restrictive practices obligations are different again.”
+              Participants were very supportive of the new restraint legislation that had been introduced for residential aged care, claiming that it had heightened awareness, made staff seek alternative strategies, ensured greater accountability and had already impacted use.
+            
+              
+                “I think it’s great. It’s reined people in. It’s made everybody think about what we’re doing as opposed to just this person is disruptive on the evening shift, we don’t have time to deal with this so let’s just give him something to shut him up. Because that’s what was happening.”
+              
+              RN 9
+            
+          
+                “I think it’s great. It’s reined people in. It’s made everybody think about what we’re doing as opposed to just this person is disruptive on the evening shift, we don’t have time to deal with this so let’s just give him something to shut him up. Because that’s what was happening.”
+              Some participants felt the legislation had directly enhanced interprofessional collaboration, particularly around the use and review of chemical restraint.
+            
+              
+                “It has made us focus on the chemical restraints a lot more, and we’ve had a lot more conversations, with General Practitioners (GPs), around ceasing, than we would have had before.”
+              
+              RN 8
+            
+          
+                “It has made us focus on the chemical restraints a lot more, and we’ve had a lot more conversations, with General Practitioners (GPs), around ceasing, than we would have had before.”
+              Although most people interviewed were supportive of the tightened regulation many admitted that they were not entirely sure what it involved. Several nursing staff said the legislative changes had imposed additional workload, such as increased documentation and ensuring informed consent had been gained. Yet, despite increased obligations on providers, most felt the additional reporting and safeguards were worth it to reduce restraint practice.
+            
+              
+                “It’s a massive pain, but I think it needed to be done and I’d rather them go ridiculous and way over-report and me have to deal with the paperwork nightmare for the next two years and then slowly reduce it and actually catch out some of the people that were doing the wrong thing.”
+              
+              RN 1
+            
+          
+                “It’s a massive pain, but I think it needed to be done and I’d rather them go ridiculous and way over-report and me have to deal with the paperwork nightmare for the next two years and then slowly reduce it and actually catch out some of the people that were doing the wrong thing.”
+              All participants were asked to read the DMT restraint resources [18,19] which incorporate the title: ‘supporting a restraint-free environment’. Yet most queried the feasibility of ‘restraint-free’ practice; expressing the view that locked doors and gates, both forms of environmental restraint, were crucial to have in aged care, particularly when many clients were highly cognitively impaired. If locks were categorized as restraint use, then using restraint in most settings was unavoidable.
+            
+              
+                “I don’t know that restraint-free practice is—I don’t know that it’s possible. I mean, we’re talking about me, here. Restraint-free means my door’s open. It’s not ideal. It’s not safe, it’s not possible. Well, it’s possible, I can do it, but what do I say? Well, he got run over yesterday, told him he shouldn’t have gone out the gate.”
+              
+              PCA day care 1
+            
+          
+                “I don’t know that restraint-free practice is—I don’t know that it’s possible. I mean, we’re talking about me, here. Restraint-free means my door’s open. It’s not ideal. It’s not safe, it’s not possible. Well, it’s possible, I can do it, but what do I say? Well, he got run over yesterday, told him he shouldn’t have gone out the gate.”
+              Similarly, many participants expressed the opinion that chemical restraint could never be completely eliminated but instead it was more important to ensure they were used appropriately when prescribed.
+            
+              
+                “I don’t believe it’s possible to have zero antipsychotics in a facility, or zero psychotropics in a facility, but I definitely think that it should be possible to have only those who have a clear diagnosis, a clear plan, and it’s all monitored.”
+              
+              Pharm 3
+            
+          
+                “I don’t believe it’s possible to have zero antipsychotics in a facility, or zero psychotropics in a facility, but I definitely think that it should be possible to have only those who have a clear diagnosis, a clear plan, and it’s all monitored.”
+              There was consensus from most stakeholders that restraint was sometimes needed and that the overarching emphasis should be on minimizing use, not to condone all use.
+            
+              
+                “I think that needs to be clear from the get-go with recognition that sometimes, restraint is necessary to prevent people from harming themselves or coming to harm or harming other people.”
+              
+              Relative 2
+            
+          
+                “I think that needs to be clear from the get-go with recognition that sometimes, restraint is necessary to prevent people from harming themselves or coming to harm or harming other people.”
+              
+            
+              
+                “The care staff know that it’s never going to be a restraint-free environment. To minimize the impact of restraint, you’re minimizing them and having as little as possible.”
+              
+              EN
+            
+          
+                “The care staff know that it’s never going to be a restraint-free environment. To minimize the impact of restraint, you’re minimizing them and having as little as possible.”
+              Some of the managers in the residential aged care focus group commented that some homes were catering to the ‘restraint-free spin’. They felt that homes voicing they were ‘restraint-free’ demonstrated a limited understanding of what restraint meant and the types of practices it entailed.
+            
+              
+                “People want to say, “We don’t have restraint here,” and that’s such a big aspirational target. I think in some instances, there’s real ignorance about what restraint is and what it looks like.”
+              
+              FG Residential
+            
+          
+                “People want to say, “We don’t have restraint here,” and that’s such a big aspirational target. I think in some instances, there’s real ignorance about what restraint is and what it looks like.”
+              During the residential care focus group and in many of the semi-structured interviews, participants referred to the use of low-level or less obvious forms of restraint. From the residential focus group:
+            
+              
+                “All of our facilities say they don’t use any physical restraint, but we found physical restraint: the pushing the chair under the table, the locked doors to outside areas, so a whole lot of things that aren’t seen as hard physical restraint but are definitively restraining.”
+              
+              FG Residential Participant 1
+            
+          
+                “All of our facilities say they don’t use any physical restraint, but we found physical restraint: the pushing the chair under the table, the locked doors to outside areas, so a whole lot of things that aren’t seen as hard physical restraint but are definitively restraining.”
+              
+            
+              
+                “I see things like call bells that are dropped on the floor or not in positions to allow the person to get assistance.”
+              
+              FG Residential Participant 2
+            
+          
+                “I see things like call bells that are dropped on the floor or not in positions to allow the person to get assistance.”
+              
+            
+              
+                “Even just simple things like leaving a tray-table across a chair that is being used for having a meal or an activity but then not removing it, so the person is free to move around.”
+              
+              FG Residential Participant 3
+            
+          
+                “Even just simple things like leaving a tray-table across a chair that is being used for having a meal or an activity but then not removing it, so the person is free to move around.”
+              Participants also referred to practices such as tucking in bed sheets tightly to restrict a person’s movement, taking cushions away from deep armchairs and the use of low beds or princess chairs as forms of restraint. One physician mentioned that not accommodating for hearing and/or sight impairment could also be viewed as restraint. Some felt that in most cases this ‘low-level’ restraint use was unintentional and spoke to a lack of awareness and the need for more education on this issue.
+            
+              
+                “It’s making people aware of what is considered to be a restraint is really important too. So things like the princess chairs that they use. Or even someone who’s got some sort of incapacity, so they can’t hear everyone, or they can’t see everyone, they’re not able to access help when they want to access help.”
+              
+              Physician 1
+            
+          
+                “It’s making people aware of what is considered to be a restraint is really important too. So things like the princess chairs that they use. Or even someone who’s got some sort of incapacity, so they can’t hear everyone, or they can’t see everyone, they’re not able to access help when they want to access help.”
+              In contrast, a few participants implied that low-level restraint was used commonly to compensate for a lack of staff or to allow staff to assist other residents.
+            
+              
+                “Everyone’s got to try and put other people to bed…one resident can sit there for half an hour after dinner’s finished all by themselves, with the wheelchair locked, because they don’t want them to get away. But no-one’s there to take them back to their room and help them out.”
+              
+              PCA 3
+            
+          
+                “Everyone’s got to try and put other people to bed…one resident can sit there for half an hour after dinner’s finished all by themselves, with the wheelchair locked, because they don’t want them to get away. But no-one’s there to take them back to their room and help them out.”
+              
+            
+              
+                “I suspect, across the board, there’s a lot of, what I would call, low-level restraint to be able to implement the care of anyone in a facility like ours, which is for people with dementia…It’s one of those things where you end up in this argument….”Well, if we can’t do that, we can’t actually implement any care.”
+              
+              Physio 2
+            
+          
+                “I suspect, across the board, there’s a lot of, what I would call, low-level restraint to be able to implement the care of anyone in a facility like ours, which is for people with dementia…It’s one of those things where you end up in this argument….”Well, if we can’t do that, we can’t actually implement any care.”
+              Community-based nursing staff and PCAs faced additional barriers when it came to identifying and minimizing restraint. In response to the question, “do you think that the practice of restraining people occurs commonly in community aged care?” the community focus group members replied:
+            
+              
+                “It’s difficult to judge. It would just be more difficult to gauge in a community setting than it would be in residential aged care because of what we’re in there for and what we’re not caring for.”
+              
+              FG Community Participant 1
+            
+          
+                “It’s difficult to judge. It would just be more difficult to gauge in a community setting than it would be in residential aged care because of what we’re in there for and what we’re not caring for.”
+              
+            
+              
+                “That’s right. And certainly, Home Care Packages, the majority of them, we don’t actually see what medications they’re on, because we’re not providing clinical care; we’re providing case management and other community services……so it could be a really hidden problem.”
+              
+              FG Community Participant 2
+            
+          
+                “That’s right. And certainly, Home Care Packages, the majority of them, we don’t actually see what medications they’re on, because we’re not providing clinical care; we’re providing case management and other community services……so it could be a really hidden problem.”
+              Participants working for community service providers said that they had limited control over what happened in a client’s own home. If the family installed a bed-rail they were unable to prevent its use when they were not there. In addition, they stressed that many people with home care packages opted not to use their funding for clinical care, including medication management, meaning that the use of chemical restraint could not be ascertained. Adding to the complexity was the use of multiple care providers by the same client, the lack of home visits made by GPs and the need to manage relationships with certain clients who were resistant to having assistance with care:
+            
+              
+                “I think you have to step so carefully with some people in the community. Even if we’ve got concerns, we’ve got to be really careful how we manage that, so we don’t affect the relationship, the provisional relationship with the client.”
+              
+              FG Community Participant 3
+            
+          
+                “I think you have to step so carefully with some people in the community. Even if we’ve got concerns, we’ve got to be really careful how we manage that, so we don’t affect the relationship, the provisional relationship with the client.”
+              
+            
+              
+                “Absolutely. We go into homes and there’s medications all over the floor, and we can’t do anything about that, we just have to report it.”
+              
+              FG Community Participant 2
+            
+          
+                “Absolutely. We go into homes and there’s medications all over the floor, and we can’t do anything about that, we just have to report it.”
+              Several community nurses and PCAs stressed that they fostered client independence and encouraged family involvement, rather than let the service ‘take over’. Paradoxically, several relatives of older people receiving community aged care reported situations where they were not consulted when restraint was proposed and subsequently used:
+            
+              
+                “The community nurse called Dad’s GP and to her credit, the GP was reluctant to put Dad on this antipsychotic but everybody else was pushing for it, so she did write a prescription and they used it and then they let me know by email after it was done.”
+              
+              Relative 3
+            
+          
+                “The community nurse called Dad’s GP and to her credit, the GP was reluctant to put Dad on this antipsychotic but everybody else was pushing for it, so she did write a prescription and they used it and then they let me know by email after it was done.”
+              
+            
+              Interviewer: “So, they didn’t ask your permission at all, it was just more informing you that it happened?”
+              
+                “That’s exactly it.”
+              
+              Relative 3
+            
+          Interviewer: “So, they didn’t ask your permission at all, it was just more informing you that it happened?”
+                “That’s exactly it.”
+              All three relatives interviewed agreed there is was a strong need for those receiving community-based care and their families to receive guidance on restraint practice.This study provides insight into the attitudes, beliefs and experiences of a diverse group of interdisciplinary stakeholders towards restraint use in aged care. ‘Interdisciplinary’ is a term used to describe healthcare practitioners from different professional disciplines who work together to manage the care needs of a person. Whenever possible the person and their family should be an integral part of this group [25]. Our interdisciplinary stakeholders were nursing staff, PCAs, physicians, physiotherapists and pharmacists, as well as relatives of people receiving aged care. Aged care clients, also known as consumers or residents, were not directly involved in this research due to ethical considerations around capacity to consent, alongside restrictions and uncertainty associated with COVID-19. However, a group of Australian researchers was able to interview community-based older people about restraint, reporting that they were conscious of this issue and concerned about being on the receiving end of such practice [26]. Those interviewed were most averse to the use of physical restraint and sedation, which were perceived to have the greatest impact concerning limiting choice and self-expression {26].Several recent systematic reviews have reported that definitions of restraint in the research literature are highly variable, with interpretations differing according to country, aged care setting, and the timeframe in which studies were conducted [3,27,28]. When we asked our participants to define what restraint meant to them we noted similar differences in understanding. The definitions given by our stakeholders tended to vary according to the professional background of the participant; for example, all the physiotherapists defined restraint as restricting the movement of a person. Likewise, all three pharmacist participants referred to restraint as a way of managing behavior. This would be expected given physiotherapists specialize in movement and pharmacists provide advice on medication that affects mood and behavior. Yet, in spite of a degree of professional variation in interpretation, more than half of the participants defined restraint as ‘limiting what a person can do’ with regards to how they act, move and their overall freedom. This broad definition aligns with the 2019 legislative definition of restraint as ‘any practice, device or action that interferes with a consumer’s ability to make a decision or restricts a consumer’s free movement’ [11] and provides some indication that most people in our study conceptualize restraint in line with the new legislation.The second most common definition of restraint cited by participants, predominantly RNs, was that restraint ensures the safety of older people receiving care and reduces the risk of them harming themselves or others. This is not, in essence, a definition but rather provides justification for why restraint is used. The rationalization for restraint; ‘under the premise of risk minimization and prevention of harm to self or others’, has been reported in Australia as far back as 2005 [28]. It has also been reported in several government enquiries conducted about Oakden, an older person’s mental health service in Adelaide, South Australia [29], despite a lack of evidence that restraint, either physical or chemical, protects residents against injuries and falls [3,14,15,28].Participants were far more likely to define restraint as restricting movement and actions than to control behavior, an aspect which relates to the definition of chemical restraint as ‘a practice or intervention that involves the use of medication for the primary purpose of influencing a care recipient’s behavior’ [30]. This may speak to the difficulty of determining if a psychotropic medication is used for a medical or mental health condition as opposed to influencing behavior. Some older people receiving care may present with a combination of behavioral symptoms and mental illness. The reluctance to define chemical restraint was also observed in a recent systematic review examining the prevalence of restraint practice in residential aged care [16]. The authors of this review could locate only four studies that provided a definition for chemical restraint, compared to 51 studies that explicitly defined physical restraint [16]. The mental health sector has also found it challenging to define chemical restraint, acknowledging ‘use remains controversial with different understandings of what it is and its role in care’ [31]. It appears more education is needed around what chemical restraint entails in the aged care sector so that it is identified, and when proposed and/or used, practice accords with the requirements set out in the Aged Care Legislation [11] and the Aged Care Quality Standards [32].Evidence is mounting that legislation appears to be one of the most effective approaches to reduce restraint practice for older people receiving residential aged care. Countries that have introduced legislation in response to high rates of restraint, including the USA and Canada, have subsequently reported significant reductions in use [16,33,34]. One consistent finding in our study was the high level of support for the recent legislative amendments that have been introduced around restraint use in Australian residential aged care [11]. Although some participants were frustrated with the increased reporting associated with the new Legislation, they agreed that documentation and enhanced oversight was necessary to stamp out poor practice and reduce reliance on restraint use. Interestingly, although highly supportive of enhanced regulation, many participants admitted they did not know the specifics of the legislative amendments relating to restraint, pinpointing a need for additional information and training for staff, healthcare practitioners and informal caregivers.Another common theme raised in this research relates to the concept of ensuring a ‘restraint-free environment’. The 2012 DMT resources state: ‘with a restraint free approach, the use of any restraint must always be the last resort’ [18,19]. Some of the participants expressed the view that in literal terms, ‘restraint-free’ meant that restraint should never be used, as opposed to being permitted in certain situations. As with restraint, the meaning of a ‘restraint-free environment’ appears to vary in different settings and between countries. For instance, a group of researchers based in a hospital in the USA claimed they were able to achieve a ‘restraint-free environment’ in their delirium unit [35]. However, they specifically defined ‘restraint-free’ as meaning no physical restraint. Antipsychotic and benzodiazepine use was still permitted and there was no reference to the use or absence of environmental restraint (i.e., locked doors, keypads) [35]. Many of our participants felt that a completely restraint-free environment was unattainable especially when providers require gates to be locked or keypads used to prevent people with severe dementia leaving services unattended. Others stressed that it was important to acknowledge there were certain times when restraint was needed but instead there should be significantly more emphasis on ensuring use was appropriate and that legislative obligations were followed. To circumvent confusion and skepticism around the term ‘restraint-free environment’ it may be more appropriate to focus instead on using terms such as ‘minimizing restraint’ or ‘restraint as the last resort’.‘Low-level’ restraint was another practice over which many participants expressed concern, citing examples such as wedging a wheelchair under a table at mealtimes or placing a walker or care-bell out of reach. The use of these less obvious methods of restraint has also been reported by others [17,36,37]. An ethnological study conducted in Norway, consisting of a mixture of field observation and staff interviews, found that low-level restraint practices were often used to avoid using more overt forms of restraint. These low-level practices also allowed staff to ‘get the care work done’. Similar to some of our participants, the Norwegian researchers reported that many of the care staff were not aware that certain practices could be construed as restraint [36]. Training the staff in what restraint means within a human rights framework, detailing the practices it entails, and encouraging open discussion around restraint is vital to mitigate this issue.Participants based in community aged care reported additional challenges with regards to restraint. They reported that due to the ethos of ‘consumer directed care’ in home care, clients now control the types of care and services they receive [38]. This means that for many clients, clinical care is not provided as part of their home care package. Community care nurses and PCAs would often only visit a client’s home briefly so had limited knowledge of the medication they were taking or if they were subject to other forms of restraint. It should be stressed that the new legislative obligations on restraint [11] only apply to residential aged care providers, although the Aged Care Quality Standards [32], which apply in all federally-funded aged care settings, do require providers to minimize restraint and report on restraint as part of governance.Our health professionals reported that families, in general, were more involved with the care of their loved one when they were living in the community and stressed the need to work collaboratively with them. However, several of our community-based participants recounted experiences where their relatives with dementia were commenced on psychotropic medication for behavior control and they only found out after the medication had been administered. Similar experiences of chemical restraint use without informed consent were recounted in a 2019 Human Rights report [8] and at the Royal Commission [10]. Additional training for health care practitioners around the legal requirement for obtaining informed consent is urgently needed to address this issue. In a recent research study [2], also presented as evidence to the Royal Commission [10], rates of psychotropic use in the community were shown to increase markedly in the year prior to aged care admission, providing an indication that the issue of chemical restraint is not just confined to residential care. More research is needed to gauge the extent of restraint practice, both chemical and other forms, in community aged care, particularly given the rapid expansion of the Home Care sector. Further, the knowledge, experiences and opinions of relatives around restraint use in all aged care settings need to be investigated in greater depth as their voice is rarely captured [39].Community care nurses and PCAs also reported that they had little influence if the client or their family decided to use devices such as bedrails or low beds, both forms of restraint, as they were only responsible for practices that occurred while they were present providing care. Similar barriers were reported in a Dutch qualitative study of restraint use in community settings that concluded that informal caregivers, especially relatives, have a dominant role in the use of restraint [40]. They also reported that relatives were less aware of the harms associated with restraint use and had limited knowledge of alternative strategies. Due to their findings, the researchers have instituted a training program on restraint for informal caregivers, with promising results to date [40]. Similar education and training may be needed in Australia.There are several limitations associated with this study. First, we found it challenging recruiting participants based in community aged care. Although we approached several large community care providers in Tasmania, requesting names of potential staff to interview, volunteers were not forthcoming. Two participants from a large Home Care provider also withdrew on the day of their focus group. We theorize that this lack of recruitment was due to several factors, including staff shortages, limited experience with aged care research, and the emerging COVID-19 situation at the time. As there were fewer participants from the community compared with the sample from residential aged care, their experiences and opinions regarding restraint may not be representative of the community aged care sector overall. Second, we need to note that all participants were asked to read a 2012 resource on restraint [18,19] before being interviewed or involved in one of the focus groups. This pre-reading may have increased knowledge and influenced viewpoints around restraint use, meaning that their opinions and familiarity about this topic may not be representative of key stakeholders overall. A final limitation involved the online nature of the interviews and focus groups owing to restrictions imposed at the initial stages of the pandemic. We inevitably experienced technical difficulties with sound and vision with some participants, resulting in several interviews having to be rescheduled, and some conducted solely using mobile phones. Participants may not have been as willing to express their opinion due to the lack of direct personal engagement with the interviewer, and with the lack of interaction with other participants in focus groups. Despite these limitations, we felt that all participants were very accommodating of the situation and that rich and insightful data regarding restraint in aged care settings were obtained.This study was conducted to inform and update Government resources on restraint to align with legislative changes and contemporary best practice. The findings suggest that many nursing and care staff, health practitioners and relatives have a broad understanding of what restraint means. However, additional education is needed for these stakeholders on a more conceptual level about what restraint involves, particularly how to recognize and minimize low-level forms and chemical restraint. Likewise, resources need to contain information about the new Aged Care Legislation relating to restraint and what this means for providers and consumers of aged care services. Resources for providers advocating restraint-free environments were considered aspirational. Instead more practical guidance on what to do to prevent and minimize restraint, as well as what to do when restraint is judged appropriate was sought. Research is needed to investigate restraint practice in community aged care as use is largely unknown.Conceptualization, J.B., B.C.W., H.C.-P., K.L., K.S., A.P. and L.R.G.; methodology, J.B., H.C.-P. and L.R.G.; formal analysis, J.B., B.C.W. and L.R.G.; investigation, J.B. and C.C.H.S.; data curation, J.B.; writing—original draft preparation, J.B.; writing—review and editing, J.B., B.C.W., C.C.H.S., H.C.-P., K.L., K.S., A.P. and L.R.G.; supervision, J.B.; project administration, J.B. and B.C.W.; funding acquisition, J.B. All authors have read and agreed to the published version of the manuscript.This research was funded by the Australian Government, Aged Care Quality and Safety Commission, Updating the Decision-Making Tools for Supporting a Restraint Free Environment, grant number B0027625.The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Tasmanian Health and Medical Human Research Ethics Committee (ID 20044) on 16 April 2020.Written informed consent was obtained from all study participants before interviews were conducted and prior to focus group involvement.Full anonymized data supporting reported results are available by contacting the lead author via juanita.breen@utas.edu.au.We would like to acknowledge extensive administrative support provided by Karin Easton from the Wicking Dementia Research and Education Centre, the College of Health and Medicine at the University of Tasmania, Hobart, Tasmania. We would like to also expressly acknowledge the contribution of all those who were interviewed as part of this research study. Thank you for taking part and sharing your insight. Thank you also for your flexibility around the delays and with online technology associated with the COVID-19 pandemic.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.Codes and Sub-themes relating to participants’ understanding of restraint. n = number of definitions.Semi-structured interview guide.To start with can I ask you what the term ‘restraint’ means to you in the context of providing care to older people?In your opinion do most people you work with have the same definition—or do you think that other people have a different definition of what constitutes restraint?Do you think the practice of restraining people occurs commonly in aged care?I sent you a copy of the restraint decision making tool—did you have a read through it? What do you think about the current tool for restraint use?Could you provide an example of a case when restraint might be needed? Could you describe this case for me? (then ask questions to find out things such as what happened? What did the staff do? Was this effective? What else was tried? etc?)With this case, do you think you were supported by the organisation’s policy?Along a similar theme, what are your thoughts about the new restraint amendments to the Aged care Legislation?These are all the questions I have listed—but would you like to add any other comments about this topic?Semi-structured interview participant demographic information.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11023.txt

@@ -0,0 +1 @@
+This study examined the effects of a non-caffeinated energy drink (ED) that contained calamansi juice, glucose, and taurine on 3-km running performance and recovery. Eleven NCAA Division I middle-distance runners (20.8 ± 1.5 years old) were randomly assigned to consume either the ED or a placebo drink 60 min before 3-km running on a 400-m official track. Performance time and speed were recorded every 500-m interval. Recovery blood lactate concentration (BLC), systolic (SBP), diastolic blood pressure (DBP), and heart rate (HR) were measured at baseline, 60-min after ingesting the drinks, and post-running measurements were performed at 1-min, 5-min, and 10-min. Repeated analysis of variance and paired t-test were applied to examine the effects of time, trials, and their interaction on performance and recovery. Statistical significance was set a priori at p < 0.05. No significant difference was observed in performance time and speed between trials (p < 0.05). No interaction effect was found on performance time, speed, recovery BLC, DBP, and HR (p < 0.05). However, an interaction effect for trial by time was observed on SBP (p = 0.01). Recovery SBP continues to decrease from 5-min to 10-min in the ED trial (∆ = −13.9 mmHg) and slightly increased in the placebo trial (∆ = 1.1 mmHg). This study suggests that acute consumption of a calamansi-containing ED can positively impact the SBP recovery but not running performance. Further studies are needed to examine the acute and chronic effects of this ED on exercise performance and recovery among different populations.Energy drinks (EDs) are often consumed as nutritional ergogenic aids to improve exercise performance and recovery [1]. These benefits are attributed to the sole and/or combined effects of EDs’ ingredients. Particularly, caffeine alone or when combined with other ingredients such as taurine, carbohydrate, and vitamins (e.g., B group, C) has been suggested to be the primary component that enhances exercise performance [2,3]. Yet, recent evidence indicates that caffeinated EDs not only can delay recovery [4,5] but can also adversely impact various health outcomes such as toxicity, insomnia, cardiac arrhythmia, and fibrillation [6,7]. Due to these adverse impacts, advocates to ban EDs sale, especially caffeinated ones, has already been established [8].An alternative strategy is to replace caffeine with other substances that may provide the intended ergogenic influence on exercise performance and recovery without similar adverse impacts. Specifically, Taurine is an amino acid that exerts favorable effects on insulin activity, glycemic regulation, lactate concentration, and anti-inflammation [9,10,11]. Administration of Taurine was found to have a favorable influence on exercise-induced fatigue and exercise performance and recovery [12,13]. Similarly, glucose is the basic form of carbohydrate that can enhance the production of the molecular unit of currency (adenosine triphosphate, ATP), reduced exercise-induced fatigue, and improve exercise performance [14,15]. In fact, both Taurine and glucose have commonly been added to EDs [1] due to their beneficial effects on exercise performance and recovery. Thus, Taurine and glucose may be optimal surrogates to caffeine in EDs.A crucial determinant that may limit exercise performance and recovery and accelerate exercise-induced fatigue even if EDs were consumed is reactive oxygen species (ROX) [16,17]. The detrimental impacts of ROX are usually observed following high intensity or prolonged exercise [16]. These harmful effects include decreasing muscle force production and damaging lipids and proteins in the contracting myocytes, all of which can impact exercise performance and recovery [18]. Antioxidants such as dietary nitrate supplementations and vitamins have been proposed to mitigate the unfavorable effects of ROX during and following exercise [17]. Of particular interest, vitamin C is one suggested antioxidant to blunt biological, oxidative changes [17], which could potentially delay exercise-induced fatigue and enhance exercise recovery. Moreover, vitamin C can enhance carnitine biosynthesis and bioavailability [19], which is known for its essential role in transporting fatty acid into mitochondria for ATP production by B-oxidation [20]. Thus, vitamin C consumption may also contribute to improving exercise performance via promoting fatty acid oxidation. Uniquely, Calamansi is a citrus fruit that grows abundantly in East Asia and is rich with antioxidant agents particularly vitamin C. Altogether, the combination of Taurine, glucose, and vitamin C as a form of ED may result in a beneficial influence on exercise performance and recovery. Vitamin C can be extracted from natural sources such as citrus fruits.To the best of our knowledge, only one study examined the ergogenic effect of an ED that contained Taurine, glucose, and calamansi juice on exercise performance and recovery in female basketball players (n = 6; 21.5 ± 1.9 years old) [21]. In a randomized controlled trial design, the authors have evaluated the influence of the ED versus placebo drink on muscular power, agility, and anaerobic using vertical jump, 10 m × 5 shuttle run, Wingate tests, respectively. Their findings suggested that the ED resulted in higher muscular power and faster fuel replenishment compared to the placebo trial. These improvements may be explained by quick absorption of the ingested Taurine and glucose which can enhance insulin activity, maintain blood glucose concentration, and spare endogenous glycogen stores, respectively [11,22]. Yet, the efficacy of such EDs on endurance performance and recovery remains to be explored.Therefore, the purpose of this study was to examine the acute effects of ED that contained calamansi juice, taurine, and glucose on 3-km running performance and recovery in NCAA division I middle-distance runners. It was hypothesized that the ED would improve 3km running performance and physiological and metabolic recovery including heart rate (HR), blood pressure (BP), and blood lactate concentration (BLC) compared to a placebo drink.This study was conducted in a randomized, double-blinded, and crossover design. All participants visited the laboratory on two different occasions at the same time of the day, separated by a 7-day washout period to ensure the carry-out effects were eliminated. All visits took place between mornings and afternoons. During visit one, participants underwent baseline measurements including body height and mass, body mass index, body fat percentage, resting HR, BP, and BLC. Then, participants were randomly assigned to the ED or placebo drink trial. The assigned drink was provided to participants who were instructed to consume it 60 min prior to their running start. Afterward, participants performed 3-km running on a 400-m official, outdoor track. Performance time and speed were recorded for the ED and placebo drink trials. Recovery HR, BP, and BLC were measured at baseline, 60-min after the ingestion of the ED/placebo drink, 1-min, 5-min, and 10-min post-running (Figure 1). After a 7-day washout period, participants performed the same experimental procedure as their first trial with consuming the ED or placebo drink upon their supplementation at the first trial. During the washout period, participants were instructed to carry on their normal activities and training.Eleven collegiate distance runners who were involved in the National Collegiate Athletic Association (NCAA) Division I voluntarily participated in this study. All participants were engaged in their regular training (twice/day for five days a week). The inclusion criteria of this study were: (a) collegiate NCAA Division I male and female distance runners, (b) had no musculoskeletal injuries in the last three months, (c) were not taking any vitamin supplementation such as vitamin A, B, C, D, E for the last three months, and (d) regularly participated in track and field training. In accordance with the seventh version of the Declaration of Helsinki ethical principles (2013), all eligible participants were provided with the study information including the purpose of the study, study procedure, possible benefits, and risks. All participants provided informed consent. This study was approved by the institutional review board (IRB) at the University of Louisiana at Monroe.Standing body height and weight were measured with a standardized balance beam scale (DetectoTM, Model 439, Webb City, MO, USA) to nearest 0.1 cm and 0.1 kg, respectively. The measurements were performed while participants wore light cloth with bare feet. Body mass index was computed by dividing weight (kg) by squared height (m2). Body fat percentage was estimated from skinfold thickness using Lange skinfold caliper (FAB12-1110, Beta technology, Santa Cruz, CA, USA). The three-site skinfold method was applied. Chest, abdominal, and thigh skinfold thickness were measured for males, and triceps, suprailliac, and thigh for females. The measurements were completed twice from the right side of each participant. The criterion for the agreement between the two measures was within 5%. Thereafter, body density was calculated by Jackson and Pollock’s equation, and body fat percentage was computed by Siri’s equation, respectively [23,24].Seated BLC was evaluated five times as follows: at baseline, 60-min after the ingestion of the ED/placebo drink, post-running at 1-min, 5-min, and 10-min. These measurements were performed by using a hand-held device (NOVA Lactate Plus, Biomedical, Waltham, MA, USA) which was previously validated [25]. The device uses 0.7 µL of a whole blood sample to determine plasma lactate concentration in the range of 0.3–25.6 mmol/L within 15 s. To perform the measurement, participants’ middle or ring finger was punctured with a disposable Lancet BD puncture. Then, the puncture site was gently pressed, and the blood sample was collected. The first drop of blood was wiped off with a sterile swab, and the second drop of blood was used for lactate analysis.Seated systolic (SBP) and diastolic (DBP) BP were measured five times as follows: at baseline, 60-min after the ingestion of the ED/placebo drink, post-running at 1-min, 5-min, and 10-min. These measurements were performed by using an electronic sphygmomanometer (American Diagnostic Corporation, Model Advantage 6021, Hauppauge, NY, USA). When seated, participants were instructed to have their back supported, feet on the floor, and arms supported at heart level. Seated HR was also measured five times as follows; at baseline, 60-min after the ingestion of the ED/placebo drink, post-running at 1-min, 5-min, and 10-min. These measurements were performed by using a Polar H7 strap monitor (Polar H7, Polar Electro Oy, Kempele, Finland), a smartphone (iPhone SE, San Francisco, CA, USA), and a phone app (MOTIFIT, MotiFIT Fitness Inc., Moncton, NB, Canada). Participants wore a Polar strap on their chest which was synchronized to the iPhone via Bluetooth. Then, the iPhone was tightly placed on the participants’ right arm to allow HR data to be transmitted to the MOTFIT app. The collected HR data were exported to a computer software program (Microsoft Excel, Microsoft, Redmond, WA, USA) for analysis.Participants performed two 3-km running trials. Running was performed on a 400-m official, outdoor track. One week prior to the first trial, participants performed 3-km running familiarization trial. On experimental days, participants underwent baseline measurements and consumed the ED/placebo drink at our laboratory. Participants were instructed to arrive at the running track 30 min prior to running time. Then, they performed their usual warm-up exercises including static, dynamic stretching, and potentiating exercises. All participants were instructed to wear the same uniform and shoes during both the ED and placebo drink trials to minimize external influences. Coaches provided maximal encouragement to all participants to achieve their best performance during both trials. Weather conditions including temperature, humidity, and wind speed were recorded on both trial days. Average temperature, humidity, and wind speed were 75.18 ± 8.80 F° vs. 75.18 ± 8.66 F°, 59.1 ± 8.54% vs. 64.1 ± 13.38%, and 10.1 ± 4.25 m/s vs. 10.4 ± 3.75 m/s on the ED and placebo drink trial days, respectively.Participants were instructed not to drink any caffeinated drinks at least one week prior to their 1st trial and during a 7-day wash-out period. The ED (240 mL) contained 8% Calamansi juice (which is rich with vitamin C), 0.1% additive vitamin C, 10% glucose, 0.8% taurine, and 0.4% branched-chain amino acid. The placebo drink (240 mL) contained 0.02% sucralose (mimicked the sweetness of the ED), 0.05% malic acid (mimicked the taste of the ED), and 0.03% oleoresin turmeric (mimicked the color of the ED). The volume, texture, and appearance were similar across both the ED and placebo drink. To ensure the randomization and crossover design, two small papers were folded after writing the following on one side: “Red” (indicates the ED) or “Blue” (indicates the placebo drink). Then, participants picked up one of the two folded papers and were assigned to their trials accordingly. Participants, coaches, and examiners were not aware of which word matched with which drink, except one examiner who was not directly involved in the measurements. Therefore, the randomized double-blind design was well controlled. Participants consumed the ED/placebo drink 60 min prior to the 3-km running trials. Once the ED/placebo drink was provided to participants, they were instructed to drink it immediately in front of a blinded examiner who ensured that participants drank the entire drink. The time interval (i.e., 60 min) between the consumption of the drink and the start of 3-km running trials was chosen according to previous studies [26,27]. Lastly, the ED had been tested by a certified company (OATC Inc., Seoul, Korea) and was officially certified as a non-caffeinated ED by the Ministry of Food and Drug Safety in South Korea (MFDS FID-2016042480).A power analysis using G*Power program 3.1.9.2 (Heinrich-Heine-Universität, Dusseldorf, Germany) was used based on the previous study [28] to determine the sample size required to detect the difference between two dependent means (matched pairs). With an estimated power of 0.95, alpha of 0.05, a total sample of 10 in each group was required to detect an effect size of 1.3 (Actual power: 0.95). SPSS (version 25, SPSS Inc, Chicago, IL, USA) was used to perform the statistical analysis. All data were expressed as means and standard deviations. A statistical descriptive test was performed on height, weight, BMI, and body fat percentage. Kolmogorov–Smirnov test was utilized for normality evaluation of all analyzed variables. A paired t-test was performed to compare average performance time and speed between the two trials. A 2 × 6 (conditions × time points) repeated measures ANOVA was performed to examine interaction effects for speed by time of the 3-km running trials. A 2 × 5 (conditions × time points) repeated measure ANOVA analysis was also conducted to evaluate interaction effects for recovery variables including BLC, SBP, DBP, and HR. Bonferroni post-hoc test was applied when significant interaction or main effects were detected. Effect sizes were calculated as partial eta-squared (η2p; small ≥ 0.01, medium ≥ 0.06, large ≥ 0.14) values within measures ANOVA and Cohen’s d (small ≥ 0.02, medium ≥ 0.05, large ≥ 0.08) for paired t-test. Statistical significance was set a priori at p < 0.05.The characteristics of the participants are presented in Table 1. Seven participants reported themselves as males. Participants were 20.8 ± 1.5 years old with normal SBP (116.7 ± 8.2 mmHg), DPB (70.9 ± 12.2 mmHg), BMI (20.0 ± 2.5 kg/m2), and body fat percentage (9.3 ± 5.1%). All analyzed variables were normally distributed. There was no significant baseline difference in BLC, SBP, DBP, and HR between the trials. There were a few missing data from recovery outcomes including blood lactate and blood pressure (single data at recovery 1 min) due to unexpected device errors. Therefore, only completed data from all time points were included in the analysis.Figure 2 displays performance time and speed in both trials. The paired t-test showed no significant difference between the two trials in both average performance time and speed (t = 0.681; p = 0.511; Cohen’s d = 0.044 and t = 0.832; p = 0.425; Cohen’s d = 0.023, respectively). The average performance time for the placebo and ED trials were 658.96 ± 94.53 s and 655.04 ± 90.42 s, respectively. The average performance speed for the placebo and ED trials were 2.34 ± 0.30 s and 2.32 ± 0.30 s, respectively. The repeated ANOVA showed no interaction for trial by time effect on performance time and speed (F = 1.455; p = 0.221; η2p = 0.127 and F = 0.815; p = 0.422, η2p = 0.075, respectively). There also was no trial effect on both performance time and speed (F = 0.708; p = 0.420; η2p = 0.066 and F = 0.692; p = 0.425; η2p = 0.065, respectively). However, there was a main time effect on both performance time and speed (F = 450.69; p < 0.001; η2p = 0.978 and F = 65.02; p < 0.001; η2p = 0.867, respectively).Table 2 presents the changes of BLC, BP, and HR during recovery. There was no main trial or interaction effect for BLC (F = 0.117; p = 0.741 and F = 0.551; p = 0.700, respectively). However, there was a main time effect for BLC (F = 136.18; p < 0.001). Though there was no main trial effect (F = 1.119; p = 0.325) for SBP, there were main time and interaction effects (F = 28.899; p < 0.001 and F = 4.117; p = 0.010, respectively). Comparing 5-min versus 10-min post-running, SBP greatly decreased in the ED (121.2 ± 17.07 vs. 107.3 ± 13.06 mmHg) compared to the placebo trial (113.1 ± 13.48 vs. 114.2 ± 10.49 mmHg) (Figure 3). In addition, there was no main trial or interaction effect for DBP (F = 0.572; p = 0.469 and F = 1.544; p = 0.210, respectively); yet there was a main time effect for DBP (F = 3.65; p < 0.037). Lastly, there was no main trial or interaction effect for HR (F = 0.884; p = 0.372 and F = 0.722; p = 0.582, respectively). However, there was a main time effect for HR (F = 95.83; p < 0.001).This study examined the effect of a non-caffeinated ED that contained calamansi juice, taurine, and glucose on 3-km performance and recovery in NCAA I division distance runners. Our results revealed no significant effects of the ED on 3-km running performance (i.e., time and speed). Though the ED did not improve BLC, DBP, and HR, it significantly affected SBP reduction during recovery compared to the placebo trial.To the best of our knowledge, this study was the first to evaluate the effect of a non-caffeinated ED that contained calamansi juice, taurine, and glucose on 3-km running performance. Though the ingredients of the ED such as glucose and taurine are believed to promote energy production which may enhance exercise performance, we did not detect improvements on any of the running performance variables. A recent meta-analysis revealed that taurine plays an essential role in EDs that induce exercise performance improvement [29]. A previous study revealed that consuming ED that contained caffeine, taurine, and glucose improved 5-km running time in healthy, recreational runners [27]. Prins et al. ED contained 80 mg caffeine with unknown concentrations of taurine and glucose. In our current study, the ED consisted of 19.2 mL taurine and 24 mL glucose. Though the minimal, beneficial concentrations of taurine and glucose in EDs remained to be examined, it is possible that these inconstant results be due to different concentrations of EDs’ ingredients. Additionally, the caffeine substance, sample characteristics, and running distance may also have roles in these equivocal findings. However, one existing study examined the acute effect of a non-caffeinated ED that contained vitamin C, taurine, and glucose on muscular (vertical jump) and anaerobic power following a simulated basketball game (SBG) in college basketball players [21]. The ED was provided to participants immediately after SBG. Then, after 20 min, the study assessments took part. The researchers demonstrated that the consumption of the ED favorably influenced vertical jump and anaerobic power comparing to a placebo drink trial. In contrast, both running time and speed (aerobic endurance) did not improve following the consumption of a non-caffeinated ED containing calamansi juice, taurine, and glucose in the current study. The ED was provided to participants 60 min prior to running. It is difficult to directly compare the efficacy of the ED on exercise performance between these two studies. Because these two studies had different study protocols such as dependent variable (anaerobic vs. aerobic performance) and subjects (basketball player vs. long-distance runner). However, it is possible that a dose–response of this ED on performance can be varied by different populations, ingestion time points, and performance variables; thus, future studies may need to be examined to find underlying mechanisms of this ED on physiological responses.Generally, glucose is considered one of the main ingredients in EDs [30]. Glucose is added to EDs due to its properties to promote the production of a molecular unit of currency (ATP) and the ability to enhance exercise performance and recovery [31,32]. Many studies demonstrated enhanced carbohydrate availability following pre-exercise consumption of glucose which eventually enhances exercise performance [33,34,35]; however, our results showed that the pre-exercise consumption of ED that contained glucose (24 mL) did not boost running performance. A few factors may explain these inconsistent results including resting blood glucose level and concentration of consumed glucose. Generally, 30–60 g/h glucose is recommended to be consumed during endurance performance to maintain the plasma glucose level in athletes [34]. However, it is important to consider the glucose recommendation in the middle-distance running event. This is because the effectiveness of the additional glucose intake could be minor if athletes’ glycogen status is sufficient prior to the event.This study revealed that the pre-exercise consumption of the non-caffeinated ED did not influence recovery BLC or HR. These results were consistent with the literature where many studies showed no effects of pre-exercise consumption of EDs on BLC and HR [4,27,36,37]. However, a previous study reported that similar non-caffeinated ED decreased BLC and maintained blood glucose level suggesting accelerated replenishment of fuel’s deficiency following a simulated basketball game [21]. These findings signify that the pre-exercise consumption of the ED may benefit glycemic outcomes during recovery. However, the present study did not measure other glycemic outcomes thus the hypothesis should be explored in future studies. Noteworthy, it is possible that inequivalent effects between the previous study and ours are due to differences in the nature of exercises performed (i.e., basketball that mostly relies on anaerobic capacity vs. 3-km running which mostly relies on aerobic capacity). Therefore, further research investigating the influence of these non-caffeinated EDs on aerobic vs. anaerobic performance is warranted.Although the BLC and HR were not influenced by the ED in this study, an interesting result was observed during the recovery period. As was expected, both SBP and DBP significantly decreased following both trials through exercise-induced hypotension (PEH); however, the greater reduction was observed on SBP in the ED trial. PEH is a fall in BP that occurs following exercise (even after a single bout) and can last up to ~13 h [38,39]. Even though these reductions were expected following both trials, it was also assumed that the ED would cause greater BP reductions due to its ingredients. The ED contained vitamin C and taurine which serve as antioxidant/inflammatory agents [9,40,41]. Indeed, several studies revealed that the consumption of both vitamin C and taurine can significantly decrease BP [42,43,44]. Vitamin C functions in vascular endothelium and smooth muscle and increases vascular dilation and eventually BP reduction [45]. On the other hand, the consumption of taurine can increase endothelium-dependent and independent vasodilation [44,46]. Thus, it would be assumed the antioxidant ingredients in the ED (calamansi juice) boosted PEH. Furthermore, a previous study suggested that sustained post-exercise vasodilation, which always accompanies PEH would benefit muscle glucose uptake especially in athletes [42]. Taken together, a calamansi-containing ED may provide a benefit on SBP reduction during recovery; however, it is premature to confirm this hypothesis. Further studies are required to confirm the role of this ED on BP during recovery.The strength of this study was systemically designed as randomized, double-blind, crossover and placebo-controlled trials. Additionally, the study was applied in a practical manner where it was performed at an official, outdoor track where running competitions are held. However, small sample size (n = 11) may be a limitation of the study. Additionally, the enrolled participants in our current study were NCAA Division I middle-distance runners. As such, the applicability of our findings is currently limited to this elite group of athletes, and the effects of this ED on less active or sedentary populations remain to be studied. Importantly, though the ED favorably influenced blood pressure, its effects, if any, on running performance and time were minor. Not controlling for the menstrual status also limits the generalizability of the current findings. Lastly, our study examined the acute effects only, and the chronic effects of this ED on exercise performance and recovery warrant further investigation.Our study found that acute consumption of a non-caffeinated ED that contains calamansi juice, taurine, and glucose does not improve exercise performance and BLC, DBP, or HR recovery. However, this drink may be effective on SBP recovery, especially in distance runners. Nevertheless, the long-term effects of this ED remain unknown. Further studies are needed to examine the acute and chronic effects of this calamansi contained ED on exercise performance and recovery among different populations. In short, the findings of this study indicate that EDs that contain calamansi juice, taurine, and glucose may be effective to improve recovery following aerobic exercises. Yet, other non-caffeinated EDs that may enhance aerobic performance remain to be explored.Conceptualization, A.B.A., J.H. and H.C.J.; methodology, A.B.A., J.H. and H.C.J.; investigation, A.B.A., J.H. and H.C.J.; data curation, A.B.A. and H.C.J.; writing—original draft preparation, A.B.A. and H.C.J.; writing—review and editing, J.-M.L., M.-W.S. and D.Y.; supervision, H.C.J. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of University of Louisiana at Monroe (ULMIRB-733-2016).Informed consent was obtained from all subjects involved in the study.The authors would like to thank participants for their voluntary participation in the study.The authors declare no conflict of interest.Study procedure BLC; blood lactate concentration, BP; blood pressure, HR; heart rate.Running performance time and speed. Note. m/s: meter per second.The changes of systolic blood pressure. The square indicates the placebo trial while the circle indicates the energy drink trial.* p < 0.05, indicates a significant change from recovery 5-min to 10 min only in the energy drink trial. mmHg: millimeters of mercury, SBP: systolic blood pressure.Participant Characteristics (n = 11).BP: blood pressure, cm: centimeter, kg/m2: kilogram per meter squared, mm: millimeter, mmHg: millimeters of mercury, mmol/L: millimoles per liter.Changes in blood lactate concentration, blood pressure, and heart rate during recovery.* p < 0.05; indicate significant time effects, + p < 0.05; indicate significant interaction effect between group by time; BLC: blood lactate concentration; BP: blood pressure; E: energy drink; G: group; G 237 × T: group × trial; mmHg: millimeters of mercury; mmol/L: millimoles per liter; P: placebo; T: time.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11024.txt

@@ -0,0 +1,15 @@
+Working towards sustainable population development is an important part of carbon mitigation efforts, and decoupling carbon emissions from population development has great significance for carbon mitigation. Based on the construction of a comprehensive population development index (PDI), this study adopts a decoupling model to explore the dependence between carbon emissions and PDI across 30 Chinese provinces from 2001 to 2017. Then, the stochastic impacts by regression on population, affluence and technology (STIRPAT) model is used to investigate the impact of population factors on carbon emissions. The results show that the decoupling relationship between carbon emissions and PDI has experienced a transformation from expansive negative coupling to expansive coupling and then to weak decoupling at the national level, while some provinces have experienced the same evolutionary process, but the decoupling state in most provinces is not ideal. Sending talent to western provinces and developing low-carbon supporting industries will accelerate carbon decoupling. At the national level, incorporating environmental protection into the existing education system as part of classroom teaching could contribute to carbon decoupling.With the continuous growth of population and carbon emissions in developing countries, the issue of sustainable population development has attracted attention. In the face of the complex climate change situation, implementing carbon mitigation and sustainable population development strategies to reduce the climate risks caused by carbon emissions has become the top priority of the Chinese government [1]. To realize the coordinated development of the economy, society, resources and the environment, China, the country with the largest population, has experienced the transformation from the family planning policy to the three-child policy and the population sustainable development policy. During the demographic transition period, changes in population dynamics and development patterns will undoubtedly affect China’s energy use and the resulting carbon emissions [2]. Meanwhile, as the world’s largest carbon emitter, China is considered a key player in the international effort to tackle climate change [3]. In order to reduce carbon emissions, the Chinese government has also proposed the goal of peaking carbon emissions by 2030 and achieving carbon neutrality by 2060 [4]. To achieve the goal of dual carbon emission reduction firstly means that carbon emissions can no longer increase with the social development; that is to say, carbon emission decoupling is the first step to achieve carbon neutrality [5]. Therefore, evaluating the degree of carbon emission decoupling is of great significance for promoting carbon emission reduction.Decoupling research originally focused on how to balance the tension between economic growth and environmental pollution [6]. Although some studies have examined this link at the national, industry or sector level, the decoupling between carbon emissions and population development is still unclear because it is rarely discussed. Quantifying the decoupling states of carbon emissions from a multiperspective can guide more detailed emission reduction strategies [7]. To achieve population sustainability and low-carbon development, it is of great importance for the government to understand the decoupling state between carbon emissions and population factors and the influencing mechanism between them [8,9].Moreover, there are significant disparities regarding energy demand, population size, residential income, urban–rural structure and population quality in Chinese provinces, which may lead to significant differences in the interaction between carbon emissions and population development [10]. An objective evaluation of regional population development will provide an important premise for exploring the decoupling the relationship between carbon emissions and population development. Therefore, China and its 30 provinces for which data are available are selected for this study, which may contribute to a better understanding of the dynamics of carbon decoupling.Investigating the dynamics of the decoupling of carbon emissions from population development and making recommendations that might facilitate the decoupling of carbon emissions are key objectives of this study. The specific objectives are shown below:(1)Develop a comprehensive indicator to reflect the multifaceted aspects of population development;(2)Explore the decoupling relationship between carbon emissions and population development;(3)Identify the driving force of population development on carbon emissions to promote carbon decoupling.Develop a comprehensive indicator to reflect the multifaceted aspects of population development;Explore the decoupling relationship between carbon emissions and population development;Identify the driving force of population development on carbon emissions to promote carbon decoupling.This study can make the contribution to the existing knowledge from the followings: Firstly, as for the research perspective, this study focuses on the dependence of carbon emissions on population development, while previous studies focus more on decoupling from the perspective of economic development. Since the population effect has no less of an impact on carbon emissions than economic development [11,12], this study focuses on the decoupling of carbon emissions from population, which is a good extension of the existing studies on the decoupling of carbon emissions and may help to promote managers to focus on high-quality transition and the sustainable development of population, rather than only sustainable development at the economic level. Secondly, regarding the research scale, this study investigates the decoupling state between population development and carbon emissions at the national and provincial level. China is the world’s most populous country, with a population of more than 1.4 billion. Of the 30 provinces for which data are available, 20 have a population of more than 30 million in 2017 [13]. As an important contributor to carbon emissions, the study on the decoupling of China’s carbon emissions will not only help China’s energy conservation and emission reduction process, but also provides a reference for sustainable population development in other populous countries, such as India and other countries, and plays an important role in promoting climate change mitigation worldwide. Finally, in terms of the research method, based on decoupling analysis and the stochastic impacts by regression on population, affluence and technology (STIRPAT) model, this study explores the impact of a series of population development factors on carbon emissions, including population size, urban–rural structure, employment structure, age structure, population quality and population affluence, which is a supplement and extension of the existing research on the driving mechanism of carbon emissions. This will help decision makers formulate measures to improve population development and reduce emissions.Researchers are interested in the study of carbon emission drivers in developing countries. Most of the current studies focus on carbon emission decomposition or decoupling from the perspective of energy consumption, economic development, technology progress and other factors [14,15,16,17]. To explore the interaction mechanism between these factors and carbon emissions, the Tapio decoupling model has proved to be an effective method and has been widely used. Shuai et al. investigated the global decoupling of economic growth and carbon emissions and concluded that high-incomes countries are more likely to have the expected decoupling relationship [18]. Zhang and Da analyzed the decoupling relationship between carbon emissions and economic growth in China, and the results indicated that economic growth is the primary driver of carbon emissions growth [19]. Song et al. utilized the decoupling model to evaluate the decoupling status and dynamic trends of carbon dioxide emissions at the provincial level in China [20]. The decoupling researches are also focused on the sector level, including construction, industry [21,22], product [23],transportation [24,25] and agriculture [26].Economic output has become the main consideration in the study of carbon decoupling in China [27], and this consideration is also reflected in the research outside China [28]. However, the dependence of carbon emissions on other factors, such as population, has received little attention, and only a few studies have assessed the decoupling of carbon emissions from population-related factors. For example, Ma et al. explored the relationship between household carbon emissions and economic growth based on decoupling indicators, and concluded that household carbon dioxide emissions were in a weak decoupling state on the whole, and changes in CO2 emissions caused by population growth and economic growth were in a weak decoupling state and expansionary decoupling state, respectively [29]. The current research has also consistently found that population factors (i.e., size, growth and other parameters) are strongly correlated with carbon emissions [30,31,32,33]. Successful environmental social science research projects, such as the Infrastructure Project Assessment Tool (IPAT), place great emphasis on the relationship between population and the environment [34]. Therefore, in the transition period of population development, it is necessary to systematically understand the dependence of carbon emissions on population factors.To clarify the interaction mechanism between population and carbon emissions, researchers examined the effect of various population factors on carbon emissions. Wang, et al. explored the impact of population size, per capita consumption, urbanization and an aging population on carbon emissions [11]. Zhu and Peng studied the impact of population change on China’s carbon emissions, and they revealed that consumption levels and population structure significantly affect carbon emissions [35]. Jorgenson and Clark argued that population size is positively correlated with carbon emissions [36]. More and more research argued that population size, population structure, quality and other indicators should be considered in the economy-environment model to fully reflect the impact of population factors on carbon emissions [37,38]. In some research, carbon emissions are related to population aging, and the working-age population is also considered an important indicator of future carbon emission mitigation [39,40]. Li et al. found that the relationship between the aging structure and carbon emissions in China can be described by an inverted U-shaped curve [41]. In some developing countries, population quality also has a significant impact on carbon emissions [42]. A cross-nation study adds to the discussion on the link between population size and other demographic factors and pollution, arguing that population increases are matched by proportional increases in emissions while a higher urbanization rate and lower average household size increase emissions [43].In terms of population development characteristics, almost all of the important population factors, including population size, population structure, population quality and population distribution, are constantly changing, which have a complex and changeable impact on carbon emissions [35]. Generally, the impact of population on carbon emissions is uncertain due to the varied population features in different regions [44]. It is certain that if the population factors are measured by population scale, it cannot fully reflect the population impact on carbon emissions. However, this is what most studies have done when exploring carbon emissions drivers by multiple regressions in the economy-environment model. The assumption behind this treatment is that each individual in a population shares the same production and consumption behavior, but this assumption may be inaccurate and misleading [45]. Thus, an integral description that utilizes the multidimensional characteristics of population development is required, which will help to understand the effect of population on carbon emissions.In summary, the impact of various population factors, including population size, population growth rate, age structure, urban–rural structure, employment structure, population quality, consumption structure and per capita GDP, on carbon emissions have been studied. Although some progress has been made, there are still some limitations, which highlight the following research gaps: (1) various population factors are simply used, and without an integral indicator to reflect multidimensional population development characteristics; (2) most studies on carbon decoupling have been typically conducted at a sectoral or country level and measured by economic outputs. This makes the relationship between population development and carbon emissions unclear.To address the research gap, this study: (1) develops a population development index (PDI) to evaluate multidimensional population development; (2) establishes a decoupling model to investigate the decoupling between carbon emissions and the PDI in 30 Chinese provinces; (3) investigates the impact of various population factors on carbon emissions and explores policy suggestions to promote the decoupling of carbon emissions from PDI.This paper is organized as follows. Section 3 describes the research methods. The study areas and the data sources are presented in Section 4, and Section 5 presents the results and discussion, which is followed by the final conclusions and policy implications.The PDI is developed through the following three steps: (1) select indicators that represent the level of population development; (2) determine the weight of each indicator; and (3) calculate the PDI.The evaluation of the population development level usually includes a series of complex index systems, including the total population index, the population structure index, population quality index, population economic activity index and these indexes objectively reflect the population development level [40]. Some important population studies provide reference for the establishment of a population development index [46,47,48,49,50]; population size, population growth, population quality, population living standard and population age structure are incorporated into the PDI in this study. The indicators that make up the PDI are shown in Table 1.Population size: As a population factor that has a significant impact on carbon, it is often included in the IPAT/STIRPAT model to investigate the environmental pressure. In terms of population development, the more people there are, the more social wealth can be created. Population growth promotes the development of the service industry and industrial transformation [51], and also has a long-term positive impact on economic development [52]. In turn, economic growth promotes population development [53]. Under the current three-child policy in China, population size can be regarded as a positive indicator to promote population development. This study uses the total population and population growth rate to measure the population size.Population structure: in this study, population structure is considered by age structure, urban–rural structure and employment structure.In terms of age structure, it reflects the structural health of the population. The population aging trends will impose challenges for China’s sustainable development on the supply and demand sides in the long term [54]. Governments around the world are also urgently formulating policies to deal with the phenomenon of population aging [49,55]. Therefore, from the perspective of long-term population development, population aging is considered as a negative indicator in the PDI establishment.Urban–rural structure reflects the distribution and migration of population in the process of urbanization. It is generally argued that the population distribution in urban areas is more concentrated, and the population density is higher than that in rural areas. The higher the urbanization level, the stronger the regional development. The level of population urbanization plays a very important role in promoting population progress and development [56], so the urban–rural structure is considered as a positive indicator of the PDI.Employment structure: It is generally believed that the greater the number of people engaged in the secondary and tertiary industries, the higher the degree of development of industrial and social services. In terms of China’s current stage of development, the employment population in the secondary industry and tertiary industry will be considered as positive indicators of the PDI because both industry and service industry contribute significantly to economic and population development [57].Population quality: population quality often represents the civilization construction level of a country or region, and is usually measured by the educational level of the population, which plays a very important role in promoting population development [58].Population wealth: population wealth reflects the ability of the population to create wealth. The higher the per capita wealth is, the higher the people’s living standard and the higher the consumption level are. In the context of rapid economic development, the level of population wealth has further promoted the shift of population development to high quality [59].All of the indicators are standardized to make different variables comparable by using the following formulas [60]:(1)y+ij=(xij−xijmin)/(xijmax−xijmin)
+          (2)y−ij=(xijmax−xij)/(xijmax−xijmin)
+          where y+ij represents the positive indicator; y−ij represents the negative indicator; xij represents the value of indicator j in province i; and xijmax and xijmin indicate the maximum and minimum value of the indicator j, respectively. Then, the entropy weight calculation is used to determine the importance of each indicator:Firstly, to calculate the sample indicator weight:(3)pij=xij/∑i=1nxij,Secondly, to calculate the entropy of indicator j:(4)ej=−k∑i=1npij×lnpij,
+          (5)k=1/ln nThirdly, to calculate the utility value of each indicator:(6)dj=1−ejFinally, to calculate the indicator weight:(7)wj=dj/∑j=1ndj,
+          where pij represents the share of province i on indicator j; ej is the entropy of indicator j; n is the number of the indicator; dj is the utility value of each indicator.The linear weighted sum method is commonly used for evaluating the performance of a system which consists of multiple dimensions of indicators. By using this method, the performance value of the PDI in province i, can be calculated as follows:(8)PDIi=∑j=1nwj×yij,
+          where wj is the PDI weight of indicator j.More than 95% of carbon emissions come from energy consumption [61], according to the IPCC National Greenhouse Gas Inventories and energy consumption [62]. The carbon emissions is calculated by following formula:(9)C=∑e=1mEefeke4412,
+        where C represents the emissions; m is the number of the energy type; Ee is the consumption of fossil fuel e; fe indicates the standard coal conversion factor of fossil fuel e; ke is the carbon emission factor for fossil fuel e [62,63]; and 44/12 is the molecular transition from carbon dioxide to carbon.The Tapio model proposed a theoretical framework of decoupling when studying the relationship between GDP, traffic volume and transport carbon emissions in the European Union, which has become a commonly used model to explore the correlation between social development and environmental impact [64]. The Tapio model described the GDP decoupling elastic of carbon emissions in transportation industry as:(10)β=%ΔC/%ΔGDPThe decoupling model is improved in this study by using the PDI instead of a single index to reveal the decoupling features between population development and carbon emissions. The decoupling elasticity, β, can be represented as follows:(11)β=EC/EPDI
+        (12)EC=ΔC/Cb=(Ct−Cb)/Cb
+        (13)EPDI=ΔPDI/PDIb=(PDIt−PDIb)/PDIb
+        where ΔC and ΔPDI represent the change of carbon emissions and the PDI from base year b to target year t, respectively; Ct and Cb denote carbon emissions in year t and year b, respectively; and PDIt and PDIb indicate the value of the PDI in year t and year b, respectively.In order not to overinterpret slight changes as significant, a ±20% variation in the elasticity values around 1.0 is still regarded as coupling. Thus, coupling is defined as elasticity values of 0.8–1.2, and decoupling and negative decoupling is defined outside of this scope [64]. The decoupling can be divided into three degrees and eight states, as shown in Table 2.The STIRPAT model is introduced to evaluate the nonlinear impact of population, the economy and technological development on the environment [65,66]. It can be expressed as follows:(14)I=aPbAcTde
+        (15)lnI=lna+blnP+clnA+dlnT+lne
+        where I represents the environmental impact, which is carbon emission in this study. P represents the PDI; A represents affluence per capita; T represents the technological level and is measured by technology market turnover and number of patents; b, c and d reflect the importance of P, A and T respectively; a and e are constants.To identify the impact of population-related factors on carbon emissions, we disaggregated population factors into the following: population size (total and trend), population structure (age, urban–rural and employment structure), population quality (education level) and population wealth (economy and consumption). The extended STIRPAT model is expressed as follows:(16)lnI=lna+b1lnTP+b2lnPG+b3lnP0−14+b4lnP15−64+b5lnP65++b6lnUR+b7lnES+b8lnET+b9lnEP+b10lnIR+b11lnGP+b12lnCE+dlnT+lne
+        where A is integrated into population wealth, other meanings are the same as the above.In this study, time series data are used for the regression of 30 provinces to clarify the driving mechanism of carbon emissions in each province.The standard form of multiple linear regression model is usually expressed as:(17)Y=Xβ+ε,
+        where Y is a (n × 1) matrix of dependent variables, X is a (n × p) matrix (rank p) of independent variables, β is a (p × 1) vector of coefficients and ε denotes the normally distributed random errors. The unbiased estimate of β is normally given by:(18)β^=(XTX)−1XTY,Due to the limitation of the social and economic variables, there are always correlations among the variables, that is, multicollinearity. When there is a multicollinearity between the independent variables, the  XTX matrix is ill conditioned, i.e., the value of its determinant |XTX| ≈ 0. The calculation of the XTX matrix is sensitive to slight variations in the data. The addition or deletion of a variable or the slight change of a variable will have a great impact on the results, leading to the instability of the regression results. To control the general instability and inflation in estimating β, the ill-conditioned problem needs to be transformed into a conformity problem by adding a regularization term to the loss function, i.e., ridge regression [67]. The ridge regression model can obtain an acceptable biased estimate with small mean square error in independent variables through a bias variance tradeoff, which is one of the effective methods to deal with multicollinearity [68]. The general expression of the ridge regression model is as follows:(19)β^=(XTX+kE)−1XTY,
+        where E is unit matrix, k is the variable ridge regression coefficient in ridge traces and the value of k varies from 0 to 1. The ridge regression estimates are computed with various increasing values of k, starting from k = 0, until an optimum value of k is determined for where all the regression coefficients appear to have stabilized.China’s 30 provinces (Figure 1) are investigated (excluding Tibet, Hong Kong, Taiwan and Macau where data are not available). These provinces have made great achievements driven by reform and an opening-up policy. For example, per capita GDP rose from less than CNY 10,000 in 2001 to more than CNY 60,000 in 2017, and the urbanization level has also improved significantly. However, there are significant differences between provinces. The economic strength and urbanization processes of the eastern provinces are higher than those central and western provinces, and the population distribution is also quite different.The population density of the eastern provinces is much higher than that of the western provinces. The development status of different provinces and regions leads to differences in the spatial distribution of population development and carbon emissions. The decoupling state of carbon emissions and population development in different provinces may also vary greatly. In order to further clarify the decoupling state of each region and provide support for decision making, this study discusses the decoupling relationship between carbon emissions and the PDI at the national and provincial scales.The PDI component data (2001–2017) are obtained from the China Statistical Yearbook and Population Census Bulletin. The energy consumption data (2001–2017) comes from the China Energy Statistics Yearbook. The data of Chinese administrative boundaries are obtained from the Resources and Environment Science and Data Center of the Chinese Academy of Sciences. These data sources are listed in Table 3.The PDI changes of 30 provinces from 2001 to 2017 are evaluated to reflect the characteristics of population development. Among the 30 provinces, Guangdong has the highest PDI score, as shown in Figure 2. As the province that contributes the most to China’s GDP, its per capita wealth is higher than that of the other 29 provinces, and the population age structure is getting younger. Beijing and Shanghai also have strong PDI competitiveness. Shandong, Jiangsu, Zhejiang, Tianjin, Henan and Fujian also saw a significant increase in the PDI between 2001 and 2017. These provinces have higher per capita wealth or population size, resulting in a higher PDI than other provinces.Jilin, Heilongjiang and Gansu are at the bottom of the PDI list. Although continuous population urbanization is occurring in Jilin and Heilongjiang, the geographical location and climate problems of these two provinces have led to a large population outflow. In addition, the natural population growth rate is low, the population scale is on the decline and the elderly population continues to grow, which is not conducive to the long-term population and regional development. There is limited per capita wealth and an aging population, which are common characteristics of these three provinces. For these provinces, it is important to formulate relevant population policies to promote population inflow, reduce the proportion of the aging population and give full play to the dividend of population agglomeration so as to promote long-term population development.The average PDI in the 30 provinces increases over time, while regional disparities are also reflected in the PDI, similar to how Table 4 shows, which contained the statistical information of the PDI. In 2001, Guangdong Province showed the optimal PDI of 0.38, while Guizhou and Qinghai had the lowest PDI, which is only 0.20, lower than the national average of 0.25. In 2017, the optimal PDI reached 0.67, while the lowest PDI was 0.28. The individual gap between the optimal PDI and the worst gradually widens each year in the sample period, indicating a huge development gap. Although the PDI in Hebei, Shanxi, Inner Mongolia, Anhui, Jiangxi, Hubei, Hunan, Guangxi, Hainan, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia and Xinjiang showed a gradual upward trend, they still failed to reach the initial value of Guangdong, which indicates that there is great potential for further improvements in these provinces.Reviewing the population development of all of the provinces, we find that Guangdong’s population wealth and population structure, regardless of age structure or employment structure, are in the best state in China. Therefore, for most provinces, improving the quality of population development should not only focus on accelerating the urbanization of population, but also pay attention to the improvement of population wealth and employment. Meanwhile, resources should be coordinated at the national level. Instead of widening the development gap gradually, provinces that develop first should lead those that develop later, and finally achieve common development.The national decoupling between carbon emissions and the PDI during the study period can be divided into four states: expansive negative decoupling (EN), expansive coupling (EC), strong decoupling (SD) and weak decoupling (WD). Table 5 provides a complete decoupling relationship dynamic, showing that the interannual decoupling state gradually changes from EN to the decoupling state at the end of the study period. In the period from 2012 to 2015, each interannual decoupling both showed a SD state, while other periods, except from 2016 to 2017, showed a coupling state, including EN and EC. EN and EC reflects the close relationship between carbon emissions and the PDI, which indicates that both carbon emissions and the PDI increased, and carbon emissions changed more than the PDI in EN state. The decoupling state, including WD and SD, indicates that the dependence of carbon emissions on PDI decreases. In particular, in the SD state, the PDI continues to rise while carbon emissions are reducing, which is an ideal state.In the early stage of the sample (2001–2007), economic globalization promoted China’s rapid development, and the economic growth rate reached over 9%, which also exacerbates the contradiction between social development driven by energy consumption and sustainable development. During this period, the carbon emissions growth rate is much higher than that of the PDI; EN is the primary decoupling state.In 2007–2008 and 2008–2009, there was EC, the economic situation was not optimistic and the industrialization process slowed down due to the impact of the economic crisis. Industries with high energy consumption and high emissions, such as construction, had been largely shut down due to the decline in market purchasing power, which greatly reduced carbon emissions. Although the population unemployment rate increased and the growth rate of population wealth slowed down, from the perspective of multidimensional population development, the impact of the economic crisis on the population development is not obvious. As shown in Figure 2, the PDI of most provinces continued to grow during this period.In each period between 2009 and 2012, the EN or EC showed that the link between carbon emissions and the PDI had strengthened after the economic crisis. In order to promote economic development and safeguard people’s well-being, China adopted a series of macro-economic regulation measures, including tax cuts and tax rebates, expanding domestic demand, etc., which led to the growth of carbon emissions. China also became the largest carbon emitter during this period [69]. In order to ease the pressure of carbon emissions, China paid more attention to the harmony with nature in the following years (2012–2017); SD was the main decoupling relationship during the period. However, it is important to note that EC also appeared during this period, which is a nondecoupling state. Our results show that the growth rate of the PDI is greater than zero in both decoupling years and nondecoupling years, but the change rate of carbon emissions in nondecoupling years is larger, while the change rate of carbon emissions in decoupling years is small or negative. In the long run, SD state may be difficult to maintain, which means that there is no real decoupling between carbon emissions and the PDI. The same decoupling trend is also reflected in the research of Shang and Luo [27]. Therefore, the key to decoupling is to effectively control the change rate of carbon emissions, while evaluating the effectiveness of the decoupling state requires examining the changes of carbon emissions and the PDI over a period, for example, taking 5 years as an evaluation cycle.Table 5 also shows the long-term decoupling relationship. The results show that although the interannual decoupling relationship between 2011 and 2015 is dominated by SD, the tension between carbon emissions and population development is not actually alleviated from the five-year assessment period because EC is a long-term decoupling state. This means that the growth rate of carbon emissions is still higher than the growth rate of the PDI, and only shows a short and small decrease in some years, making the decoupling state vulnerable to variable changes in the short term. The change of the decoupling state undoubtedly shows that both active and passive emission reduction need a long-term process, and decoupling can only have a discernible effect on the climate if it is consistent over a number of years. The long-term change trend of the decoupling relationship from EN to EC and then to WD means that carbon emissions are gradually decoupling. However, WD also indicates that carbon emissions have not shown a downward trend, and there are still many efforts to be made at the national level, such as developing clean energy and improving population welfare, so as to achieve a stable strong decoupling state in some period in the future.To further promote the decoupling of carbon emissions from the PDI, and prevent the transition from decoupling to coupling again, on the one hand, the national level should continue to adhere to the green and sustainable development, gradually eliminate the industries with high energy consumption and high emissions, and promote the transition from the consumption structure based on fossil energy to the utilization of renewable energy. On the other hand, the government should continue to create employment opportunities, improve the level of per capita education and the quality of the population, formulate population policies, improve the aging phenomenon and promote high-quality population development.The interannual decoupling state of each province is determined by calculating the decoupling coefficients of each province from 2001 to 2017. The evolution trend of decoupling coefficient in most provinces is basically consistent with the national level. Meanwhile, the decoupling coefficient across most provinces also shows a similar trend without considering the decoupling state. However, the provincial interannual decoupling state includes not only the EN, EC, WD and SD state, but also the SN state. Ningxia, Qinghai and Gansu are significantly different. These provinces showed more SN state, indicating that the growth rate of carbon emissions is higher than that of the PDI, and there is unbalanced development between the two. The main reason is that these provinces, located in western China, are rich in fossil resources. In the context of the urgent need of local governments to improve their development level and the Great Western Development Strategy of China, the energy-driven development model will undoubtedly lead to a large amount of carbon emissions. Although the PDI has also improved, it also pays a high environmental cost (carbon emissions).As mentioned above, a long-term state of decoupling might make more sense. The long-term evolution of decoupling is shown in Figure 3. The research period is divided into 2001–2005 (10th Five-Year Plan), 2006–2010 (11th Five-Year Plan), 2011–2015 (12th Five-Year Plan) and 2016–2017 (13th Five-Year Plan). It can be clearly seen that EN is the main decoupling state during 2001–2005 and 2006–2010, while EC state appeared in a few provinces and WD only appeared in Beijing and Shanghai. During the 12th Five-Year Plan period (2011–2015), more and more provinces began to show EC and WD states, and only Shanghai is in the SD state. The coexistence of EN, EC and WD is the main decoupling feature in this period. Most of the central and western provinces are in the EN state, indicating that carbon emissions and the PDI is still unbalanced, but the tension between carbon emissions and the PDI has eased compared with the previous two periods. Provinces in the EC state are mainly distributed in the central region, while most eastern coastal provinces are in the WD state.In the 13th Five-Year Plan period (2016–2017), the number of provinces in the EN state has further decreased, and some provinces, including Beijing, Shanghai and Chongqing, appeared in the SD state, showing an ideal direction of decoupling evolution. However, there are also some provinces that showed the opposite direction of evolution. For example, in the 12th five-year period, Shandong and Henan are in the WD state, Yunnan is in the EC state and from 2016 to 2017, the three provinces are in the EN state again. This may be due to the lack of complete data for the 13th Five-Year Plan (2016–2020). As mentioned above, the short-term decoupling relationship is susceptible to the effects of variable tiny changes. Still, it is a reminder to managers that they need to continue to reduce carbon emissions as they develop to prevent carbon emissions from rising again.From the perspective of the decoupling evolution of each province, we found some noteworthy phenomena. The decoupling status of some provinces, including Inner Mongolia, Gansu, Ningxia and Qinghai, did not changed during the four periods, which is more reflected in EN. In terms of geographical location, these provinces are all located in the central and western regions of China. Due to the population flow, especially some young labor force to the eastern provinces, there are obvious differences in population development between these provinces and eastern provinces of China. In addition to geographical conditions, economic policies and other reasons, although the PDI of each province is on the rise, the development gap between regions is gradually widening, which is consistent with the results of Section 5.1. Meanwhile, these provinces are key players in China’s power grid supply. Due to the large population and high energy demand of central and eastern provinces, coupled with the resource mismatch between provinces in China, the stable supply of electricity requires the export of resource-rich provinces, such as Inner Mongolia, Gansu, Qinghai and Ningxia [70].The existing studies show that the power sector is one of the largest carbon emission sectors in China [71]. If China’s power sector was considered as a country, it would be the third largest carbon emitter in the world [72,73]. However, the embodied carbon emissions associated with power transfer are not considered in our study. Large-scale electricity production generates carbon emissions locally, so decoupling remains a challenge for these provinces. The decoupling changes of Henan and Shandong are also worth paying attention to because their decoupling states have undergone a transition from coupling to decoupling and then to coupling. To further promote the decoupling of carbon emissions, two major efforts may be possible: on the one hand, optimize the energy production structure and gradually replace the current coal-dominated secondary power generation structure; on the other hand, formulate policies to attract talent and improve the population welfare to promote the improvement of the PDI.A ridge regression is used to eliminate the influence of multicollinearity among the variables on the regression results. Supported by time series data, the extended STIRPAT model results of 30 provinces are obtained, as shown in Table 6. For each of the 30 provinces, the regression equation is significant (F statistic sig < 0.05), and the fitting degree (R2 ≥ 0.9) is good. However, some of the variables in some provinces do not pass the significance test of the ridge regression with 90 percentile confidence intervals, for example, PG in Tianjin, P65+ in Hebei and other specific significance results are also presented in Table 6.The STIRPAT model is utilized to explore the impact of different population factors on carbon emissions, and on this basis to explore policy recommendations to promote decoupling. From our empirical results, we identify several meaningful phenomena.First, compared with other factors, population growth has no significant impact on carbon emissions in most provinces and the total population has an impact on carbon emissions in all of the provinces. As one of the main driving factors of carbon emissions, the total population promotes the growth of carbon emissions in most provinces, while the inhibiting effect is only in a few provinces (Liaoning, Jilin, Heilongjiang, Anhui, Hubei and Chongqing). For these provinces that have the effect of population inhibiting carbon emissions, attractive talent introduction policies can be formulated to promote population transfer, further play the emission reduction effect of population and promote decoupling between carbon emissions and the PDI. In addition, according to our results, controlling rapid population growth is obviously beneficial to carbon emission reduction in most provinces, but it should be noted that it may accelerate the emergence of other social problems, such as the phenomenon of population aging. The results show that the aging phenomenon in most provinces promoted the growth of carbon emissions, which means the carbon emissions are not mitigated and is not conducive to population development.Second, the labor-oriented age structure contributes to the growth of carbon emissions, and the aging population is negatively correlated with carbon emissions in provinces with a higher PDI, while positively correlated with carbon emissions in provinces with a lower PDI. This is consistent with the results of Zhang and Tan [40]. Even after retirement, as the older individuals continue to look for other jobs, the swelling labor force led to the growth of carbon emissions. In addition, they are less willing to pay for environmental protection because the costs are immediate, but they will not benefit from a high-quality environment in the future. It may be helpful to promote carbon decoupling by build more green leisure places for the elderly to ease the labor glut.Third, the obvious improvement of the urban–rural structure means that economic development is effective, and the population is richer and has a stronger purchasing power. On the one hand, the improvement of the living standard lead to more direct and indirect carbon emissions, including more direct energy demand and fuller range purchases of home appliances, as well as more entertainment and leisure spending.On the other hand, a higher industry and technology level promotes the consumption of more commodities and stronger purchasing power, and demand further promotes the development of the industrial and technological level. Many are choosing to shift from agriculture to higher-paying secondary and tertiary industries. The changes in employment structure also have an impact on carbon emissions, and the employment in both secondary and tertiary industries has contributed to the growth of carbon emissions during the current development period in almost all of the provinces. This means that the development level of China’s tertiary industry still needs to be improved because, theoretically, the more people engaged in the tertiary industry, the more developed, cleaner and more efficient the tertiary industry will be, and the lower the carbon emissions will be.Therefore, in order to promote the decoupling of carbon emissions from the PDI, it is necessary to establish low-carbon supporting industries based on the characteristics and needs of the provinces. For example, for those provinces that are in the SN decoupling state, most of which are in central and western China, they can make use of their location and resource advantages to vigorously develop wind power, natural gas, new energy and other industries to shift the way of people’s life towards reliance on clean energy. This is not only conducive to promoting the decoupling of carbon emissions but is also conducive to optimizing the employment structure. While promoting China to achieve the carbon peak and carbon neutral goals, it will also raise the level of population development.At last, population quality contributes to carbon emissions, although it is not significant in some provinces. Studies have shown that education is positively correlated with carbon emissions [74,75]. In China, improving population quality can promote economic prosperity, which, in turn, contributes to more carbon emissions. High-quality people also tend to have the ability to do more consumption and other behaviors that contribute to carbon emissions [76,77,78]. Environmental protection, therefore, should be integrated into the existing teaching system as a classroom teaching content. The government should guide people to adopt a green and low-carbon consumption pattern, such as introducing free buses to replace private cars, so as to promote the change of population’s consumption concept and promote carbon decoupling. This is not only for the provinces in SN decoupling state, but also for other provinces in China.As working towards sustainable population development is an important part of carbon mitigation efforts, this study conducts a decoupling relationship analysis between carbon emissions and the PDI and investigates the influential mechanism between them. The following objectives are achieved in this study: (1) an integral population-related indicator, the PDI, is constructed to reflect the population development features, including population size, age structure, urban–rural structure, employment structure, population quality and personal wealth; (2) the decoupling model is established to investigate the decoupling relationship between carbon emissions and the PDI; and (3) the impact of population factors on carbon emissions are investigated and some suggestions are put forward for promoting carbon decoupling. The main findings and policy implications are as follows:There is a significant increase in the PDI in all of the provinces, however, the inter-provincial gap has widened in terms of population development. In order to narrow the gap, the local governments should pay attention to the multidimensional population development process, and the central government should the coordinate resources and talent to favor China’s western provinces.The decoupling relationship between carbon emissions and the PDI at the national level has experienced a transition from EN to EC, and then to the decoupling state, showing an ideal evolution process. The decoupling degree at the provincial level has also strengthened from 2001 to 2017, but some provinces are still in the EN state. These provinces can promote the decoupling of carbon emission from the PDI by developing clean energy supporting industries and increasing subsidies for clean energy markets.The influence of population factors on carbon emissions is different in different provinces, but the total population, population wealth, population urbanization, labor force population and elderly population in most provinces are almost always positively correlated with carbon emissions. To promote the decoupling of carbon emissions from the PDI, provinces should develop low-carbon-supporting industries according to their own characteristics.Although our research is focused on China, given that it is the world’s largest carbon emitter and most populous country, this study may help to prompt managers to focus on sustainable population development, not just high-quality economic development, as China shifts to high-quality development. Meanwhile, these implications may also promote some studies on population decarbonization in other countries of the world, thus promoting sustainable human development at the international level. In addition, due to the complexity of the influencing factors of carbon emissions, decoupling research can be further extended to other factors in the future, so as to promote the development of overall decoupling.Still, this study also has limitations: On the one hand, in the construction process of the PDI, this study only focused on several major aspects of the current population, which can represent the development degree of population to some extent, but it is not comprehensive from the perspective of all-round evaluation. On the other hand, the carbon emission accounting in this study is based on the end-energy consumption of each province. Given the large-scale electricity trade at the provincial level in China, this will lead to a large amount of embodied carbon transfer, which is not considered in our study. In future studies, we will explore more comprehensive indicators of population development and explore the possible impact of embodied carbon transfer on decoupling.K.Z.: conceptualization, methodology, formal analysis, investigation, writing—original draft preparation, visualization and writing—review and editing; X.C.: conceptualization, validation, formal analysis, writing—review and editing and funding acquisition; Z.Z.: conceptualization, methodology, software, investigation, data curation and visualization; P.H.: conceptualization, methodology, investigation and data curation; D.L.: conceptualization, methodology, investigation and data curation. All authors have read and agreed to the published version of the manuscript.This research was funded by the Project of National Social Science Fund of China [grant number 21BGL186]; the Project of National Natural Science Foundation of China [grant numbers 71603062, 71601042]; the Project of Philosophy and Social Sciences in Guangdong Province [grant number GD14CGL02]; the Project of Guangzhou University’s 2017 training program for young topnotch personnel [grant number BJ201723]; and the Key Special Project for Introduced Talents Team of Southern Marine Science and Engineering Guangdong.Not applicable.Not applicable.The data used in this study can be obtained from the China Statistical Yearbook, the China Energy Statistical Yearbook and the provincial statistical yearbook.The authors declare no conflict of interest.Study areas.The PDI changes in 30 provinces from 2001 to 2017.The decoupling evolution between carbon emissions and the PDI.PDI indicator.Decoupling degrees.Data sources.The PDI Statistical information.Decoupling trends at the national scale from 2001 to 2017.Extended STIRPAT model results for 30 Chinese provinces.Note: ***, ** and * represents significant at 1%, 5% and 10% levels, respectively.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11025.txt

@@ -0,0 +1 @@
+Endometriosis is a common disease, affecting up to 60–80% of women, with pelvic pain or/and infertility. Despite years of studies, its pathogenesis still remains enigmatic. Genetic, hormonal, environmental, and lifestyle-related factors may be involved in its pathogenesis. Thus, the design of the review was to discuss the possible role of environmental factors in the development of endometriosis. The results of individual studies greatly differ, making it very difficult to draw any definite conclusions. There is no reasonable consistency in the role of environmental factors in endometriosis etiopathogenesis.Endometriosis is a condition that occurs in women of childbearing potential, although it is only occasionally seen in prepubertal girls and postmenopausal women. It is defined as the presence of endometrial glands and stroma outside the uterine cavity. The prevalence of endometriosis reported in publications varies greatly, ranging from 5% to as much as 50% [1,2,3,4]. This wide variation may have several reasons. Not all cases of endometriosis are detected: in some women, where endometriosis is clinically asymptomatic, the condition can go undiagnosed or is incidentally discovered during surgical procedures performed for other reasons. Subtle endometrial lesions may go unnoticed even during surgery. On the other hand, the condition is more thoroughly looked for in patients undergoing surgery for infertility or chronic pelvic pain [1,2,3]. Endometriosis is estimated to be present in 2–22% of women without any clinical manifestations of the condition, 40–80% of patients with lower abdominal pain, and 13–48% of women with infertility [4]. Despite the discrepancies in the evaluation of its prevalence, endometriosis is undoubtedly one of the most commonly diagnosed and treated disorders of the female reproductive system.Of the numerous theories attempting to explain the underlying mechanisms that lead to the development of endometriosis, several merit closer attention and may be grouped in three categories: in-situ theory, implantation theories, and induction theories. According to the in-situ theory, ectopic endometrium develops de novo from the multipotential peritoneal epithelium, from the germinal epithelium of the ovary, or from remnants of the Wolffian or Mullerian ducts. Implantation theories are based on the premise that the menstrual debris contains viable endometrial cells that can migrate, become implanted, and grow. Dissemination of endometrial cells may occur by extension, through the lymphatic vessels or blood vessels, through the ovaries during “retrograde menstruation”, or it may be iatrogenic and take place during surgical procedures. A hypothesis that combines these two groups of theories is the induction theory proposed by Levander. Based on his studies, Levander suggested that the exfoliated endometrial cells that arrive in the peritoneal cavity secrete active substances that stimulate the peritoneal epithelium to transform into endometrial tissue [4,5,6,7]. The aetiology of endometriosis remains, however, unclear and seems multifactorial. Genetic, hormonal, environmental, and lifestyle-related factors are involved in its pathogenesis [8].Review on the literature suggests that there is no reasonable consistency in the role of environmental factors in endometriosis etiopathogenesis. The results of the trials presented by different authors are incoherent, and the conclusions are often contradictory. Thus, the design of the review was to discuss the possible role of ecological factors in the development of endometriosis. The aim was to collect the most recent data on environmental factors involved in the pathogenesis of the disease. We searched the MEDLINE and Embase databases to May 2021. The studies included were published during the last decade. A combination of the following medical subject headings or text words were used: endometriosis, dioxins, polychlorinated biphenyls, estrogens, diet, endocrine-disrupting chemicals, bisphenol A, and environmental factors. Endometriosis is a disease that, despite many years of research, is still surprising. Scientific papers describing its course are often contradictory and usually do not provide a consistent answer to puzzling issues. Due to the lack of uniform standards that adequately describe the disease, it is not possible to group scientific papers describing the topic.Intrauterine exposure of female fetuses to certain substances is associated with a higher risk of endometriosis. Endometriosis is an estrogen-dependent disorder, and the ectopic endometrium, similarly to the normal uterine mucosa, expresses estrogen receptors [9]. There are currently no substantiated scientific data to demonstrate the benefits of estradiol to human fetuses. Administration of ethinyl estradiol, a common ingredient of oral contraceptives, to pregnant mice results in an increased incidence of adenomyosis, or endometriosis of the uterine muscle [10]. Ethinyl estradiol is a hormone that falls into category X, which means that studies in animals or humans have demonstrated fetal abnormalities and/or there is positive evidence of human fetal risk based on adverse reaction data from investigational or marketing experience, and the risks involved in use of the drug in pregnant women clearly outweigh potential benefits. Diethylstilbesterol (DES) is a drug that was used to prevent premature labor until 1971, when it was confirmed to exert teratogenic effects [11]. Analysis of the Nurses’ Health Study unequivocally demonstrated an increased risk of laparoscopically confirmed endometriosis in women who had been exposed to DES as fetuses. Interestingly, women who were born of twin pregnancies had additionally increased risk of endometriosis, which, hypothetically, could have been due to the higher estrogen concentrations that typically accompany such pregnancies [12]. The association between in-utero exposure to DES and the risk of endometriosis has also been confirmed in other studies [13]. The exact mechanism responsible for this association remains unknown. Golden et al. demonstrated that in utero exposure to DES interferes with the expression of estrogen receptors [14]. Exposure of mice to DES results in abnormal expression of estrogen receptor-dependent genes that encode lactoferrin and epidermal growth factor in the uterus and vagina [10]. Exposure to DES during pregnancy also leads to structural abnormalities of the uterus in female fetuses. The classification of congenital anomalies of the uterus includes a separate category of “developmental anomalies associated with diethylstilbesterol use” [15]. Congenital anomalies of the uterus, by augmenting “retrograde menstruation”, are a recognized risk factor for endometriosis [5].Exposure of pregnant women to toxins can also be associated with a higher risk of endometriosis in their daughters. This association is particularly well understood in cases of exposure to dioxins and polychlorinated biphenyls. A study in an animal model demonstrated that prenatal exposure to 2,3,7,8-tetrachlorodibenzodioxin (TCDD) increased the risk of surgically induced endometriosis in adult mice and rats [16]. This phenomenon may be associated with progesterone resistance [17]. It was hypothesized that exposure to TCDD and other dioxins in fetal life could induce endometriosis as a result of epigenetic modifications, interfering with the expression of genes involved in the pathogenesis of this condition [18]. In-utero exposure of mice to bisphenol A (BPA) is also associated with the risk of endometriosis in later life [19]. This effect is most likely due to the unbalanced hyperestrogenism caused by the inhibited expression of progesterone receptors on the one hand and the increased estrogen concentration on the other [20].It is estimated that approximately 1.7% pregnant women worldwide smoke [21]. Smoking during pregnancy negatively impacts embryofetal growth and development. It increases the risk of miscarriage, premature labor, intrauterine growth restriction, and many other complications of pregnancy [22,23]. Smoking has, however, been shown to reduce the risk of endometriosis in later life among female fetuses [24]. The underlying mechanism is unknown. Nicotine, along with its metabolite cotinine, may suppress aromatase-dependent conversion of androgens to estrogen, stimulate apoptosis, and inhibit angiogenesis [20]. Both hyperestrogenism and neovascularization play a fundamental role in the development of this disorder, which is why nicotine may inhibit the development of endometriosis (see Table 1).Results of many studies point to an association between diet and the development of endometriosis. This association is, however, unclear. It may be speculated that dietary factors affect the formation of sex hormones and exert pro- or antioxidant effects and proinflammatory effects. All these factors play a considerable role in the pathogenesis of endometriosis. At the same time, the effect of contaminants accompanying food production on the development of this condition cannot be ruled out. In 2004, it was shown that women with laparoscopically confirmed endometriosis consumed smaller quantities of fresh fruits and green vegetables compared to women without endometriosis [25]. Harris et al. also demonstrated that consumption of fresh fruits, including, in particular, citrus fruits, reduced the risk of this disorder [26]. A diet rich in fruits provides large amounts of provitamin A, lower levels of which have been reported in patients with endometriosis [27]. Vitamin A suppresses the formation of the proinflammatory interleukin-6, a cytokine whose high levels are found in the amniotic fluid in women with endometriosis [28,29]. Citrus fruits also contain high quantities of vitamin C, which inhibits inflammation and exerts antioxidant effects. Vegetables and particularly fruits may contain organochlorines, which in turn have been positively associated with the risk of endometriosis [30,31]. Paradoxically, Trabert et al. showed an increased risk of endometriosis in American women consuming large amounts of fruits [32]. Hypothetically, this may be associated with the large quantity of pesticides used during cultivation in the United States.A diet rich in red meat increases the risk of endometriosis. The Nurses’ Health Study II, which included nearly 82,000 American nurses, demonstrated over 22 years of follow-up that consumption of both processed and unprocessed red meats increased the risk of endometriosis. This association was strongest among women who had never reported infertility and thus were more likely to present with pain symptoms. On the other hand, intakes of poultry, fish, shellfish, and eggs were shown to be unrelated to endometriosis risk [33]. This study confirmed the findings of Parazzini et al., who also demonstrated an association between the consumption of red meat and the risk of endometriosis [25]. This effect may be due to the pro-oxidative effects of haem released from red meat. Haem is involved in the initiation of lipid peroxidation, in which unsaturated fatty acids or other lipids are oxidated by free radicals to their respective peroxides [34]. Disturbances of the pro-/antioxidant balance that lead to oxidative stress are a recognized mechanism responsible for the development of endometriosis [35]. A diet rich in red meat can also be responsible for higher estradiol levels by which it leads to endometriosis. Harmon et al. demonstrated that vegetarian women have lower serum levels of estrone and estradiol compared to women who consume large amounts of red meat [36]. Palmitic acid was the only saturated fatty acid that was significantly associated with increased risk of endometriosis. High intake of trans-unsaturated fats was also a risk factor for endometriosis, whereas intake of long-chain omega-3 fat acids decreased the risk of endometriosis (see Table 2) [37].Endometriosis is an estrogen-dependent disorder that develops in women of childbearing potential. Its pathogenesis is therefore affected by endocrine-disrupting chemicals (EDCs), which are defined as a group of exogenous chemical compounds that affect the function of the endocrine system. The endocrine-disrupting chemical is defined as “an exogenous chemical, or mixture of chemicals, that interferes with any aspect of hormone action” [38]. The continuously updated list of EDCs maintained by TEDX (The Endocrine Disruption Exchange) has grown to include almost 1500 substances at the moment [39]. Endocrine-disrupting chemicals enter the human body with food, water, dust by inhalation, with the inspired air, and via the transdermal route after using cosmetics and creams. Transplacental transfer of these substances to the developing fetus has also been demonstrated [40]. They are also found in human milk [40]. EDCs accumulate in the adipose tissue, and the effects of their metabolites persist for a long time [41]. Because of the potential association between exposure to EDCs and the development of endometriosis, many studies have been devoted to this topic. Such studies are difficult to design, as it is difficult to identify both the study group and the control group and to measure the exposure to EDCs and the effects of other factors on the development of this condition. An additional difficulty is associated with the fact that endometriosis may take several different forms (ovarian endometrioma, peritoneal endometriosis, deeply infiltrating endometriosis, and adenomyosis—or endometriosis of the uterine muscle), which not only differ in location but also have different clinical presentations. In some cases, endometriosis remains asymptomatic, and a certain diagnosis can only be established by invasive evaluation and histopathological confirmation. Silent endometriosis is a condition in which the patient does not experience any discomfort resulting from the development of the disease. Symptoms may appear in later life or remain dormant.Cosmetics and personal care products (PCPs) contain numerous EDCs [42]. They are increasingly used worldwide. In the USA alone, the cosmetics and PCPs market is estimated at €78.6 billion [43]. EDCs released by cosmetics and PCPs are mainly associated with the benzophenone (BP) and paraben (PB) families. BPs include BP-1, BP-3, and 4-hydroxybenzophenone (4-OH-BP) and are used as UV blockers in suntanning creams and other cosmetics [44]. Frequent use of cosmetics and PCPs results in higher urinary concentrations of these substances and may potentially lead to endometriosis. This is likely associated with the estrogenic and pro-oxidative effects of these compounds [45]. Oxidative stress is a recognized factor involved in the pathogenesis of endometriosis [34]. By damaging germinal cells and disrupting folliculogenesis, EDCs adversely affect fertility, which further exacerbates the problem of infertility in women with endometriosis [46].Of the many EDCs, compounds that are best understood in terms of potential involvement in the pathogenesis of endometriosis are polychlorinated biphenyls (PCBs) (congeners 118, 138, 153, and 156) and dioxins (2,3,7,8-TCDD, 1,2,3,7,8-PeCDD, 1,2,3,4,7,8-HxCDD, and 1,2,3,6,7,8-HxCDD).Bisphenol A (BPA) disrupts the pulsatile secretion of GnRH, which negatively affects the function of the hypothalamic-pituitary-ovarian axis. Pre- and postnatal exposure to BPA induces future structural changes in the ovaries, which further contributes to abnormal folliculogenesis. It may also adversely affect the structure of the uterus [40], and by increasing retrograde menstruation, it may increase the risk of endometriosis. BPA is able to exerts its endocrine effects, as it displays affinity for the estrogen receptors Erα and Erβ [47]. A review article by Anger and Foster suggests that BPA, despite its high environmental distribution, is not the leading cause of endometriosis. The main reasons are the short half-life in the human body and too-low doses to which the studied women were exposed. Researchers suggest that unequivocal results may be due to the long time lag between diagnosis and measurement for exposure [48]. BPA, phthalates, and perfluoroalkyl substances (PFAS) present in water and foods adversely affect puberty in teenagers, and in addition to being involved in the pathogenesis of endometriosis, they increase the risk of polycystic ovarian syndrome (PCOS) and recurrent pregnancy in humans and animals [49]. Studies in mice have demonstrated that exposure to BPA induces endometriotic foci in the adipose tissue that surround the reproductive organs in these animals. BPA has also been shown to induce cystic ovaries, adenomatous hyperplasia with cystic endometrial hyperplasia, and atypical hyperplasia [18]. Results of studies investigating the potential involvement of BPA in the pathogenesis of endometriosis are conflicting. Most studies have, however, reported higher urinary levels of BPA in women with endometriosis compared to those without it [50,51,52]. Higher urinary levels of BPA may only be associated with the peritoneal form of the condition and may not be present in endometriosis located elsewhere [53].Dioxins are considered to be the most toxic industrial pollutants that are highly resistant to degradation and, due to their lipophilic nature, show considerable bioaccumulation [54]. The studies showing the influence of dioxins on the occurrence and development of endometriosis are not unequivocal. In Rier SE et al., the increase in the concentration of 2,3,7,8-TCDD was significantly correlated with endometriosis; however, one should remember about the differences between the organism of the rhesus monkey and that of humans. In the human organism, the adipose tissue accumulating 2,3,7,8-TCDD is significantly better developed, which means that the accumulation effect may be less intense [55]. Studies in women have generated conflicting results on the association between the exposure to 2,3,7,8-TCDD and the development of endometriosis. Mayani et al. found higher blood levels of 2,3,7,8-TCDD in women with endometriosis compared to those without this condition while demonstrating no correlation between the levels of this dioxin and the severity of endometriosis [56]. In 2002, results were published of a study evaluating the risk of endometriosis 20 years after the chemical plant explosion near Seveso, Italy, that resulted in the highest populational exposure to 2,3,7,8-TCDD ever reported. The authors found twice as high but statistically non-significant risk of endometriosis in women whose blood samples collected after the accident displayed 2,3,7,8-TCDD concentrations exceeding 100 ppt [57]. An association has also been shown between increased dioxin accumulation in the adipose tissue and the risk of deeply infiltrating endometriosis [58]. It has been suggested in recent years that the development of endometriosis may be associated with epigenetic alterations [59]. In this mechanism, dioxins may be responsible for the pathogenesis of this condition, as they affect histone modification, DNA methylation, and expression of non-coding RNA (ncRNA) [18]. A study in an animal model has shown that the suppression of the epigenetic histone modification caused by 2,3,7,8-TCDD delays the progression of endometriosis [60]. However, not all the studies have confirmed the association between the exposure to dioxins and the risk of endometriosis [61,62,63,64]. Epigenetic changes occurring in the genetic material modulate susceptibility to endometriosis. Research varies in terms of impact, as epigenetic changes are not constant and are constantly changing. This is due to the regular change in environmental factors [65].Endometriosis is a condition of multifactorial etiology. Despite years of studies, its pathogenesis has not been fully elucidated. Gaining an understanding of its underlying mechanisms would enable effective prevention and treatment of this condition. Genetic factors play a huge role in the pathogenesis of endometriosis. The risk of endometriosis is 6–9% higher in first-degree relatives of women with endometriosis and 15% higher when they had severe disease [59]. The genes responsible for the heritability of the disease have not, however, been identified. The involvement of epigenetic factors in their modification cannot be ruled out. Such effects may be exerted by environmental findings and, by controlling gene activity, may be involved in the pathogenesis of endometriosis (see Table 3).Designing studies to investigate the potential involvement of environmental factors in the pathogenesis of endometriosis is extremely difficult. Endometriosis is a heterogenous disorder that occurs in various forms and locations and has varied symptomatology. In some cases, it is asymptomatic; for a definite diagnosis to be established, histopathological examination needs to be carried out, which in turn requires an invasive approach. Other difficulties are associated with the selection of an appropriate control group and the exact evaluation of the dose and duration of exposure to the substance of interest. The time elapsed from the exposure to the substance of interest to the development of endometriosis induced by it is unknown. It is also impossible to determine the exact onset of the condition. In addition, due to the polyetiological and still unclear nature of endometriosis, it is impossible to eliminate the impact of other factors involved in its pathogenesis, which is why the results of individual studies may greatly differ, making it very difficult to draw any definite conclusions.Conceptualization, G.P.; data curation, G.P. and B.B.; formal analysis, M.R.; investigation, M.F.; methodology, A.W.; supervision, G.P.; writing—original draft, G.P. and B.B.; writing—review and editing, A.W. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The study did not require approval from the Medical Ethics Committee.Not applicable.The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.The authors declare no conflict of interest.In-utero factors of endometriosis.Diet factors of endometriosis.EDC factors of endometriosis.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11026.txt

@@ -0,0 +1,37 @@
+Managing post-traumatic stress reactions in the first few days after exposure to a potentially traumatic event in the course of one’s work remains a challenge for first responder organizations such as Emergency Medical Services (EMS). Psychological First Aid (PFA) is an evidence-informed approach to reducing initial distress and promoting short- and long-term coping strategies among staff in the aftermath of exposure. PFA provided by peer helpers is considered a promising solution for first responder organizations. Unfortunately, first responders may encounter stigma and barriers to mental health care. Therefore, a deeper investigation is needed regarding adherence over time to implemented PFA intervention. The purpose of this study is to qualitatively explore factors that influence adherence to PFA intervention of recipients and peer helpers. EMS workers (n = 11), working as PFA peer helpers for one year, participated in semi-structured interviews. Data were analyzed using thematic analysis; intercoder reliability (κ = 0.91) was also used. Researchers identified four themes and 11 subthemes influencing adherence to PFA intervention: (1) individual perceptions and attitudes of peer helpers and recipients about pfa intervention; (2) perceived impacts on peer helpers and recipients; (3) organizational support to pfa intervention; and (4) congruence with the occupational culture. Study findings herein suggest that it is conceivable to act on various factors to improve adherence to PFA intervention among peer helpers and recipients within EMS organization. This could lead to enhanced understanding of the challenges involved in sustaining a peer led PFA program for first responders.First responders such as Emergency Medical Service (EMS) workers respond to a broad range of emergencies as part of their mandate. These emergencies are unpredictable and recurrent in their nature, increasing the risk of exposure to highly disturbing events (i.e., seeing someone die, a badly beaten adult, or completing a death notification; [1]), and even more so during a prolonged crisis (i.e., disease pandemic; [2]). Mental health problems resulting from exposure to a traumatic event may include acute stress disorder, post-traumatic stress disorder (PTSD), anxiety, depression, and substance abuse [1,3]. Among first responders (i.e., police officers, EMS workers, and firefighters), EMS workers appear worldwide to be the population most at risk of developing PTSD [3]. A meta-analysis estimated prevalence rates of 11% for PTSD, 15% for depression, 15% for anxiety, and 27% for general psychological distress among EMS workers [4]. In addition to mental health problems, meta-analyses have highlighted the negative health impacts of PTSD on physical well-being in adults [5] and among first responders specifically [6,7]. Repercussions are also felt at the organizational level (e.g., increased sick leaves, lowered operational performance; [8,9]). Despite these alarming consequences, a gold standard for early organizational intervention among exposed EMS workers has not yet been established [4,10].One of the challenges in implementing early post-trauma interventions in organizations might be the mental health-related stigma and the barriers to mental health care experienced by a significant proportion of first responders, including EMS workers [11], which interfere with them getting help even when help is readily available. Several studies have shown how organizational support might be a protective factor against mental health symptoms after exposure to workplace trauma [12]. Likewise, EMS workers receiving organizational support report greater emotional stability and belongingness, a steeper decrease of post-traumatic symptoms, and a great likelihood of achieving post-traumatic growth [13,14]. International guidelines also recognize the protective role of peer support following a traumatic event in first responder organizations [15,16].If peer support appears to be a promising solution to operationalizing social support within EMS organizations, the management of post-traumatic stress reactions in the first days following a traumatic event remains difficult. Organizations are confronted with a lack of effective early post-traumatic intervention options to suit and cover all individuals exposed to a traumatic event, regardless of symptomatology, as recently stated by a meta-analysis [10]. Since the 1990s, psychological debriefing has been proposed to address adverse reactions in the aftermath of a traumatic event [17]. This type of intervention is still widely used in first responder organizations [18]. However, considerable controversy exists regarding the potential effects of this form of intervention [18,19]. Instead, multiple health organizations and international trauma experts recommend providing Psychological First Aid (PFA) after a traumatic event since it is an evidence-informed early intervention approach [20,21].Initially developed as a response to natural disasters and terrorist attacks, PFA aims to reduce the initial distress caused by traumatic events and foster short- and long-term adaptive coping strategies. It was designed to help victims of traumatic events and first responders on-site and can be delivered with minimal training in mental health intervention [20]. However, a systematic review concluded that further investigation was needed to establish the effectiveness of PFA on mental health [22]. Indeed, relatively few studies have evaluated this type of intervention, especially regarding the potential reduction in post-traumatic stress symptoms [23,24,25,26]. On this note, it is important to remember that PFA is not a form of treatment, but an early intervention; therefore, more than symptom reduction, PFA aims to foster better adaptation in the short and long term. Therefore, it remains relevant to investigate its implementation in high-risk organizations before examining its effectiveness on multiple outcomes.Researchers and practitioners view peer-led PFA as a promising solution in first responder organizations, but empirical evidence as to its implementation in high-risk organizations is still in its infancy [27,28]. Forbes and colleagues [27] developed a framework to implement PFA within high-risk organizations, but no studies have yet demonstrated the ecological validity of their framework. The few studies evaluating PFA generally investigate the effectiveness of PFA training on providers’ confidence and the skills required to respond after a traumatic event appropriately [28,29,30,31]. However, no study has evaluated PFA implementation, denying researchers important information on the potential facilitators and barriers to implementing and ensuring PFA adherence over time (i.e., peer facilitators’ and service users’ active engagement with PFA support).As first responder organizations increasingly move toward the adoption of evidence-informed mental health interventions, researchers should study how they are implemented and adhered to by EMS workers. On this note, a recent scoping review identified three key components using meta-ethnography analysis [12]. Richins and colleagues [12] found that respect for the organizational culture, support from the host organizations, and building on existing social cohesion and peer support systems made implementation easier, in emergency response organizations. Although this review did not include any PFA intervention, those recommendations for delivering and evaluating early post-trauma interventions may be helpful to put into perspective with our investigation about factors that influence a key part of sound implementation, namely adherence. Adherence is a crucial concept in health practice [32,33] as interventions evolve positively or negatively according to the context in which they occur. Investigating implementation without considering adherence over time regarding participation appears to be in vain [34]. Even if there is still little agreement on a conceptual definition of adherence in the health disciplines [32], it is generally referred to as “the extent to which patients follow the instructions they are given for prescribed treatments” [35] (p. 2). In the present study, we use the concept of adherence in a slightly larger perspective, as the extent to which EMS recipients as well as peer helpers both follow the organizational recommendations to be involved in PFA intervention after exposure to a traumatic event. In order for PFA programs to be sustainable and to become routine inside EMS organizations, it is important that those who benefit from and those who volunteer to provide PFA intervention maintain their involvement over time. As we previously stated, there is still little scientific knowledge about PFA intervention in organizational setting. No studies have yet demonstrated the ecological relevance or transferability of the suggested frameworks. Such a problematic can be better approached in its complexity by a qualitative research design. Qualitative approaches are particularly valuable when you want to explore a new field of knowledge and remain as close as possible to the participants’ experience [36]. Therefore, the present study aims to qualitatively identify factors that influence the adherence of peer helpers and recipients to PFA intervention in an EMS organization by exploring peer helpers’ perspectives.Our 11 participants were recruited from a pool of 37 PFA trained peer helpers of one EMS organization [Urgences-santé], one of Canada’s largest emergency medical services. These peer helpers were active EMS workers trained, using Brymer’s PFA manual [20], to provide PFA to their colleagues after exposure to a traumatic event in the course of their work, doing so in the early hours after the event (i.e., on site, or back at the station). From June 2018 to September 2018, the organization’s in-house psychologist conducted a 14-h PFA training in group sessions (8–10 participants). The research interviews took place one year after implementation of PFA. Considering our small population, we have tried to ensure internal diversification using opportunistic sampling (according to sex, age, type of job, and professional experience) [36,37]. To capture potential differences in perspective due to the changing culture of mental health care among first responders, we ensured that the number of years of experience was diversified among our participants. Furthermore, because PFA was a new intervention in the organization, few PFA interventions had occurred each month since the peer helpers were trained, so we needed to ensure that our participants had provided a minimum of 3 PFA interventions before recruiting them to ensure their ecological experience as PFA peer helpers. All participants reported that they had provided between three and 20 PFA interventions between August 2018 and September 2019. The researchers did not seek to obtain a statistically representative view of the study population, but rather a portrait representative of the diversity of possible experiences.The interviews took place in August and September 2019. All PFA-trained peer helpers were informed about the study by the organization’s psychologist. The researchers did not have access to the reasons for potential participants’ refusals at this stage. Subsequently, the researchers sent more information about the study through emails and phone calls to those who voluntarily gave preliminary agreement. The project had ethical approval (CER-CEMTL 2019-1884) from the ethics committee of Integrated University Health and Social Services Centre for the East Island of Montreal. Participants were informed of the process and intentions of the study, signed a consent form, and were each given a personal code for anonymity. No participant refused or dropped out of the study at this stage or afterwards. Interviews were conducted over the phone at only one point in time per participant. They were asked to be in a quiet and confidential setting outside of their working hours. Telephone interviews were selected as the data collecting method for practical reasons, especially since it facilitates scheduling meetings for the workers and the research team. Moreover, current evidence shows that telephone interviews do not produce lower quality data than face-to-face interviews [38]. We stopped recruiting after a satisfactory level of information power was reached. Information power dictates that the more information the sample holds, the lower number of participants is needed, based on five criteria (i.e., aim of the study, sample specificity, established theory, quality of dialogue, analysis strategy) [39]. Without being able to claim the data saturation, the authors considered that the collected data allowed to add substantial information to the studied phenomenon.The first author conducted individual semi-structured interviews lasting between 30 and 45 min each. All interviews were audio-recorded and later transcribed for qualitative analysis purposes. The interview grid was developed based on PFA intervention implementation, practicability, acceptability, and the consequences of PFA intervention. The research question was inductively developed from the participants’ answers to these questions. The interviewer used reflection and reformulation strategies to explore further following the interview schedule flexibly. Participants also filled out a short sociodemographic questionnaire.Researchers used an inductive thematic analysis approach to analyze the interview data. Thematic analysis is especially useful in understudied and descriptive qualitative research. It allows for the identification, analysis, and the report of patterns within data while being independent from theoretical frameworks. This flexibility allows a more accessible sharing and understanding of data—process and results—with people from all research backgrounds (i.e., researchers of different theoretical orientation, stakeholders, knowledge users) [40]. Inductive thematic analysis was favored by the authors because it is data-driven, meaning that it allows researchers to move away from the pre-existing coding framework and possible preconceptions [40]. The six phases of thematic analysis were followed: familiarization with the data, generation of preliminary codes, searching for potential themes, reviewing themes, defining and naming themes, and producing the report [40]. Only one interviewer, information power, and sample diversity were employed as strategies to ensure the reliability and transferability of the results [36,41]. Moreover, all transcripts were coded systematically by the first author. The last author reviewed the suggested themes, and disagreements led to reexamining the data until the raters reached an agreement. Subsequently, the second author performed double coding. Cohen’s kappa was chosen to measure the intercoder reliability of themes as it is a strict measure [42]. Eleven suggested subthemes were coded using 27% of the interview materiel (i.e., three transcripts), all randomly selected [43]. The global Cohen’s Kappa was κ = 0.91 for inter-rater correlation (SD = 0.0925). Kappa values between 0.40 and 0.60 are commonly considered satisfactory agreement, and values above 0.80 suggest perfect agreement [44]. Our intercoder reliability is therefore considered sufficient to continue data interpretation. Thematic analysis was performed using QDA Miner 5.0 software package (Provalis Research, Montreal, QC, Canada) and intercoder reliability was supported by the use of the QSR-NVivo V.12 software package (QSR International, Burlington, MA, USA). Furthermore, using the previously described analysis, an alternative presentation of the data from a cross-sectional perspective will be suggested in the Discussion section, on the basis of the categories of workers identified.The final sample consisted of 11 participants (nine paramedics and two emergency medical dispatchers), representing 37% of the entire population of trained peer helpers for this organization. Six of the 11 participants were male, with their average age being 43 years (SD = 6.1), see Table 1.The analysis of the interviews found four overarching themes and 11 subthemes regarding factors that may influence adherence to PFA intervention, see Figure 1. Anonymized excerpts are provided in the following sections to illustrate each theme.On the basis of the participant’s responses, it was found that they held certain interpretations which oriented their actions. This finding seemed to be well illustrated by the concepts of perception and attitudes. Perception is understood in the present study as “the process by which organisms interpret and organize sensation to produce a meaningful experience of the world” [45] (p. 52) and attitude as “mindset or a tendency to act in a particular way” [45] (p. 44). Participants reported their perceptions and attitudes regarding this new PFA intervention as peer helpers. They also shared what they knew about perceptions and attitudes toward the intervention from the recipients. Those individual perceptions and attitudes helped shed light on how EMS workers perceive and act during PFA intervention. Their perceptions and attitudes also offered insight into their willingness to adhere its principles, as well as avenues for future service improvement stemming from the negative perceptions that were expressed.Regarding peer helpers’ perceptions and attitudes, all participants described some positive perceptions toward PFA intervention. Some mentioned how simple and easy it was to provide PFA interventions. Many perceived PFA interventions to be well suited to EMS workers’ needs.
+              “It was responding to actual needs, so it’s well adapted. I think the tools that have been put in place are targeted, they are simple and because it’s simple, effective, and allows me to go straight to where I need to go, which is to respond to a need […]. Peer helpers have proven that they are a good influence for not doing well.”
+            Some negative perceptions were also mentioned by a few participants, suggesting areas of improvement. Specifically, some peer helpers shared concerns regarding their own ability to adequately provide PFA intervention or concerns about how they would be accepted as peer helpers by co-workers.
+              “At the beginning, it’s pretty worrisome as a program because you don’t know what you’re getting into […]. Actually, I had my doubts before applying it. I questioned its applicability; I doubted the responsiveness. I was afraid that the reactions would not be in line with what we’d learned. To get bogged down in answers or a slippery slope.”
+            Concerning the attitudes stated by peer helpers, they favored adherence to the PFA intervention. They reported their investment in their role as peer helpers and their feeling of being confident when delivering PFA. Most of them were mindful to follow the directives of PFA intervention, and some described their flexibility and ability to take ownership of the intervention, enhancing their sense of self-efficacy.
+              “Sometimes I don’t do the actions in order; it’s more fluid in the form of conversation. Afterward, I put everything back when I complete the form. Maybe I’m a bit rebellious; I don’t do it to the letter. I have also forgotten certain things because I didn’t have the form in front of me. I contacted people afterward or when I was doing the follow-up.”
+            Most participants highlighted that recipients have positive perceptions toward the PFA intervention, which is a positive predictor of a good adherence to the PFA intervention. According to these participants, EMS workers seemed to give credibility and trust to the new PFA intervention and their peer helper role.
+              “The credibility of the process and the program was based on accessibility, the recruitment process, and the helping relationship offered […]. The fact that there were no cases where the peer helper broke the bond of trust between the recipient and the peer helper. These elements explain, in my opinion, why the program gained credibility.”
+            The majority of participants also described some negative perceptions. According to participants, some recipients seemed to question the sustainability of PFA intervention and its credibility because of its closeness to the organization. Some also expressed concerns around potential breaches of confidentiality.
+              “In the beginning, there was a lot of mistrust on the part of the paramedics because people were saying: “It’s not going to work, [Name of organization] is going to promptly drop this project, it’s not going to work.”
+            Both favorable and unfavorable attitudes toward PFA interventions were identified in recipients, which have the potential to affect long term adherence. Participants predominantly perceived positive attitudes like openness, trust, and recognition of PFA interventions among recipients. Still, they reported sometimes closed attitudes (i.e., because of pride of shyness) or unwarranted personal abuse of PFA interventions (e.g., work stoppage).
+              “I didn’t witness it, it’s hearsay, but people were somehow arranging to take advantage of the peer helping program to get a day off, spend two, three hours talking.”
+            
+              “People’s openness to this is good. Of course, I didn’t make a hundred interventions, but people were open in the ones I did; they were understanding and communicative.”
+            All participants described impact arising from PFA interventions after one year of implementation, allowing researchers to identify facilitators and barriers for adherence to PFA intervention over time. Some of these effects directly impact peer helpers, while others are more relevant for recipients.The majority of participants mentioned the extent to which PFA intervention created an additional workload for them, as peer helpers, regularly infringing on their personal time. Unreasonable workloads may lead to some peer helpers feeling depleted, possibly affecting their involvement in providing PFA intervention in the long term.
+              “Another point that I would improve, but I don’t know how … Maybe it’s the workload it gives us. I give an example: at one point, I met four paramedics at the same time. After that, contact all four people 24 to 48 h later. I know that’s part of our role; it comes with it, you can’t pass it up, but I don’t know if there’s a way maybe to lighten it up.”
+            A few participants reported knowing peer helpers who were overly solicited for PFA interventions due to the uneven distribution of peer-support workload. In their opinion, this may have led to psychological fatigue or secondary traumatic stress reactions and therefore endangered their involvement as peer helpers over time.
+              “There are peer helpers who have been overused. Was it because they were often available? Because the system was solicitating them all the time? Because people did not advertise themselves as peer helpers while some did? As a result, they were the ones who received calls for peer helpers all the time. I can say that some colleagues were overly solicited, and, in my opinion, this contributed to burnout.”
+            Many participants brought forward the idea that the training they received in the PFA program positively influences the way they routinely interact with their co-workers (i.e., authenticity). They also appeared to be more prevention-focused and careful about co-workers’ needs and limits when interacting with them. Therefore, the PFA intervention appeared to have improved informal psychosocial support for workers after a traumatic event, possibly encouraging future participation in mental health interventions.
+              “If we see that there hasn’t been a traumatic event, but we see that there is a co-worker who is not doing well, […] we ask him if he wants to talk about it, then we can get pulled out of work if the person agrees to talk about it […]. They give us tips in training to detect people who would have problems, signs that would show a co-worker’s psychological difficulty, and try to see that. If they don’t want to talk to us, we say that we can change help out.”
+            One participant even reported that the PFA training positively impacted their interactions with patients, encouraging him to maintain his participation as a peer helper.
+              “It helped me in my work [as a paramedic], sometimes we tend to extrapolate a little too much, but focusing mainly on listening and a little less on dialogue (is important). Yes, it has served me well.”
+            Finally, some participants indicated how PFA intervention after a traumatic event brought to light workers’ demands for psychological support and behaviors to seek this support. It appears that, by reducing stigma and barriers to help, PFA intervention may favor adherence toward mental health interventions among EMS workers.
+              “In my opinion, by creating this project [PFA intervention], we discovered many people who needed a lot of help. Initially, it was clear that it was for high-stress incidents, but soon we realized other needs.”
+            Participants highlighted several elements falling under the heading of organizational support toward PFA interventions. It appeared essential to participants that the organization demonstrates endorsement of, support for, and commitment toward the PFA program and its peer helpers to ensure long-term success.A few participants named the importance of symbolic or financial recognition for their role as peer helpers. They would like to see the volunteer work they have accepted as a peer helper being recognized and valued by the organization in order to maintain their commitment over time.
+              “Peer recognition is rewarding. Though, I wonder if recognition by the organization for either an improvement or an identification more. You know there are titles for everything at [Name of the organization]. I wonder if that could be something interesting or something to consider.”
+            Most participants valued the clinical support they receive from the organization’s psychologist and colleagues when they have questions about their intervention with a recipient. However, clinical support is not sufficient for some of them and is not easily accessible for every peer helper. Some participants call for more regular supervision to facilitate the pursuit of their practice as PFA peer helpers.
+              “I am lucky to meet regularly [the psychologist] and exchange quickly on questions or issues, but, on the other hand, I don’t think that everyone has this chance. I wonder about more meetings, closer follow-ups with her, in a group, for example […]. I think that latitude is peer helpers’ greatest strength and weakness, but you’re also left on your own, and that creates a certain emptiness.”
+            The majority of participants expressed their willingness to maintain or even develop their skills as PFA peer helpers. Indeed, they suggested that the organization provide them with continuing training for PFA core actions and additional training to ensure that the quality of their interventions with co-workers is high. From a monitoring perspective, several participants reported the need for more feedback from the program, such as the number of interventions per year, impacts of their interventions, adjustments to be made to help them maintain their motivation as peer helpers.
+              “It would be nice to make “wrap-ups” of the year, quarterly “wrap-ups.” To say: What worked well? What didn’t work well? What should be adjusted? Whether it’s every six months, or whether we meet once a year and say: “This year, there were that many interventions…”, “there was that much business…”. We had periods of overdraft; what could we do to cover them? I would have liked us to look back at real cases, find out how others did it and what the strengths and weaknesses were so that we could refine our interventions.”
+            Most of the participants highlighted that the organization’s hierarchy (from managers in the field to more senior managers) is currently supportive of this new practice. Massive efforts appeared to be made in terms of technicality (i.e., road clearance, dedicated room) to ensure that PFA intervention is and remains a priority and an essential component of the organizational services offered to EMS workers. This favorable positioning of the hierarchy is perceived to facilitate their work as PFA peer helpers.
+              “Most of the time, on the job, we are free to do the interview and the report without really any problem, it’s unbelievable! Never a problem. I have even been removed from Priority 1 [First priority call] on occasion to be a peer helper. It’s really at the top of the ladder; I felt like it was at the top of the assignment ladder … the leaders never have a problem offering us a room or an environment.”
+            However, most participants also noted episodes of interference by some managers making it more difficult for them to provide PFA interventions adequately. They reported a rejection of intervention from some managers, intrusive manager behaviors in the peer helper/recipient relationship, or uncertainty about procedures changing from shift-to-shift depending on the manager. Such inequalities in treatment may lead to frustrations among peer helpers and recipients who may be less likely to adhere to PFA principles as a result.Adherence to PFA intervention was also influenced by the level of congruence that the intervention holds with professional culture (e.g., EMS) and the specific organizational culture.The vast majority of participants described how they perceived PFA intervention to be compatible and conforming to EMS culture. For some, PFA intervention falls in line with their professional culture. Participants reported that PFA intervention is quick and easy to provide as it can be performed anywhere, facilitating participation for these workers. Moreover, PFA core actions are described as being relevant for EMS workers’ characteristics.
+              “We are not people who are strangers to the intervention, so for us, it was easily assimilated, I think. When you have a certain number of years of experience, putting words and gestures on intervention is much easier than “Mr. and Mrs. Everybody” […]. We do a lot of psychological interventions within the population, so offering psychological interventions with co-workers is still part of the subject.”
+            On the other hand, for some of the participants, PFA intervention may have led to role conflict between EMS workers and peer helper positions. Core actions of a peer helper are sometimes incompatible with professional obligations of EMS workers, particularly regarding detection and reaction to physical symptoms (i.e., hyperventilation, hypertension) or the report of suicidal thoughts or high-risk behaviors. This discrepancy with their EMS-worker role and responsibilities may lead to cognitive dissonance or a loyalty conflict, which can, in turn, be detrimental to their adherence to the PFA intervention process over time.
+              “In theory, if I do an intervention, it remains confidential at the peer helper level. However, I still keep my paramedic hat on if there are clinical aspects, such as chest pain. I have to assess the chest pain. When I finish the procedure, I still have to fill out a report that the person has refused transport.”
+            Beyond the EMS culture, several participants indicated that adherence to PFA intervention is made difficult by the specificity of their relation to the organizational structure and past organizational responses to the psychological distress of EMS workers. Participants mentioned the extent to which some members of the EMS workers reject PFA intervention due to bitterness or lack of trust toward the organization that offers it due to past organizational response.Using a qualitative inductive approach, the present findings reveal factors that may foster or hinder adherence regarding participation in PFA intervention among EMS workers, according to peer helpers’ perspectives. Researchers identified four themes and 11 subthemes influencing adherence to PFA intervention: (1) individual perceptions and attitudes of peer helpers and recipients about PFA intervention; (2) perceived impacts on peer helpers and recipients; (3) organizational support for PFA intervention; and (4) congruence with occupational culture.With a cross-sectional lens, it appears that some factors can influence the adherence of both peer helpers and recipients, such as the congruence of PFA intervention with EMS culture and organizational culture as well as the approval of higher management. Other factors appear more specifically related to peer helpers’ adherence, such as perceptions and attitudes regarding the adaptability of the intervention and their sense of self-efficacy, the additional workload and mental load for peer helpers, and the recognition of the peer helper role. The need for more supervision, training, and monitoring is also likely to influence adherence of peer helpers. Finally, it appears that some factors can influence recipients’ adherence, such as perceptions and attitudes regarding the matter of credibility and trust and the perceived impacts concerning improved informal psychosocial support and the unveil psychological support needs of EMS workers. Adherence of both peer helpers and recipients is essential for PFA intervention to exist and be sustained over time in an organizational context, so it may be useful to categorize the identified subthemes in accordance with these two categories of workers; see Figure 2. Specific guidelines for each category could then emerge.Participants identified factors that influence their adherence to PFA intervention and that help maintains their adherence over time. The question of adherence is, therefore, in the middle ground between the concepts of implementation and sustainability [33,46]. Indeed, the factors inductively gathered in the present study are consistent with the literature in both directions.Present results regarding individual perceptions and attitudes of peer helpers and recipients suggested elements that could influence adherence to the intervention over time. For peer helpers, perceptions and attitudes regarding adaptability and self-efficacy, such as confidence to provide, the feeling of being sufficiently equipped, and providing an adapted intervention, seems to be crucial elements to maintain their participation as peer helpers over time. These are the most studied elements regarding the delivery of evidence-based psychosocial interventions [47], especially in PFA intervention [28,29,30,31]. This suggests they are the first and most essential factors organizations should consider regarding adherence to PFA intervention. For recipients, concerns about credibility and trust toward this intervention may lead to rejection or abuse and thus affect their adherence over time. In accordance with the present study, Forbes stated that PFA providers should feel confident in applying PFA, PFA core actions should be well implemented, confidentiality should be ascertained, and recipients should feel positively supported by PFA intervention [27]. Those results are also congruent with a recent systematic review of influences on implementation of peer support work for adults with mental health problems [48] as well as the first level (i.e., use of the intervention) of the Dynamic Sustainability Framework (DSF), which was developed for health interventions and implemented in organizations [46].Further looking at the first level of the DSF [46], early perceived impacts were described on peer helpers and recipients after one year of PFA intervention. The first perceived impacts appear to be additional workload and mental load for peer helpers. Being overused and more exposed to traumatic content, by providing PFA intervention to colleagues, may lead to greater post-traumatic stress symptoms, as reported in some studies on peer support in high-risk organizations [15,49]. This may well jeopardize their adherence to FPA intervention as providers in the long run. Those results are also in accordance with Richins’s scoping review [12], which stated that early post-trauma intervention success is increased when specific need and logistical issues (i.e., workload) are identified and overcome. On the other hand, participants reported an improvement in their capacity to offer informal psychosocial support. They explained that they were more attentive to behavioral and mood changes in co-workers, more active in suggesting support, and providing a more empathetic, authentic, and listening ear. This positive impact may be linked to a generalization of the interpersonal skills learned during PFA training [28,50]. They are known to lead to a better form of support [51], facilitating recipients’ adherence to mental health intervention in the long run. As the last perceived impact, PFA intervention by peer helpers appeared to have unveiled psychological support needs among EMS workers. These last two positive perceived impacts suggest that frequent use of the intervention over time (e.g., adherence to PFA intervention), using a peer helper model, may be a good way to reduce stigma and barriers to mental health care in first responder organizations [12,15].In findings related to overall organizational support for PFA intervention, participants described three main aspects of organizational support: recognition of the peer helper’s role, supervision, training, and monitoring of PFA intervention, and favorable position of the hierarchy. In Richins’ review, a synthesis of study outcomes found that early post-trauma interventions help emergency responders manage post-traumatic events when they are delivered in a manner supported by organizations [12]. These empirically induced results also echoed the model proposed by Forbes. Forbes and colleagues [27] suggested that organization policy supporting PFA interventions, organizational procedures, regular supervision, training follow-ups, and monitoring providers’ activity ensures a successful implementation. The theme is also congruent with the second level of the DSF (Practice setting). This level focuses on setting characteristics such as policies and procedures, human and capital resources, providing training and supervision [46]. Finally, peer-support literature also underlies that training and detailed role definition favor successful program implementation [48].With respect to the last theme related to congruence with occupational culture participants, it was indicated that PFA intervention appears compatible with EMS occupational culture, because of its quick and flexible nature. However, they reported some tension with EMS workers’ usual role and responsibilities (i.e., reaction to physical symptoms) that need to be solved to ensure adherence over time. They also mentioned that specific organizational culture elements (i.e., lack of trust due to past organizational response toward mental health) are helpful in ensuring adherence to a new and disruptive intervention. Addressing distinctive organizational culture is also described by Richins and colleagues [12] as an important factor in order to make early post-trauma intervention models successful in emergency and other high-risk organizations. Likewise, organizational culture was identified by Ibrahim and colleagues [48] as an important factor of influence in peer support intervention’s implementation. Congruently, in Forbes’s implementation model, the culture and context within which PFA will be delivered are highlighted as an important factor to evaluate [27]. This theme also echoes Chambers and colleagues’ [46] ecological system level of the DSF, as this DSF level includes general policies or population characteristics that act as supplementary drivers for the successful sustainability of a mental health intervention.With regard to limitations of the present study, we must recall that, although the objective of qualitative research is by no means the generalizability of its results [40], all participants come from a single EMS organization, which limits the widespread use of the present results. Organizational diversification would be critical to understanding the outcome as we found that organizational support and culture influence adherence to the PFA intervention. In addition, the transferability of the results needs to be supported by further studies to assess whether our themes are applicable to other first responders such as firefighters or police officers, and whether they can be applied to different types of mental stress in this population. A further limitation may be the opportunistic sampling strategy [36], which may have led to a selection bias. Indeed, volunteer selection may enhance the fact that those who choose to participate may share a characteristic that makes them different from non-participants leading to possible blind spots in the results. The self-report method, as well as the relational dynamics between the interviewer and the interviewee, may have led to social desirability effects that can affect the results. Moreover, deliberate inclusion in the sample of negative cases might have enhanced the quality of ours results leading to new perspectives on adherence to PFA in this context. Despite these limitations, the present results contribute to the literature regarding implementation and adherence to PFA intervention. Specifically, this study has good ecological validity as results came from peer helpers’ perspectives after one year of PFA implementation in an EMS setting. In conclusion, exploration of peer helpers and recipients’ adherence appear to be particularly relevant in studying PFA within an organizational context, as suggested by previous studies [26,52]. It will inform the organization on how to modify PFA implementation to allow to improve adherence and ensure sustainability in the long run. It may provide guidelines for adjustment in order to ensure that adherence is maintained over time, as the conditions of adherence must be present for the intervention to be sustainable [53]. Moreover, some of the themes are in line with Forbes’s implementation model of PFA in high-risk organization [27], and our themes appear to fit with the Dynamic Sustainability Framework [46] to reflect the experience of EMS peer helpers, when it comes to exploring the factors that influence adherence to PFA intervention over time. Therefore, this model appears relevant for use in future studies to optimize adherence to PFA intervention.These study findings suggested that it is possible to act on various factors to improve peer helpers and recipients adherence to PFA implementation and ensure sustainability in EMS organizations. First, individual perceptions and attitudes of peer helpers and recipients about the PFA intervention could be worked upon by focusing on confidence for peer helpers, credibility, confidentiality, and feeling supported for recipients. Second, perceived impacts of PFA intervention should be monitored for additional workload, additional mental load, and improvement in informal psychosocial support for peer helpers, and unveiled psychological support needs among recipients. Third, organizational support should provide role recognition, supervision, training, and monitoring for PFA intervention and favorable and sustained positioning of the hierarchy. Fourth, congruence with occupational culture should be considered and strengthened. These findings will help inform both program and research design for evaluating such a program for other first responder organizations worldwide. They suggest that PFA intervention’s implementation should be flexible, tailored to the setting-specific needs, and refined if needed to favor adherence over time.Conceptualization, M.T. and S.G.; methodology, M.T.; validation, M.T. and S.G.; formal analysis, M.T., J.L. and S.G.; investigation, M.T.; data curation, M.T.; writing—original draft preparation, M.T.; writing—review and editing, S.G., J.L. and M.T.; visualization, M.T. and S.G.; supervision, S.G.; project administration, M.T.; funding acquisition, S.G. and M.T. All authors have read and agreed to the published version of the manuscript.This research was funded by the Canadian Institutes of Health Research, Catalyst Grant: Post-Traumatic Stress Injuries among Public Safety Personnel/Subvention, grant number: PPS-162535. The first author received a scholarship from the Fonds de Recherche du Québec-Société et Culture (FRQSC), grant number: B2Z-299287. This study was also funded by a career grant awarded to the last author by the Fonds de recherche du Québec—Institut Robert-Sauvé en Santé et Sécurité au Travail, grant number: 268274. The APC was funded by Canadian Institutes of Health Research.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Integrated University Health and Social Services Centre for the East Island of Montreal (protocol code CER-CEMTL 2019-1884, 2019-05-14).Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the subjects to publish this paper.To protect the confidentiality of participant information, the University Institute of Mental Health of Montreal will not allow the authors to make data publicly available. Data are available upon request from Marine Tessier at Trauma Studies Center, University Institute of Mental Health of Montreal, for researchers who meet the criteria for access to confidential data.The researchers would like to thank the peer helpers who participated in this study as well as the EMS organization that entrusted us with this study, and especially Josée Coulombe, Thérésa Choisi and Luc de Montigny, for their help and support. 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.Themes and subthemes following thematic analysis.Themes and subthemes graded by category of workers.Participant demographic and professional information.Note: N = 11.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11027.txt

@@ -0,0 +1,7 @@
+With the development of science and technology, buying has become much easier. At the same time, however, impulsive buying has many negative consequences for college students, such as dissatisfaction and debt; the causes of impulsive buying should, therefore, be explored urgently. There are numerous empirical studies indicating that social exclusion may be a potential factor of impulsive buying, and the underlying mechanisms of this association remain unclear. In this study, we used the Social Exclusion Scale, Self-Esteem Scale, Risk Preference Scale, and Impulsive Buying Scale, as well as a cross-sectional design to investigate the roles of self-esteem and risk preference in the relationship between social exclusion and impulsive buying among 768 college students (387 were female, Mage = 20.25 years). The results were as follows: (1) when controlling for gender, age, family monthly income, and monthly living expenses, social exclusion significantly and positively predicted impulsive buying; (2) self-esteem played a mediating role between social exclusion and impulsive buying; (3) risk preference moderated the relationship between the second half of the mediating path and the direct path. These results reveal the mechanism underlying impulsive buying in college students, that is, social exclusion will predict the decrease in college students’ self-esteem, and low self-esteem will further predict college students’ impulsive buying, which is a way for them to gain a sense of self-worth. Relatively low risk preference can well alleviate the negative impact of social exclusion and low self-esteem on impulsive buying. What is more, these results have implications for impulsive buying interventions and preventions. Schools should aim to create a good peer atmosphere by implementing certain rules that help to reduce social exclusion, and parents and education departments should cultivate students’ risk awareness to avoid risk behaviors in college students, such as impulsive buying behavior. This study fills the research gap regarding college students’ impulsive buying and explores its internal psychological mechanism.Shopping has become an indispensable part of college students’ lives. According to an official report, the proportion of Internet users who shop online has reached 79.1% in China [1]. Most of the college students are still in their late adolescence and they have gradually become increasingly targeted by marketers [2]. Moreover, with the rapid development of science and technology, such as online shopping and express service, shopping has become even more convenient. However, it has also caused some problems for college students, one of which is impulsive buying. In China, college students have more disposable money than junior high school students and senior high school students, but the correct concept of consumption has not been fully formed, so the possibility of impulsive buying is also higher. What is more, a large number of previous studies have shown that social exclusion will have a serious and far-reaching negative impact on college students, such as internet addiction [3] and depression [4]. Therefore, it is of great significance to explore the relationship between social exclusion and college students’ impulsive buying.Impulsive buying is defined as sudden and unplanned buying behavior that is driven by a strong and persistent impulse, after which consumers experience a series of emotional, cognitive, and/or behavioral traits [5]. Consumers are now more inclined than ever to be utilitarian and engage in hedonic buying [6], and they enjoy the feeling of shopping more than buying what they really need [7]. While impulsive buying can bring immediate enjoyment and satisfaction [8], it is also closely related to some adverse consequences, such as low self-esteem, dissatisfaction [9], and debt [10]. The changes in an individual’s emotion and cognition following impulsive buying can lead to the subsequent recurrence of impulsive buying [5]. College students in late adolescence, who have not yet achieved mature thinking or economic independence, might engage in continuously excessive and uncontrollable buying if they have an improper shopping style, which could, in turn, cause more negative consequences, as mentioned. Although a large number of previous studies have discussed the negative consequences of impulsive buying on college students, its internal mechanism is not clear. Therefore, it is vital to explore the causes of impulsive buying among adolescences to intervene and prevent it.Ecological systems theory regards development as a process of “individual–environment” interaction [11,12]. For adolescents, peers are the group environment with which they interact, and which might have a direct or indirect impact on their behavior and mental health, manifesting in antisocial behavior and depression symptoms [13]. For college students in late adolescence or just after adolescence, it may have a similar effect. Social exclusion as a kind of peer relationship deserves attention as a potential cause of college students’ negative behaviors or moods [14]. Social exclusion is a negative social phenomenon that manifests as exclusion, isolation, and rejection. Being excluded might stop individuals from developing relationships and pursuing a sense of belonging [15]. From the perspective of emotion and mood, previous studies have shown that individuals described experiencing pain when they were socially excluded, even though they were not physically injured [16]. College students might tend to distract and relieve themselves from this pain by engaging in other activities, such as Internet or alcohol use [17,18,19]. According to Twenge et al. [20], social exclusion makes individuals unconsciously choose out-of-control behaviors, such as high-risk and unhealthy behaviors.As a high-risk behavior, impulsive buying is also highly likely to be predicted by social exclusion. Indeed, given that impulsive buying can make individuals feel immediately satisfied [8] and has the characteristics of high-risk behavior, which could lead to many negative consequences [10], it might be caused by social exclusion. Previous studies have also shown that customers can improve their negative mood by immediately buying products that bring satisfaction [21]. From a cognitive perspective, individuals may use cognitive resources to repair the negative effects of social exclusion, such as low self-esteem [22]. As a result, individuals might invest fewer cognitive resources into cognitive tasks, which affects their reasoning and decision-making abilities, and further increases impulsive buying. Furthermore, social exclusion can impair an individual’s self-regulation ability [23] and weaken their ability to inhibit impulsive behaviors [24], including impulsive buying. In addition, social exclusion could weaken intelligent thought [25]. In that case, individuals’ reasoning might be less rational, such that they are more easily controlled by their emotions and resulting in impulsive behaviors, such as impulsive buying [9]. Therefore, we proposed the following hypothesis:
+            Social exclusion has a positive predictive impact on impulsive buying.
+          Although social exclusion may be a direct predictor of impulsive buying, more research about the internal mechanisms, such as mediating influences, is needed to improve our understanding of impulsive buying and aid the development of effective interventions. Based on social exclusion theory and sociometer theory [14,26], individual’s interpersonal relationship will be reflected through self-esteem and individuals are eager to be accepted by peers to get a sense of belonging and security. A lack of peer group acceptance will lead to a series of negative consequences, such as anxiety, loneliness, depression, and low self-esteem [14]. Self-esteem is an attitude one has toward one’s self, and it is a mental representation of self-worth and self-acceptance [27]. Previous research has found that social exclusion induced by the Cyberball paradigm is also associated with decreased self-reported self-esteem, as well as reduced implicit self-esteem [28]. This shows that negative social feedback can affect an individual’s self-esteem [29]. Previous studies have also found that self-esteem is closely related to an individual’s interpersonal relationships [30], and that student–student relationships will directly affect individual self-esteem [31]. Therefore, it can be speculated that social exclusion might directly impact college students’ self-esteem. Moreover, low self-esteem might lead to self-doubt and a lack of self-confidence, such that college students might pay more attention to being accepted by society. To dissipate the negative emotions caused by low self-esteem, college students may gain self-worth through impulsive buying to compensate for a lack of social relationships [32]. Furthermore, Dittmar et al. [33] reported that buying products has become a way to obtain and express a sense of self-identity. College students who are excluded from society might buy impulsively to achieve a level of self-expression and form social ties through shopping [34]. Empirical research has also found that self-esteem mediates the relationship between mindfulness and impulsive buying tendencies [35]. Therefore, we proposed the following hypothesis:
+            Self-esteem will mediate the relationship between social exclusion and impulsive buying.
+          Although social exclusion may significantly impact college students’ impulsive behavior, not all college students with high levels of social exclusion will develop impulsive behaviors. According to the ecological systems theory [11,12], individuals’ development stems from the interplay between the environment (such as their social environment) and their intrapersonal characteristics (such as risk preference). Risk preference is another potentially important factor that influences impulsive decision-making and risk decision-making, and may, thus, affect impulsive buying. Risk preference refers to a person’s preferred reaction in the face of risk choice and safety choice [36,37]. High risk preference is usually associated with non-adaptive and impulsive behaviors, including drinking, taking drugs, smoking, gambling, and engaging in unsafe sexual behaviors [38,39,40]. As an impulsive behavior that can bring immediate satisfaction [8], impulsive buying also has certain risks; individuals often need to spend more money to get temporary happiness, and this process may be affected by differences in individual risk preference. Farley [41] has divided consumers into risk-seeking and risk-averse consumers. Risk-averse consumers focus on minimizing risk; their actions lead to hesitation and consideration. In other words, they seek security and stability. On the contrary, risk-seeking consumers are willing to face risks [42]. College students with high risk preference may make more impulsive decisions, such as impulsive buying, and prioritize the short-term benefits of impulsive buying over the losses. Therefore, college students with high risk preference who are excluded by society may find it more difficult to inhibit impulsive buying. However, college students with low risk preference may engage less in impulsive buying and may pay more attention to the negative effects brought by impulsive buying, such as short-term economic debt [8]. Even if these individuals experience high levels of social exclusion, they will consider the negative consequences of impulsive buying so as to restrain this behavior. To a certain extent, low risk preference may, therefore, restrain impulsive buying. We hypothesized that low risk preference is a positive factor that influences impulsive behavior and can largely overcome the negative impact of social exclusion and a lack of self-esteem. Previous research has also shown that low risk preference is directly related to high self-control [43] and that high self-control can inhibit impulsive behavior [44]. Therefore, we proposed the following hypothesis:
+            Risk preference will moderate the direct and indirect link between social exclusion and impulsive buying. Specifically, the indirect association and the direct association between social exclusion and impulsive buying via self-esteem will be stronger in college students with high risk preference and will be weaker in college students with low risk preference.
+          In summary, this study proposed a moderated mediation model to explore the internal psychological mechanisms underlying the effect of social exclusion on impulse buying. The present results enhance our understanding of the mechanism underlying impulsive buying and provide a theoretical basis for the development of preventative measures and interventions for impulsive buying in college students. Figure 1 illustrates the proposed research model.In this study, 811 college students from Guangzhou University, Guangdong University of Technology, Guangdong University of Finance and Economics, and other schools were surveyed using convenience sampling. The survey was conducted anonymously. After completing the questionnaire, respondents received a reward of CNY 1. The valid sample used in the analysis comprised 768 respondents (94.7% response rate; Mage = 20.25 years, SD = 1.52 years), of which 387 (50.4%) were female (Mage = 20.24 years, SD = 1.53 years).The research materials and procedures were approved by the ethics committee of Guangzhou University (protocol code: GZHU2019007; date of approval: 27 May 2019). In this study, the data were collected between 9 November and 9 December 2020. Before the formal test, the data collectors informed the participants that participation was voluntary and that they can refuse to answer questions if they feel uncomfortable. Participants were assured that their responses would be kept confidential and that they would only be used for academic research.Mediation and moderation effects were tested with Mplus 8.3 (Muthén and Muthén, Los Angeles, CA, USA) [45]. Bootstrapping analysis with 5000 replicates was performed to verify the significance of the paths. If the confidence interval does not include 0, the path coefficient is significant. A model fit is considered acceptable when χ2/df is less than 5, CFI and TLI are greater than 0.90, and when RMSEA is less than 0.08, according to Hoyle’s suggestion [46]. Age, gender, family monthly income, and monthly living expenses were included in both models as control variables. We bootstrapped with 5000 samples to generate bias-corrected 95% confidence intervals. If the confidence interval excludes 0, it indicates that the parameter is statistically significant.The social exclusion questionnaire of college students was developed by Wu et al. [47]. The scale is divided into two dimensions, namely direct exclusion and indirect exclusion, and includes a total of 19 items, such as “others speak ill of me behind my back and influence other people’s views about me” (direct exclusion) and “my mistakes are coaxed or impolitely criticized” (indirect exclusion). College students were asked to report how often they experienced these situations using a 5-point Likert-type scale from 1 = very inconsistent to 5 = very consistent. The average of all items was calculated for the total score, and a higher score indicated a higher social exclusion level of college students. In this study, the scale demonstrated excellent reliability (α = 0.96).We used the Self-Esteem Scale compiled by Rosenberg [27] to measure self-esteem in college students. The scale includes 10 items, such as “I am able to do things as well as most other people” and “I take a positive attitude toward myself”. College students were asked to respond to items on a 4-point Likert-type scale, from 1 = highly agree to 4 = highly disagree. Reverse coding was used for some item scores and the average of all items was calculated. A higher score indicated a higher level of self-esteem in college students. In this study, the scale demonstrated excellent reliability (α = 0.86).The 14-item risk preference questionnaire developed by Hsee and Weber [36] was used to assess the risk preference of college students, including seven profit scenario items and seven loss scenario items. Each item has two possible responses that represent a conservative choice and risk-taking choice, such as “A: 100% probability to get CNY 400, B: 50% probability to get CNY 2000, 50% probability to get CNY 0” (profit situation) and “A: 100% probability to lose CNY 600, B: 50% probability to lose CNY 2000, 50% probability to lose CNY 0” (lose situation). The probability of choosing risk option B (risk score) was used as the index of individual risk preference, that is, risk preference = the number of B/14. A higher score indicates a stronger risk preference. In this study, the scale demonstrated good reliability (α = 0.78).The impulse buying intention scale was developed by Jing et al. [48] and assesses the six following dimensions: impulse buying, mood regulation, purchasing experience, consideration of the future, quick decision-making, and unplanned decision-making. There are 26 items in total, such as “I want to get what I like immediately”. College students were asked to respond to items on a 4-point Likert-type scale ranging from 1 = not at all to 7 = exactly. Some item scores were reversed and the average of all items was calculated. A higher average score indicated a stronger impulsive buying tendency. In this study, the scale demonstrated excellent reliability (α = 0.92).Previous studies have shown that gender, age and income are the important factors affecting impulsive buying [49,50]. In addition, because the income of Chinese college students is linked to their family monthly income and monthly living expenses, we have a certain degree of statistical control on them. In the present study, gender was dummy coded (0 = female, 1 = male). Family monthly income and monthly living expenses were both divided into four levels.We conducted a Pearson’s correlation analysis on the total average scores of social exclusion, self-esteem, impulsive buying, and risk preference. Means, standard deviations, and Pearson’s correlations (r) were calculated for all study variables in Table 1. The results show that impulsive buying was positively correlated with social exclusion (r = 0.36, p < 0.001) and risk preference (r = 0.12, p < 0.001). Self-esteem was negatively correlated with social exclusion (r = −0.43, p < 0.001) and impulsive buying (r = −0.22, p < 0.001). The means and standard deviations of the four main variables are as follows: social exclusion (mean = 1.83, SD = 0.65), impulsive buying (mean = 3.00, SD = 0.85), self-esteem (mean = 2.74, SD = 0.46), and risk preference (mean = 0.41, SD = 0.22). These findings suggest that social exclusion and low self-esteem may be predictive factors of impulsive buying and that low risk preference may be a protective factor of impulsive buying.The mediation model represented in Figure 2 revealed an excellent fit to the data: χ2 = 3.37, df = 2, χ2/df = 1.68, CFI = 0.99, TLI = 0.98, RMSEA = 0.03. The results are displayed in Figure 2. Social exclusion negatively predicted self-esteem (b = −0.29, SE = 0.02, p < 0.001, 95% CI = [−0.34, −0.24]) and significantly positively predicted impulsive buying (b = 0.45, SE = 0.05, p < 0.001, 95% CI = [0.35, 0.56]). Self-esteem negatively predicted impulsive buying (b = −0.17, SE = 0.07, p < 0.05, 95% CI = [−0.31, −0.03]). Moreover, bootstrapping analyses indicated that self-esteem mediated the relationship between social exclusion and college students’ impulsive buying (indirect effect = 0.05, SE = 0.02, p < 0.05, 95% CI = [0.01, 0.09]). As covariates, gender, age, family monthly income, and monthly living expenses were included in the regression equation for control.The moderated mediation model represented in Figure 3 displayed a good fit to the data: χ2 = 36.28, df = 13, χ2/df = 2.79, CFI = 0.93, TLI = 0.91, RMSEA = 0.05. The bias-corrected percentile bootstrap results indicate that the indirect effect of social exclusion on college students’ impulsive buying through self-esteem was moderated by risk preference. Specifically, risk preference moderated the association between self-esteem and impulsive buying (b = −0.70, SE = 0.29, p < 0.05, 95% CI = [−1.29, −0.15]) and the association between social exclusion and impulsive buying (b = 0.36, SE = 0.18, p < 0.05, 95% CI = [0.00, 0.70]). As covariates, gender, age, family monthly income, and monthly living expenses were included in the regression equation for control.In order to further understand the essence of moderation, a simple slopes test was conducted in this study, and, as depicted in Figure 4 and Figure 5, the negative link between self-esteem and impulsive buying was much weaker for college students with low risk preference (1 SD below the mean; b = −0.05, SE = 0.10, p > 0.05, 95% CI = [−0.24, 0.14]) than college students with high risk preference (1 SD above the mean; b = −0.35, SE = 0.09, p < 0.001, and 95% CI = [−0.53, −0.17]). What is more, the positive link between social exclusion and impulsive buying was weaker for college students with low risk preference (1 SD below the mean; b = 0.36, SE = 0.07, p < 0.001, 95% CI = [0.22, 0.50]) than college students with high risk preference (1 SD above the mean; b = 0.51, SE = 0.05, p < 0.001, and 95% CI = [0.41, 0.61]).Moreover, the positive indirect links between social exclusion and impulsive buying via self-esteem were much weaker for college students with low risk preference (indirect effect = 0.01, SE = 0.03, p > 0.05, 95% CI [−0.04, 0.07]) than for college students with high risk preference (indirect effect = 0.10, SE = 0.03, p < 0.001, 95% CI = [0.05, 0.16]).Previous empirical research has revealed there to be a relationship between social exclusion and impulsive buying. However, the internal mechanism underlying this relationship has remained unclear. Based on social exclusion theory and ecological systems theory, this study revealed the mechanism underlying the influence of social exclusion on impulsive buying. The results of this study demonstrate that social exclusion affects impulsive buying through the mediating effect of self-esteem and that the second half path and direct path of this mediating process are moderated by risk preference. The indirect and direct associations between social exclusion and impulsive buying via self-esteem were stronger among college students with high risk preference and weaker among college students with low risk preference. These research results have important theoretical significance and practical value for the prevention and intervention of impulsive buying.The results support the conclusion that social exclusion has a negative impact on individuals [25]. They are consistent with those of previous studies, verifying the negative impact of social exclusion on individuals as a negative factor. Previous studies have shown that social exclusion can lead to loneliness [51], depression, and anxiety [14], among other problems. College students with these problems may try to reconstruct their relationship with society by consuming luxury goods so as to enhance their sense of existence and superiority and alleviate the anxiety caused by social exclusion [52]. Combined with the results of the present study, these findings indicate that college students who are excluded from society hope to obtain instant happiness and satisfaction through impulsive buying, which makes them feel that they have reestablished contact with society. This kind of satisfaction can heighten their mood [21], which further shows that the negative impact of social exclusion has cross-domain consistency. Impulsive buying may be a compensatory mechanism for college students in the situation of social exclusion. In daily life, socially excluded college students find it more difficult to achieve a sense of existence and satisfaction. Therefore, impulsive buying makes it possible to heighten mood and gain a certain sense of value. To summarize, social exclusion is an important predictor of impulsive buying.The present study found that self-esteem played a mediating role in the relationship between social exclusion and impulsive buying. Social exclusion not only directly affected impulsive buying but also indirectly affected it through self-esteem. These results indicate that social exclusion can lead to the decline of college students’ self-esteem, which will, in turn, lead to more buying behavior.The results of this study are consistent with those of previous studies. Namely, social exclusion has been reported to affect college students’ self-esteem, reduce their sense of self-worth [14], and make them more eager to get in touch with the outside world. Adolescence is a critical period of individual psychological development, during which the influence of social relationships is crucial, even more so than that of family relationships [53]. College students in late adolescence or just after adolescence might also be more influenced by social relationships than family relationships. On the one hand, college students are eager for the acceptance and respect of peer groups; on the other hand, they can lack the ability to build strong social relationships, and so may instead gain a sense of value through impulsive and risk-taking behaviors. Social exclusion causes self-doubt and lower self-esteem in college students [14]. Compared to college students with a higher level of self-esteem, college students with lower self-esteem might seek to enhance their self-esteem; thus, they may be more likely to engage in impulsive buying behaviors. Impulsive buying can, to a certain extent, alleviate college students’ lack of a sense of value caused by lower self-esteem. Our results also show that positive self-esteem is more conducive to the ability of college students to restrain impulsive behavior, while negative self-esteem is not conducive to college students overcoming adverse reactions. Therefore, it is important to create a good social environment, which could improve the self-esteem of college students so as to avoid a series of problems. Self-esteem is of great significance to the growth and development of college students.This study found that risk preference played a moderating role between the direct path and the second half path of the impact of social exclusion on impulsive buying. Specifically, the direct impact of social exclusion on impulsive buying was stronger among college students with high risk preference, and the second half path of the impact of social exclusion on impulsive buying via self-esteem was also stronger among college students with high risk preference. That is, for college students with high risk preference, social exclusion had a stronger direct and indirect predictive effect on impulsive buying. Social exclusion predicted the process of impulsive buying through the mediating role of self-esteem, which was also stronger among college students with high risk preference, who are more vulnerable to low self-esteem and more social exclusion. This is consistent with previous studies, proving that lower risk preference has a certain positive effect [41].Previous studies have shown that risk preference is a direct predictor of risk-taking and impulsive behaviors [36,37], and that risk preference has an important influence on impulsive decisions in college students. The influence of high risk preference and low risk preference on college students are also different. Namely, college students with high risk preference consider more immediate benefits brought by their decisions, including material and psychological benefits (happiness and satisfaction); college students with low risk preference pay more attention to the negative consequences of impulsive decisions which may increase their economic burden and can lead to a series of problems, such as debt [10], and so will adopt a more conservative approach when making decisions. Therefore, we believe that low risk preference, as a positive factor, can to some extent restrain college students’ impulsive behavior and the resulting series of negative consequences. College students with low self-esteem will exhibit more impulsive buying behaviors. This could be because college students with low self-esteem want to build contact with the outside world and gain a stronger identity and sense of value; thus, they tend to engage in impulsive buying behavior. As a positive factor, low risk preference can restrain these impulses, such that college students can pay more attention to the negative consequences of impulsive buying.To summarize, it is important to educate college students about risks, especially when they make decisions that could have serious negative consequences, such as whether to engage in impulsive buying or alcohol or drug abuse. This provides a new perspective and inspiration for the development of education programs for young people. Schools and teachers should set up relevant courses to improve risk awareness in young people. We should encourage college students to not only focus on the negative consequences of impulsive behavior but also to cultivate their awareness of future prospects.While we found that social exclusion can directly predict impulsive buying, the present study has some limitations. First, although the cross-sectional design used in this study had a theoretical basis that was built on previous work and uses a self-report method, the causal relationship and internal mechanisms between variables could not be determined. Future research should choose an experimental method to test the mediating theoretical model of this study by comparing results from an experimental group and a control group. Alternatively, future research could perform longitudinal research to better test causality. Second, the subjects of this study were all college students in late adolescence or just after adolescence with small age differences, and so, the research results cannot be applied to early and middle adolescence. Future research should continue to explore the influencing mechanism underlying impulsive buying in early and middle adolescence.Limitations aside, the findings of this study have important theoretical and practical implications. This study extends the knowledge in the impulsive buying field and contributes to our understanding of the cause of impulsive buying. Specifically, we highlighted the environmental cause (social exclusion) and cognitive cause (self-esteem) of impulsive buying, and take self-esteem into account, thus enriching the ecological model of impulsive buying. Future research should further explore the affective mechanism of college students’ impulsive buying. Moreover, the findings of this study could help to guide the targeted intervention of college students’ impulsive buying. First, the government and social organizations should strive to create a harmonious social environment and reduce the social exclusion of college students, and schools should aim to create a good school atmosphere by making certain rules [54] that help to reduce social exclusion, thereby maintaining the self-esteem of college students and reducing impulsive buying. Second, parents and education departments should cultivate students’ risk awareness, because encouraging appropriate low risk preference could help to avoid many risk behaviors in college students. Third, effective social support and psychological guidance would also help college students to make more rational shopping decisions. To summarize, this study draws the following conclusions: (1) social exclusion positively predicts college student impulsive buying; (2) self-esteem plays a mediating role in the relationship between social exclusion and impulsive buying; (3) social exclusion is moderated by risk preference through the indirect effect of self-esteem on impulsive buying, and the direct effect of social exclusion on impulsive buying is also moderated by risk preference. Specifically, in the direct path, the social exclusion of college students with high risk preference has a greater predictive effect on their impulsive buying than in college students with low risk preference. In the second half of the mediating path, the impact of social exclusion on self-esteem is stronger in college students with high risk preference, and low risk preference can significantly inhibit impulsive buying behaviors caused by low self-esteem.Conceptualization, H.L. and Y.N.; methodology, H.L. and Y.N.; validation, H.L., J.C., S.L., Y.N. and G.W.; formal analysis, H.L. and Y.N.; investigation, H.L., J.C. and Y.N.; resources, H.L., J.C., S.L., Y.N. and G.W.; writing—original draft preparation, H.L., J.C. and Y.N.; writing—review and editing, H.L., J.C., S.L., Y.N. and G.W.; visualization, H.L. and Y.N.; supervision, Y.N.; project administration, H.L., J.C., S.L., Y.N. and G.W.; funding acquisition, H.L., J.C., S.L., Y.N. and G.W. All authors have read and agreed to the published version of the manuscript.This study was supported by the National Natural Science Foundation of China (32071067) and the National Innovation Training Program for College Students (202111078030).This study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the Ethics in Human Research Committee of the Department of Psychology, Guangzhou University (protocol code: GZHU2019007; date of approval: 27 May 2019).Informed consent was obtained from all the subjects involved in the study.The data presented in this study are available upon request from the corresponding author.In particular, thanks go to Guangzhou University for my education, to Yangang Nie for his inimitable support in my research career, as well as to Xiaolin Zhou and Xiaoxue Gao for their precious appreciation and recognition of me.The authors declare no conflict of interest.The proposed moderated mediation model.Model of the mediating role of self-esteem between social exclusion and impulsive buying. Values are unstandardized coefficients and the standard error. Paths of gender, age, family monthly income, and monthly living expenses in the model are not displayed. Of those paths, the following were significant: gender to impulsive buying (b = −0.20, SE = 0.06, p < 0.001, 95% CI = [−0.31, −0.10]), monthly living expenses to impulsive buying (b = 0.31, SE = 0.06, p < 0.001, 95% CI = [0.20, 0.42]), and family monthly income to self-esteem (b = 0.04, SE = 0.02, p < 0.05, 95% CI = [0.01, 0.09]). * p < 0.05, *** p < 0.001.Model of the moderating role of risk preference on the direct and indirect relationship between social exclusion and impulsive buying. RP, risk preference; SE, self-esteem; SOE, social exclusion. Values are unstandardized coefficients and the standard error. Paths of gender, age, family monthly income, and monthly living expenses in the model are not displayed. Of those paths, the following were significant: gender to impulsive buying (b = −0.20, SE = 0.06, p < 0.001, 95% CI = [−0.30, −0.09]), and monthly living expenses to impulsive buying (b = 0.31, SE = 0.05, p < 0.001, 95% CI = [0.20, 0.42]). * p < 0.05, ** p < 0.01, *** p < 0.001.Risk preference among college students as a function of self-esteem and impulsive buying.Risk preference among college students as a function of social exclusion and impulsive buying.Descriptive statistics and correlations for all variables.Note: Gender was dummy-coded: 1 = male, 0 = female; * p < 0.05, *** p < 0.001.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11028.txt

@@ -0,0 +1 @@
+(1) Background: As cities densify, researcher and policy focus is intensifying on which green space types and qualities are important for health. We conducted a systematic review to examine whether particular green space types and qualities have been shown to provide health benefits and if so, which specific types and qualities, and which health outcomes. (2) Methods: We searched five databases from inception up to June 30, 2021. We included all studies examining a wide range of green space characteristics on various health outcomes. (3) Results: 68 articles from 59 studies were found, with a high degree of heterogeneity in study designs, definitions of quality and outcomes. Most studies were cross-sectional, ecological or cohort studies. Environment types, vegetation types, and the size and connectivity of green spaces were associated with improved health outcomes, though with contingencies by age and gender. Health benefits were more consistently observed in areas with greater tree canopy, but not grassland. The main outcomes with evidence of health benefits included allergic respiratory conditions, cardiovascular conditions and psychological wellbeing. Both objectively and subjectively measured qualities demonstrated associations with health outcomes. (4) Conclusion: Experimental studies and longitudinal cohort studies will strengthen current evidence. Evidence was lacking for needs-specific or culturally-appropriate amenities and soundscape characteristics. Qualities that need more in-depth investigation include indices that account for forms, patterns, and networks of objectively and subjectively measured green space qualities.Green spaces are a crucial aspect of urban cities. They protect against many of the harmful impacts of rapid urbanisation on health. They also permit social and economic benefits by providing preferential settings for relaxation, building social connections, engaging in physical activity and feeling closer to nature, including resident wildlife [1]. Therefore, urban greening is an important strategy for addressing complex global issues such as climate change, sustainable urbanisation and health inequality. This is recognised via the United Nations Sustainable Development Goal (SDG) 11 target 7, which states “by 2030, providing universal access to safe, inclusive and accessible, green and public spaces, in particular for women and children, older persons and persons with disabilities” [2].Substantial research is dedicated to revealing the health benefits of green spaces [3]. While more green space tends to be good for health, such conclusions are not universally reported. Most research in this field tends to use measures of ‘greenness’ such as the normalised difference vegetation index (NDVI) to quantify green space exposure [4], ignoring substantial heterogeneity in the constituent qualities of green spaces that make them attractive for visiting and, in turn, support health and wellbeing. For example, green spaces may vary in terms of objectively measurable good qualities (e.g., presence of certain attractive elements, such as tree canopy, footpaths and seating) and others that are more subjective in nature (e.g., an emotional or spiritual connection to a particular green space). Bad qualities (e.g., proximity to a busy road and lack of accessibility) may discourage visitation and negate health benefits. Ignoring the constituent qualities that attract or discourage people to spend time in green spaces holds back the field from having more substantive impacts as a catalyst for improving community health and reducing inequities. Examining these qualities, both good and bad, may solve a missing link in our understanding of the relationship between green spaces and health [5].Moreover, studying green space qualities has practical implications for urban planning. Driven by rapid densification, the compact, high-density city has become the dominant urban design worldwide. Not only does a compact city warrant multifunctional green spaces that can serve its diverse citizen population. It also presents a complex set of trade-offs between green space creation, regeneration and expansion on one hand, and the development of new, often competing land-use on the other (e.g., housing, infrastructure and commercial) [6]. Within space constrained contexts, modifying qualities of existing green spaces may offers an important way to maintain and improve quality of life in urban communities.Research on the health benefits of green space qualities is still emerging and there are no consensus definition what green space quality is. We do not know which qualities can be modified, and which health benefits these modifications will bring (if any). To build capacities for research that attends to these issues, we conducted a systematic review to take stock of what research has been performed on green space qualities and health, with the broader aim of charting possible paths forward to strengthen the policy relevance of this research.This systematic review aims to:(a)Evaluate whether improving certain qualities of green space provides health benefits to the population;(b)Identify and categorise all qualities of green space that have been investigated in previous primary studies; and(c)Explore the extent of variations in design characteristics of these studies.Evaluate whether improving certain qualities of green space provides health benefits to the population;Identify and categorise all qualities of green space that have been investigated in previous primary studies; andExplore the extent of variations in design characteristics of these studies.The reporting of this review was guided by the updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [7]. This review was not registered a priori, nor was a protocol published separately.We searched the following databases for articles from inception up to 8 December 2020: MEDLINE via Ovid, Embase via Ovid, PsycINFO via Ovid, CINALH via EBSCO and Scopus. No language or publication date restriction was applied. An updated search was performed on 30 June 2021. The search was supplemented by a manual search of the reference lists from relevant systematic reviews.The search strategy was a combination of three components: (health outcomes AND green space quality AND green space types). For health outcomes, we used both generic and specific search terms to capture all dimensions of physical and mental health, drawing from previous systematic literature reviews on green space and health [8,9], obesity and physical activity [10,11], birth outcomes [12], mental health [13,14,15], puberty timing [16] and menopause [17]. For green space quality, we combined the word “quality” and other determinant terms adapted from audit tools used for assessing the physical environment of parks [18]. For green space types, we used both generic and specific search terms to capture all types of green space in both urban and rural settings. The full search strategy is available in Supplementary File S1.We included all human studies meeting the following criteria:(a)Population: green space users of all ages and genders;(b)Exposure: In the context of our review, green space quality refers to any attribute that can affect willingness to use and interaction of users with that space, including but not limited to intrinsic characteristics (size or patterns), features (vegetation, facilities or amenities), conditions (maintenance or safety) or user perception of its usefulness or quality. All types of natural and man-made green environments, including parks, streetscape greenery, urban open spaces, playgrounds, coastal parks with vegetation, etc., were included as long as they were defined by authors as green space. Studies where participants viewed digitalised renderings or photographs of green spaces without actual exposure were excluded. Studies that did not investigate any aspect of green space quality were excluded. The percentage of overall vegetation coverage and “greenness” (e.g., the normalised difference vegetation index) were not eligible as they are considered measures of green space quantity, unless specific vegetation types were analysed (e.g., tree canopy);(c)Outcomes: Studies that investigated health outcomes, including but not limited to cardiometabolic, respiratory, reproductive, neurological and psychological health, and child development, were included. Studies that only measured behaviours (park usage, park-based activity, etc.) without assessing health outcomes were excluded;(d)Study design: All observational and intervention studies, including randomised, quasi-randomised and non-randomised trials. We excluded non-English language studies, study protocols, conference abstracts, dissertations, reviews, qualitative studies, editorials, case studies and opinion pieces.Population: green space users of all ages and genders;Exposure: In the context of our review, green space quality refers to any attribute that can affect willingness to use and interaction of users with that space, including but not limited to intrinsic characteristics (size or patterns), features (vegetation, facilities or amenities), conditions (maintenance or safety) or user perception of its usefulness or quality. All types of natural and man-made green environments, including parks, streetscape greenery, urban open spaces, playgrounds, coastal parks with vegetation, etc., were included as long as they were defined by authors as green space. Studies where participants viewed digitalised renderings or photographs of green spaces without actual exposure were excluded. Studies that did not investigate any aspect of green space quality were excluded. The percentage of overall vegetation coverage and “greenness” (e.g., the normalised difference vegetation index) were not eligible as they are considered measures of green space quantity, unless specific vegetation types were analysed (e.g., tree canopy);Outcomes: Studies that investigated health outcomes, including but not limited to cardiometabolic, respiratory, reproductive, neurological and psychological health, and child development, were included. Studies that only measured behaviours (park usage, park-based activity, etc.) without assessing health outcomes were excluded;Study design: All observational and intervention studies, including randomised, quasi-randomised and non-randomised trials. We excluded non-English language studies, study protocols, conference abstracts, dissertations, reviews, qualitative studies, editorials, case studies and opinion pieces.All retrieved data were imported into Covidence (Veritas Health Innovation, Australia) to remove duplicates. Two reviewers (PYN and HR-A) independently screened all titles and abstracts in duplicate and excluded studies that did not meet the inclusion criteria. Studies that were included from title/abstract screening had their full text reviewed in duplicate by the same two reviewers and reasons for exclusion were noted. Disagreement was resolved by discussion with senior reviewers (XF and TA-B). All stages of study screening were conducted in Covidence. One reviewer (P-YN) extracted the data using a standard data extraction form and a second reviewer (HR-A) validated 10% of the studies for accuracy. The data extracted included: study characteristics (location, time, settings), population’s demographic and clinical characteristics, green space types, green space quality domains, health outcomes and corresponding measures of association. We also recorded the tools used to assess green space quality and health outcomes, effect measures reported, types of statistical analyses conducted and any adjustment for confounding factors. Based on the effect measures and 95% confidence intervals, we recorded the direction of effect for each study, i.e., whether the study presented some evidence of protective associations, some evidence of risk associations, or no significant associations at all.One reviewer (P-YN) appraised the methodological quality of all included studies using the quality assessment tools for the appropriate study types [19] and the second reviewer (HR-A) validated 10% for accuracy. Because these tools do not provide for ecological studies, the existing tool for observational cohort and cross-sectional studies were adapted by adding 3 criteria addressing ecological fallacy, spatial autocorrelation and uncertainty in fitting spatial data [20,21]. Based on the list of applicable criteria, each study was given an adjusted quality score of 0–10 (Supplementary File S2). Disagreement was resolved with consensus via discussion with senior reviewers (XF and TA-B), if required. We used inductive categorisation to develop a set of domains of green space quality based on definitions reported in the included studies and stratified the findings of the studies based on these quality domains. Due to the heterogeneity of exposure, intervention and outcomes, meta-analysis was not conducted.In the initial search, we identified 30,220 records, and 7 additional records were added through manual searching. After removing duplicates, 23,745 studies were included for title/abstract reviews, from which 118 full texts were selected for further screening. Fifty full texts were excluded (Supplementary File S3). The final sample comprised 68 articles from 59 studies (Figure 1). The 59 studies (68 articles) were conducted in 19 countries/territories and were published from January 2009 to April 2021. Most articles were based on studies conducted in the United States (US) (n = 17), Australia (n = 12) and United Kingdom (UK) (n = 10). The mean age of the participants ranged from 4.5 to 76.5 years. A total of 5 studies included only people aged 55 years or older [22,23,24,25,26]; 11 studies included only people under 16 years old [27,28,29,30,31,32,33,34,35,36,37]. Most studies were balanced in gender distribution, with proportions of female participants ranging from 32 to 67%. Four studies exclusively examined female participants [38,39,40,41].Cities and inner-city neighbourhoods were the predominant settings. Seven studies took place in multi-ethnic and/or socioeconomically deprived areas [29,30,31,37,42,43,44]. One study specifically examined the differential impact of green space on children of South Asian descent versus Caucasian children [37]. The characteristics of included studies are summarised in Table 1.Most included studies were cross-sectional (n = 32), followed by ecological studies (n = 16) and cohort studies (n = 15). Before-after (n = 1), quasi-experimental (n = 3) and case-crossover designs (n = 1) were rare (Table 1). The latter were relatively newer approaches published from 2015 onwards (Figure 2). All cohort studies were nested in existing longitudinal studies, usually with an additional cross-sectional survey for green space use and perceptions conducted after the initial survey waves. The follow-up time for longitudinal studies ranges from 2 to 18 years [56]. The quasi-experimental studies [55,57,87] had intervention and control groups selected in a non-random manner from two neighbourhoods with pre-determined green space qualities. The before-after study [22] was conducted among participants who participated in outdoor nature walks. The cross-over study [79] bi-directionally matched case days with the highest symptom severity scores to control days with the lowest scores, hence participants served as their own control. Among cross-sectional surveys, eight studies used convenience sampling by recruiting from park visitors [23,44,51,59,61,73,89,90]. The mean adjusted quality score among 68 articles was 0.49 ± 0.12 (scale 0–1).Most studies (n = 42) used a loose definition of green space to include any natural or open space, encompassing urban green space, private and community gardens, public open spaces, bushland and forest reserves, etc. Eleven studies included playgrounds and sports fields [25,35,36,37,52,53,57,60,67,71,84]. Seven studies included streetscape greenery, which referred to any vegetation cover that gave the street a green appearance [52,53,54,68,69,74,81]. Forty-seven studies used data from a geographic information system (GIS) to identify green spaces or evaluate green space characteristics. One study examined neighbourhood vegetation as viewed from within the house [22]. The most common buffer size for GIS analysis was 0.5 mile (approximately 800 m), generally aligning with a 10-min walk [82]. Detailed definition of green space in each study is outlined in Table 1.A range of health outcomes were reported, which were classified into physical (reported by 34 studies), psychological (n = 25), combined physical/psychological (n = 10), quality of life (n = 5), or developmental outcomes (n = 3). Twenty-seven studies used objective measures of outcomes, mainly assessing physical outcomes (Table 2).The most common tools used for physical outcomes were body mass index (BMI) (n = 9) [29,31,32,38,50,75,76,78,80], together with its associated anthropometric measures such as the percentage of truncal fat [27] and obesity/overweight [32,70]. Six studies investigated cardiovascular conditions such as hypertension, diabetes and coronary heart diseases [47,49,59,70,71,82]. Ten studies investigated respiratory outcomes, such as asthma and other allergic respiratory diseases [28,34,35,56,63,67,69,77,79,82]. The most common tools used for psychological outcomes were the Kessler psychological distress scale (K6-PD or K10-PD) [39,45,60] and the mental health inventory scale (MHI-5) [54,68,81]. All questionnaires used to measure psychological outcomes were self-reported by participants, indicative of the inherent subjectivity of this outcome domain. The strengths and difficulties questionnaire (SDQ) was used in studies assessing developmental outcomes. Lastly, five studies used various versions of the short form survey (SF-8, SF-12, SF-36) [23,52,54,65,81], which assess up to eight domains of health status, including physical functioning, physical role, bodily pain, general health, vitality, social functioning, emotional role, and mental health [91]. Detailed definitions of health outcomes and assessment tools are outlined in Table 3 and Supplementary File S4.Green space qualities were classified into 10 domains. Detailed definitions of green space qualities in each study is outlined in Table 3.There was one before-after study, seven cohort studies, one case-cross over study, seven cross-sectional studies and six ecological studies under this domain. All studies used different land cover or environment classification, commonly via adopting definitions of the data sources, some adapting [39,53,83,85] or developing their own typologies [22,86]. Detailed definitions of environment types were outlined in Table 3.Overall, a higher land-cover diversity in the neighbourhood was protective for chronic morbidities [53] and childhood asthma [56]. Some environment types were more likely to provide health benefits than others. Vegetation patches such as grassland and tree canopy was not associated with reduced sudden unexpected deaths, but formal green spaces such as greenways and forests were [85]. People who spent time outdoor recalled greater mental restoration following visits to coastal locations and rural green space than urban green space [86]. Some environment types (“broadleaf woodland”, “arable and horticulture”, “improved grassland”, “saltwater” and “coastal” environment) were positively associated with prevalence of good health among UK citizens [83]. The observed relationship between land cover types and BMI varied across age and gender. A positive relationship with lower BMI was found with high coverage of impervious surfaces among middle-aged adults and high forest coverage among young adult males. In other age and gender groups, the relationships were non-significant [76]. More rigorous studies, however, did not report significant findings. In a before-after study, the environment type of an outdoor walk did not have significant influence on emotional states of participants [22]. A sibling matched case-control analysis of Scottish mothers and their children (1991–2010) found that infant birth weight was associated with the quantity of natural space around the mother’s home, but was unrelated to specific types of natural space (parks, woods or open waters) [41].Similarly, the type of vegetation within green space potentially modulated its health benefits. There was consistent evidence of forests being a protective factor for obstructive airway diseases [28,34], cardiovascular diseases [49], allostatic overload [59], psychological distress and general health [45,65,74] while grassland and herbaceous vegetation were not. On the other hand, some studies showed superior benefits of shrubs and grass compared to trees in improving mental health [64] or severe allergy [79]. In low-diversity areas, certain vegetation types presented higher risks for asthma or other allergic conditions, typically non-native shrubs [56] or coniferous trees [35]. No difference in benefits between vegetation types was observed in studies of memory and dementia [46,48], depression and anxiety [26,45]. In one study, all vegetation types were shown to be protective against autism, which was potentially driven by their shared function of buffering against traffic noise and air pollution buffering [33].There was one before-after study, one cohort study, ten cross-sectional studies and three ecological studies under this domain. Natural features refer to characteristics of vegetation, animals, water bodies, and the overall naturalness of green space. Trees, flowers and fresh air [73] conferred restorative benefits to park visitors, with differential effects between genders. The higher density of trees among park vegetation was associated with lower rates of cardiovascular conditions [47,70] and a higher quality of life [24,51], but not overall general health [23]. The presence of dense shrubs, which implied lower security and safety, reduced the restorative benefits of parks [89]. Green spaces perceived as being more “natural”, such as protected areas or bushlands, provided greater benefits on mental restoration [86] and physical health [52,83]. A combination of habitat, plant, bird and insect biodiversity exhibited restorative effects, but each biodiversity component alone did not [22,42,44]. Interestingly, neither quantity or diversity of neighbourhood vegetation alone was significant predictor of stress levels, but vegetation diversity could modify the relationship between vegetation quantity and stress levels [62].Certain green space characteristics were potentially associated with health risks. Streetscape with tree species of high allergenicity was associated with an increase in local asthma hospitalisation rates in vulnerable populations [69]. Freshwater quality was identified as an indicator of poor health status [83].There were nine cross-sectional studies, two quasi-experimental studies, one prospective cohort study and two ecological studies under this domain. Infrastructure and amenities refer to the availability of facilities for various purposes (recreation, resting, socialisation, etc.), the quality of paths within and leading to green space, and general maintenance. Park facilities did not reduce rates of depression [40], BMI [31,32,50,75] nor general health status of park users [23,24]. High maintenance was not associated with lower psychological distress [43] or BMI [50]. However, parks that function as recreational or sports venues may provide some cardiovascular and mental health benefits [71,84]. Mixed results were reported on the relationship between walking paths’ conditions and quality of life [25,51]. A natural experiment was conducted in disadvantaged suburbs of Melbourne, Australia, tracking psychological wellbeing of park visitors for 3 years after adding refurbishments (playground equipment, walking paths and shade) to selected parks. When compared to control parks, park refurbishments did not improve emotional states of park visitors [55]. Similarly, in the Netherlands, neighbourhoods that implemented interventions to increase accessibility and useability of green space did not see an improved general health compared to control neighbourhoods [57].There was one prospective cohort study, six cross-sectional studies and four ecological studies and under this domain. Ten studies used spatial analysis to measure green patch size. Most studies found evidence for health benefits of larger green space for a wide range of outcomes: BMI [29,75], cardiovascular mortality [82], chronic morbidities [53], depression [42], general health status [23] and quality of life [30]. In a prospective cohort study in Perth (Australia), where residents were followed up after settling into a new neighbourhood, the increases in numbers of small parks, district parks and regional parks were each positively associated with mental wellbeing, but not the mid-sized local and neighbourhood open spaces [84]. However, some studies reported inconclusive evidence for these health benefits [24,32,78]There were six ecological studies and two cross-sectional studies under this domain. While all studies used spatial analysis to quantify green space patterns, six studies combined health data at the spatial block level [63,67,76,77,80,82] while others conducted regression analyses using individualised data [29,30]. All studies reported positive correlation between indices measuring the shapes and distribution patterns of green patches and a wide range of outcomes, including BMI [29,76], paediatric quality of life [30], respiratory health [63,67,77] and all-cause mortality [82]. The indices include the fragmentation index (higher values indicate more fragmented green space areas), mean area of greens space (higher values indicate averagely larger green space areas), connectedness index (higher values indicate more connection between individual green spaces), aggregation/isolation index (higher values indicate more clustering of individual green spaces), shape irregularity index (higher values means more irregular shape of each green space, as opposed to round/oval shape). When stratified by gender, age and retirement status, differential benefits were observed for female and younger users [76].There were six cross-sectional studies under this domain. The safety of green space was associated with better quality of life [23,25,51], reduced psychological distress [43] but did not have significant effects on BMI [50] of residents. In a mediation analysis, park crimes reduced the benefits of parks on mental health [72].There were three cross-sectional studies and one ecological study under this domain. Park cleanliness, either ranked by park visitors or assessed by trained auditors, was associated with lower rate of depression [42]. Evidence was inconclusive for BMI [50,78] or quality of life [24].There were three cross-sectional studies under this domain. A lower level of “nuisance” (defined as presence of dogs, dog fouling, or young people) was not correlated with better life satisfaction nor physical health among the elderly [25]. Park users did not consider a private environment in the park important in improving their mood states [73]. On the other hand, soundscapes in parks triggered positive feelings and reduced stress [61].There were four nested cohort studies, two cross-sectional studies, and one ecological study under this domain. In these studies, participants were asked to rank their perceived quality or aesthetics of green spaces, without a priori definition of factors to be considered. All studies examining “perceived quality” demonstrate positive association of green space’s perceived quality with health. Women living near good-quality local parks had lower rates of postpartum psychological distress or serious mental illnesses [39]. The effect on postpartum weight gain was less clear, with significant benefits only observed in areas with high vegetation coverage (≥40%) [38]. Parents’ satisfaction with green space was also linked to improved prosocial behaviour of their children [36,37]. Analysis of the Netherlands’ population data found a modest increase in life expectancy among residents living near high-quality green spaces [66]. However, perceived aesthetics of parks was neither a predictor of mood states [73] nor BMI [50].There were one quasi-experimental study, two cohort studies, eight cross-sectional studies and two ecological studies under this domain. These studies use a mix of features from the previous domains to evaluate park quality. Detailed definitions of these composite scores were outlined in Table 3.Five studies determined objective quality based on audits by trained assessors [32,42,54,60,68] while others asked participants to rank quality based on a set of criteria [23,25,52,58,60,68,81,87,88]. Park quality had positive benefits on reducing BMI and truncal fats in young children [27,32]. Evidence on benefits for general health were mixed [8,23,25,42,52,54,81]. Zhang et al. introduced a concept of multi-sensory experience, suggesting that visual, auditory and tactile sensation, provided by different park features, all contributed to the restorative effects of parks [88].Three studies investigated both objectively measured and the perceived quality of green spaces, and compared their effects on health. When comparing two neighbourhoods with different socioeconomic status, the residents’ perceived quality of a green space statistically mediated the relationship between its objective quality and neighbourhood satisfaction, but did not have any direct effect of wellbeing [87]. Only objective quality reduced psychosocial distress (K6-PDS questionnaire) in one study [60] while only perceived quality improved mental wellbeing (MHI-5 questionnaire) in another study [68].Overall, our review demonstrates evidence of health benefits associated with a wide range of green space qualities. Increasing research interest in green space qualities was demonstrated (Figure 2) and this aligns with rising interest in urban greening to counter the health and climate impacts of urbanisation [6]. The COVID-19 pandemic may also have amplified attention on this topic from academics and policymakers, as communities in many countries have flocked to green spaces as a means of coping with lockdowns and socioeconomic disruption [92,93]. After excluding results with a study quality assessment score under 50 (N = 32), evidence showed consistent positive associations with health with the green space qualities we classified as “environment types”, “natural features”, “shape and connectivity”, and “objective quality scores”. Limited evidence was found on the health benefits of improving infrastructure or amenities in green spaces. Research gaps were identified for the following green space qualities: peacefulness, safety and absence of incivilities; needs-specific or culture-appropriate amenities, and soundscape characteristics.The most commonly assessed qualities of green spaces were the environment types of the natural space, as well as vegetation types and other natural characteristics. Our review shows that some environment types were linked to positive health outcomes more than others [41,83,86]. Health benefits were observed when the environment type facilitated age- and gender-appropriate physical activities. For example, middle-aged adults group preferred built facilities with paved paths for exercising whereas young adults prefer forested areas with unobstructed grounds for athletic, adventurous activities such as hiking, trail-running or mountain biking [76,82]. Therefore, preserving diversity in land cover types (e.g., structured versus natural) may be a potential option to enhance health benefits of green spaces, especially in dense urban areas with limited options for expansion. Moreover, green space designs might be optimized for health through tailoring to local community profiles, to bring people together and to enable them to do what they find nourishing. This requires consultation and it is likely that certain qualities may be a source of conflicting views. For example, accommodating for birds in green spaces may be viewed positively for their provision of restorative soundscapes and an enhanced feeling of connectedness with nature, but also negatively due to the timing of their sounds, impacts on property (e.g., droppings) and occasional swooping that may create a lack of felt safety [94].Evidence indicated that some vegetation types may be more beneficial towards particular health outcomes than others. Tree canopy and forests were more consistently associated with better cardiovascular and respiratory health than grassland [47,59,67,79]. A reason may be that trees permit and promote restoration while also providing shade that helps to activate walking and active transportation (particularly in hot climates), whereas grass and shrubs might not convey the same range and levels of benefit [76]. Moreover, because of their foliage, evidence indicates that forests have the capacity to intercept airborne pollutants and buffer against traffic noise, alleviating oxidative stress and reducing risks of atherosclerotic diseases [95]. On the other hand, shrubs may impede visibility and reduce levels of felt safety, while large areas of open grass may reduce walkability (especially if it is walled or fenced-off, as can be the case for private green spaces like golf courses) [74,76,89]. Importantly, this may reflect an interaction between vegetation type and other contextual factors, such as levels of crime, nearby land-use and transport infrastructure. Further research that examines potential contingencies of association between vegetation types and health outcomes within the context of other land-uses is warranted.Interestingly, many studies in the facilities/amenities domain show no statistically significant associations with physical or mental health, despite evidence that some of these qualities are associated with physical activity [90]. This might be because different types of facilities may result in different forms of behaviour, some of which may instead promote sedentary forms of leisure (e.g., seating) or detract from the perceived ‘naturalness’ appealed by certain park users (e.g., some sports facilities that use synthetic materials) [96]. Moreover, some studies may log the availability facilities but not their condition and usability. For instance, access to areas of parks and particular buildings may be difficult for people with functional limitations, while there may also be cultural or social factors that influence whether a particular facility is considered accessible [97].Some qualities have a small evidence base, such as safety, tranquility or absence of incivilities. Most of this evidence focused on psychological wellbeing or quality of life using Likert-type rankings or the number of unwell days. Future studies on these aspects will benefit from using robust, validated questionnaires featured in other quality domains, such as MHI-5, PRS or PANAS [22,55,68,89]. Moreover, perceived safety of public spaces can be influenced by neighbourhood characteristics and social vulnerabilities, which need to be accounted for in future studies.Some quality domains were not featured among the included studies. Availability of needs-specific amenities, such as for people living with particular disabilities, may encourage more inclusive park usage and increase the potential to reduce health inequity [3,98]. Tailoring park amenities and features to the local communities, such as instructions in multiple languages, accommodation (and celebration) of cultural traditions and rituals, etc., may be particularly important in multi-cultural neighborhoods [99,100]. One study examined the feelings evoked by soundscape [61], but the constituents of soundscape that provide therapeutic effects, such as sounds of nature, human activities or traffic noise, were not elucidated. Types of bird songs were previously studied, but as sound clips rather than actual exposure inside parks [89].Physical health is the most commonly assessed set of health outcomes. Most studies showed evidence of potential benefits for anthropometric measures (BMI and obesity) and respiratory health (allergic diseases). Understandably, there are established frameworks explaining how green spaces reduce obesity via promoting physical activities [96,101] and protects against respiratory diseases via regulating temperature and air pollution [77]. Only 7 out of 34 studies on physical health examined associations with cardiovascular diseases.Based on existing evidence, higher quality green space may reduce cardiovascular mortality and incidence of cardiometabolic diseases [47,70,71]. However, evidence for associations with specific cardiovascular diseases was small. Consistent evidence from this review indicated a range of probable mental health benefits linked with various green space qualities. This aligns with existing conceptual frameworks, which suggest green spaces can confer mental health benefits via reducing stressor exposures and replenishing mental resources for coping [3].Although the evidence base is substantial for physical and psychological health outcomes, there is granularity in the quality of outcome measurement tools. For physical health, a number of studies relied on general health questionnaires such as SF-12 and SF-36. These have a low administrative burden and good internal validity [91], but responses may differ among age, education or ethnicity subgroups [102], which may explain conflicting findings among these studies. For mental health, some studies used self-ranked Likert-type questions, which lacked reliability and consistency compared to validated questionnaires like the MHI-5, K6-PDS or CED-S. A potential approach for future studies is to use quantifiable biological measure to validate subjective questionnaires, such as hair cortisol levels as a proxy for stress [62].Few studies investigated child development. This could be the focus of future studies, as evidence suggests possible health benefits linked to reduced maternal stress during pregnancy [33] and opportunities for play and socialisation during time spent in green spaces [36,37].Certain outcomes were not featured in the included studies. Vegetation types and structure influence their ability to regulate pollution and local climate, and thus will have differential effects on heat-related health risks [103]. Postpartum distress was examined [39], but the effects on antenatal depression or neonatal outcomes were not investigated. This is an important topic, as the greenness of the environment was associated with reduced risks of low birth weight and preterm delivery [104].Overall, the level of evidence certainty for health benefits of green space quality remains low.This is due to two important reasons. Firstly, there was a high degree of heterogeneity in study designs, green space and green space quality definitions, and outcome measurements. Some studies use factor analysis to derive the qualities, which make it difficult to find out the definitions behind the derived terms, especially when the survey questionnaires were not included [40,52]. Many studies ask participants to rank certain qualities or report health outcomes on a Likert scale-type questions, without defining the quality being surveyed for the participants. These potentially introduce bias in response and are a major limitation among studies in this topic. Even with GIS methods, which are deemed more reliable and reproducible in quantifying green space exposure, variations in proximity radius and buffer zones make it difficult to compare results across studies.Secondly, none of the included studies were randomised trials, which resulted in a lower overall quality of evidence.Only 6/10 domains featured evidence from longitudinal cohort studies or interventional studies (before-after and quasi-experimental studies), namely the domains of environment types, natural features, infrastructures and amenities, size, perceived quality, and combination of features. Within each domain, cross-sectional and ecological studies often accounted for more than half of the evidence base. The prevalence of observational studies is characteristic of environmental health research, which faces intrinsic logistical and ethical challenges in designing rigorously controlled trials [105]. Nonetheless, observational studies have their limitations [106]. Cross-sectional surveys do not permit inference of causation. In our review, many cross-sectional surveys used convenience sampling, which could introduce selection bias due to seasonal weather, site of surveys or time of day. Longitudinal studies can factor in temporal relationship between green space exposure and health outcomes. They also enabled adjustment for factors that can influence health outcomes, such as demographic characteristics, measures of poverty and deprivation, and socioeconomic status (income, education and employment) (Table 1). However, many cohort studies in our review were nested in longitudinal health surveys that did not routinely collect data on green space quality, and only achieved so via a cross-sectional survey or geospatial analysis [36,38,39], again making it impossible to establish temporal causation. Although ecological studies echo the principles of environmental health policies, their generalizability is limited. By assuming that green space exposure applies uniformly to all individuals within a census tract or administrative area, these studies do not control for individual health and preference, and thus may lead to incorrect inferences (“ecological fallacy”). The use of multiple databases in GIS analysis, featured in many of our studies, also raises the possibility of spatial autocorrelation and mismatched data sources, etc. [34,83]. Before-after studies and quasi-experiments are pragmatic designs that support causal inference by establishing a clear temporal relationship between exposure and outcomes and controlling for confounding factors. They provide real world effectiveness of complex interventions, and are thus compatible with population policies [107].It is important to note that, although longitudinal cohort studies and interventional studies were less prevalent, they have methodological strengths that cross-sectional and ecological studies do not. In our review, limiting analysis to these studies did not change the overall conclusion across all quality domains.Innovative trial designs have been featured in this review, namely quasi-experimental studies using controlled parks or neighbourhoods [55,57]. In addition, controlled intervention design had been used in forest therapy trials, which allowed for robust pre-post measurements of cardiovascular outcomes such as blood pressure, heart rate and oxygen saturation [108]. However, high logistical demands often limited the duration of these trials and precluded studies of long-term (child development) or high-risk outcomes (childbirth, cardiovascular events). Studies nested in cohort follow-up studies [28,36,37,41,49,60,84] are a promising approach by leveraging on well-designed longitudinal studies with annual follow ups, comprehensive baseline data collection, and large sample sizes for robust statistical power. Where randomisation is not possible, study data could be analysed using interrupted times-series analysis, which adjusts for some effects of context and individual health variations over time [69].Satellite imagery and GIS should still be part of the essential toolbox for green space quality studies, as long as GIS data is linked to patient-level data instead of being aggregated at ecological unit levels. GIS has proven useful in combining cartographical datasets, identifying and classifying land cover types. Recent advances in geospatial big data also introduced new approaches to assessing green space exposure, such as eye-level exposure (street view imagery) as opposed to overhead exposure (satellite imagery) [109]. In addition, GIS technology has enabled new indices for quantifying green space size, shape and connectivity [30,82]. By virtue of defined formulae, these indices were reproducible and reliable, and could be used in various statistical analyses.Our findings showed that perceived green space quality, even without any judging criteria, can predict health benefits [36,37,39,66]. This is an important consideration, given that spatial environmental indicators (size, greenness, aesthetics) do not always corresponded with user perceptions [110]. Therefore, it is advisable for future studies to measure both perceived and objective quality when assessing health benefits. This approach has the dual benefits of ensuring internal validity of the subjective quality measurement, while accounting for any mediating effect of user perceptions on the objective quality [60,68].Several studies used a composite quality score that aggregated across several domains (e.g., Public Open Space Tool). Although a composite score approach can reflect the complexity of green space quality, coverage can be restricted to attributes related to facilities, safety and cleanliness, which are shown in our review to have little association with health so far. RECITAL, the latest quality assessment index developed to address this gap, incorporates other quality domains such as suitability for activities, land cover types and biodiversity [111], which generally aligns with our classification. This index can be stratified into single-item or sub-section scores, allowing researchers to investigate specific aspects of quality, which is a shortcoming commonly associated with aggregated scales. Comprehensive indices such as this should be explored in future studies. Last but not least, there is a need for a new index that aggregates qualities across networks of multiple green spaces of various shapes and attributes. This may be particularly salient within higher density contexts, where multiple smaller green spaces exist with each containing a small number of qualities, but larger ones that may incorporate many more qualities do not.The strength of our review is its breadth of coverage, as we formulated our search strategy intentionally to capture across a range of health outcomes, potential qualities and green space types. Our review is the first to capture the diverse evidence conducted in this area and map them into domains of quality. Nonetheless, our review was not without limitations. As the concept of green space quality was not well-defined, we took a holistic approach but our review could still potentially miss out relevant studies that did not use conventional descriptors of quality. Our review only included studies written in English, and in view of more emerging research on park designs from China in recent years [112], publication bias due to exclusion of non-English articles was possible. Although our review was structured based on established protocols, the screening process was subjected to some degree of subjectivity due to a lack of standardized definitions in this topic.Research on green space quality and health has increased in volume, especially since 2016. A high degree of heterogeneity was observed in study design, and the definitions of quality and outcomes measured. Environment types, vegetation types, and the size and connectivity of green spaces, were associated with physical and mental health outcomes, with differences by age and gender. The associations indicative of health benefits were more consistent in populations with more tree canopy, but not more grassland. Qualities such as safety, cleanliness and aesthetics tended to be investigated with weaker study designs. Both objective and subjective quality demonstrated positive effects on health outcomes. There is a need for more experimental studies or well-designed prospective studies that incorporate longitudinal measures of green space qualities and outcome-appropriate confounders. Green space indices should account for form, pattern, networks, and both objective and perceived qualities.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111028/s1, Supplementary File S1: Search strategy; Supplementary File S2: Quality assessment of included studies; Supplementary File S3: List of excluded studies from full-text review. Supplementary file S4: Summary of findings (expanded).Conceptualization, T.A.-B. and X.F.; methodology: P.-Y.N.; validation: T.A.-B. and X.F.; formal analysis, P.-Y.N. and H.R.-A.; data curation, P.-Y.N.; writing—original draft preparation, P.-Y.N.; writing—review and editing, P.-Y.N., X.F., and T.A.-B.; supervision, X.F.; funding acquisition, X.F. and T.A.-B. All authors have read and agreed to the published version of the manuscript.This study was supported by a National Health and Medical Research Council Boosting Dementia Research Leader Fellowship 1140317 (Astell-Burt) and the National Health and Medical Research Council Career Development Fellowship 1148792 (Feng). Astell-Burt and Feng were also jointly supported by grant 1101065 from the National Health and Medical Research Council project and grant GC15005 from the Green Cities Fund—Hort Innovation Limited, with co-investment from the University of Wollongong Faculty of Social Sciences, the University of Wollongong Global Challenges initiative, and the Australian Government. The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.Ethical review and approval were waived for this study, as the review does not use personal data.Patient consent was waived for this study, as the review does not use personal data.The data presented in this study are available in the main article and Supplementary Files.The authors declare no conflict of interest.PRISMA flow diagram of study selection.Published studies over the years, by study design.Study characteristics.* Abbreviations: CCS: case-crossover study; CSS: cross-sectional study; CS-Retro: retrospective cohort study; CS-Pros: prospective cohort study; QES: quasi-experimental study; BAS: before-after study; Eco: ecological study; ** Default unit is person unless specified otherwise. Abbreviations: DA: dissemination area; LSOA: lower layer super output area; MSA: metropolitan statistical area. ≠ analysis was stratified by sex.Mapping of measures used for assessment of green space qualities and outcomes.* Expressed as a percentage of all studies under respective green space quality domain.Summary of findings.Notes: Within each quality domain, studies were arranged by study design, and then by sample size. A full version of this table is available as Supplementary File S4. * Abbreviation: DA: dissemination areas; LSOA: lower layer Super output areas; MSA: metropolitan statistical areas ** AMI: acute myocardial infarction; ART: attention restoration theory; BMI: body mass index; CES-D: Center for Epidemiologic Studies-Depression; CIDR: comparative illness and disability ratio; CVD: cardiovascular diseases; EUROHIS-QOL-8: EUROHIS 8-item quality of life questionnaire; GDS: geriatric depression scale; GHQ-12: 12-item general health questionnaire; GIS: Geographic Information System; K10-PDS: Kessler ten-item psychological distress scale; K6-PDS: Kessler six-item psychological distress scale; MHI-5: 5-item mental health inventory; PANAS: positive and negative affect schedule; PedsQL: paediatric quality of life inventory; POST: Public Open Space Tool; POSDAT: Public Open Space Desktop Auditing Tool; PRS: perceived restorativeness scale; PSS: perceived stress scale; SDI: Shannon’s diversity index; SDQ: strengths and difficulties questionnaire; SF-8: eight-item short form survey; SF-12: 12-item short form survey; SF-12v2: short form 12 item (version 2); SF-36: 36-item short form survey; SF-36v2: short form 36 item (version 2); SWLS: satisfaction with life scale; WEMWBS: Warwick Edinburgh mental well-being scale; WHOQOL-BREF: World Health Organization quality-of-life scale. ≠ (+) Some evidence of protective associations; (–) some evidence of risk associations; (o) no significant associations observed.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11029.txt

@@ -0,0 +1 @@
+Landfill leachate is a complex mixture of organic and inorganic molecules, as well as environmental pollutants that can cause harm to ecosystems and living beings. The micronucleus test in peripheral blood erythrocytes was used to evaluate the genotoxic and cytotoxic effects of exposure to a landfill leachate from an outdoor solid waste storage system on Wistar strain rats at different developmental stages, pre-adolescents and young adults, and the heavy metal content of the leachate was determined by atomic absorption spectrometry. Contents of arsenic, cadmium, chromium, mercury, and lead in the landfill leachate were outside the allowable international standards, and the exposure to the landfill leachate caused genotoxic and cytotoxic effects on Wistar rats, where the pre-adolescent animals were more susceptible to the toxics contained in the landfill leachate than young adults. Heavy metals contained in landfill leachate, individually or synergically with other molecules can be responsible for clastogenic and cytotoxic effects that can be harmful to humans and ecosystems.Urban solid waste storage systems (USWSSs), such as the open-air landfill, generate leachates that pollute the air, soil, and water, which threatens the health of ecosystems and living beings [1]. These landfill leachates are formed as a result of solutions percolating through trash and undergoing a biochemical transformation process. In addition, this mixture contains considerable concentrations of heavy metals and substances classified as pollutants of worldwide concern [2]. Toxicogenetic effects and consequences of landfill leachate exposure on microbes and plants as well as birds and fish are well-documented [1,3]. Nevertheless, the study of genotoxic and cytotoxic effects exclusively in terrestrial species, such as mammals, is limited [1,4,5,6,7,8,9]. Therefore, in this study, the genotoxic and cytotoxic effects of exposure to a landfill leachate from an outdoor solid waste storage system on Wistar strain rats during their pre-adolescent and young adulthood developmental stages were evaluated.An instantaneous sampling of the leachate from an outdoor USWSS located in the city of Guadalajara, Jalisco, Mexico was carried out. The sample was transported to the laboratory in sterile 1 L glass containers. The leachate was filtered with Whatman® (Maidstone, UK) 0.2 µM to remove any suspended solid impurity.Leachate was centrifuged at 3000× g for 10 min; the supernatant was considered as the 100% stock sample and stored at 4 °C until use. For the biological test, two concentrations of the leachate (LCH5%: 5% v/v, and LCH25%: 25% v/v) were prepared in distilled water.The concentrations of heavy metals, arsenic (As), cadmium (Cd), chromium (Cr), mercury (Hg), and lead (Pb), were determined using the atomic absorption spectroscopy technique. Measurements were carried out at the Analytical and Metrological Services Unit of the Center for Research and Assistance in Technology and Design of the State of Jalisco.A total of 60 male Wistar rats were provided by the Centro de Investigación Biomédica de Occidente (CIBO), Instituto Mexicano del Seguro Social (IMSS). The animals were housed in translucent acrylic boxes (44 × 33 × 25 cm) placed in a living room at room temperature, with a 12 × 12 h light/dark cycle (light was turned on at 7:00 am) and ad libitum access to water and food. Animals were handled following the animal care guidelines by Federal Government of Mexico (NOM-062-ZOO-1999) and the study was approved by the Committee for Ethics and Health Research of CIBO-IMSS with registration number R-2019-1305-6.The experimental design involved three developmental stages in rats. For each stage of the experiment, four groups of five animals were used in each group. At the beginning of the experiments, pre-adolescent rats were 30 days old, whereas young adults were 60 and 90 days old for the second and third stages [10]. Five percent landfill leachate was administered to the first group (LCH5%) and 25% to the second group (LCH25%), while 20 mg/kg cyclophosphamide was administered to the third group (positive control), and distilled water was given to the control group (negative control). In total, 500 uL of each treatment were administered through the intragastric route with a stainless-steel cannula for 30 consecutive days. After the treatment was completed, blood was drawn from the tail vein of each animal (each group at that time was 60, 90, and 120 days old). An intraperitoneal administration of cyclophosphamide was performed 48 h prior to collecting the blood sample for the positive control (blood collection was performed at 60, 90, and 120 days old as well).The genotoxic and cytotoxic effect was evaluated by quantification on the erythrocyte micronucleus formation [11,12]. Peripheral blood drops were placed on prestained slides with acridine orange and the micronucleated cells were scored by a fluorescence microscope (model CX-40, Olympus, Hamburg, Germany) under a fluorescence illumination system (DMB-2 Olympus blue filter).In total, 2000 polychromatic erythrocytes (PEs) were analyzed and the frequencies of micronucleated polychromatic erythrocytes (MNPEs) were scored on three slides per animal; the genotoxic effect was expressed as MNPE/PE. In order to assess the cytotoxic effect, the number of PE per total erythrocytes (TEs) as well on three slides from each animal was counted; cytotoxic effect was expressed as PE/TE.All data were expressed as mean ± standard error on triplicate. Genotoxicity and cytotoxicity data were analyzed by one-way ANOVA followed by Dunnett’s post hoc test to find groups with significant differences using GraphPad Prism 8.0.1 software (GraphPad Software, San Diego, CA, USA). p < 0.05 was considered significant.Leachate from the sanitary landfill from the open-air storage of solid waste is a mixture of organic compounds, inorganic macromolecules, and heavy metals. The mercury level found in the landfill leachate analyzed in this study was five times higher than those allowed by the Mexican standard and 50 times above international standards (Table 1). Other heavy metals, such as arsenic, cadmium, chromium, and lead, were found to be within the Mexican standard. Nevertheless, arsenic and cadmium were found to be 10 and 50 times above the permissible limits of the international standard, as well as for chromium and lead, which were found to be 10 and >65 five times above the allowable limits (Table 1).As a primary source of toxic substances emitted by landfills, leachates could be the main source. Heavy metals as pollutants can be found in low concentrations, but ill effects have been reported for organisms exposed to these trace quantities [1,4,6,7,14]. According to the United States Environmental Protection Agency, chromium causes allergic dermatitis and is carcinogenic, while arsenic damages skin, causes circulatory problems [15], and has neurotoxic effects, affecting speech, visual perception, and cognitive functioning [16].Cadmium damages the respiratory system and causes neurodegenerative diseases, including Parkinson’s and Alzheimer’s diseases [17]. Cd accumulated in the renal cortex leads to renal damage [18] and contributes to bone demineralization, diabetes, and hypertension [19,20], and is carcinogenic as well.The adverse effects of mercury exposure during gestation include neural tube defects, cleft palate, eye abnormalities, as well as neurobehavioral changes and classic birth defects, such as mental retardation, cerebral palsy, deafness, blindness, and dysarthria [21]. Moreover, Hg impairs sensory, motor, and cognitive functions; reduces kidney function [22]; and can increase the β-amyloid proteins that are associated with Parkinson’s and Alzheimer’s diseases [23,24].The presence of lead affects intellectual functioning, memory, and visual perception abilities [25,26]; causes hemolysis, anemia, and kidney damage; reduces sperm production; and reduces vigor and libido [15]. The toxic effects of exposure to Pb among females have also been associated with miscarriages and stillbirths [27].The genotoxicity of landfill leachate is generally measured by exposing the substance to living organisms or cells in vitro and phenotypic damage is evaluated [2]. However, sometimes, phenotypic changes or signs of toxicity cannot be easily detected, thus the micronucleus technique is an excellent biomarker to evaluate genotoxicity, recommended internationally [11] as part of safety assessment, as the acridine orange staining allows the differentiation of micronucleated polychromatic erythrocytes from polychromatic erythrocytes, indicating chromosome damage was induced.After exposure to landfill leachate, pre-adolescents and 90- and 120-day-old young adults Wistar rats showed a significant increase of micronucleated erythrocytes in peripheral blood (Figure 1). The MNPE incidence induced by landfill leachate at 5% and 25% in 60-day-old pre-adolescent rats was 2.92- and 5.50-fold higher (p < 0.01) than the control (Figure 1a). Interestingly, landfill leachate at 5% did not show significant differences when compared to the negative control in 90- and 120-day-old young adults Wistar rats; however, landfill leachate at 25% showed significant differences (p < 0.01) by 3.09- and 2.5-fold compared to the negative control at this developmental stage (Figure 1b,c).A cytotoxic effect is associated with a smaller proportion of polychromatic erythrocytes per total erythrocytes. A statistically significant cytotoxic effect of landfill leachate was observed in Wistar rats of different developmental stages (Figure 2). A cytotoxic effect was observed in pre-adolescent rats (p < 0.001; Figure 2a), but none of these effects were observed in young adult rats of 90 or 120 days of age using landfill leachate at 5%, while 25% landfill leachate had a cytotoxic effect on both pre-adolescent and young adult rats (Figure 2).The significant increase in the formation of MNPE among PE in the peripheral blood of the animals at different developmental stages exposed to the landfill leachate demonstrated and confirmed the clastogenic effects of the environmental contaminants present in the open-air storage of solid waste (Figure 1). Furthermore, the significant decrease of the ratio of PE to total erythrocytes in landfill leachate-treated rats (Figure 2) suggests a toxicological effect on normal bone marrow cell proliferation induced by the constituents of the leachate [6].This study provides important and useful information on adverse effects of exposure to landfill leachates on mammalian rodents at developmental stages of pre-adolescent and young adulthood, which has not had been studied previously. Throughout history, it has been debated whether environmental pollutants influence mammalian reproduction in the later stages more than early stages. Because certain stages of development are more critical, it is very difficult to evaluate the possible effects of an exposure to a metal in terms of the chemical form, the dose, or the administration route [28,29]. Notwithstanding, the results of this study demonstrate that the LCH25% group that was administered a 5 times higher dose of landfill leachate than the LCH5% group, which therefore contained a quantity of metals five times greater, presented a higher number of micronucleated polychromatic erythrocytes and a lower number of polychromatic erythrocytes, which indicates greater genotoxicity and cytotoxicity (Figure 1 and Figure 2).The fact that immature animals, such as is the case in this study for pre-adolescent Wistar rats, have greater absorption processes and more vulnerable organs makes them more susceptible to the effects of toxic metals and other pollutants. Senescent or aging animals are also more susceptible to genotoxic and neurotoxic contaminants, which can cause negative effects on their organs and systems.Further studies are needed to determine the phenotypic damage and potential effects on the immune and endocrine systems.The approach used to test genotoxicity and cytotoxicity in this investigation demonstrates a valuable method for determining the toxic effects of exposure to landfill leachates. Heavy metals as pollutants, outside of the allowed international limits contained in landfill leachate, could individually or synergically with other molecules be responsible for the genotoxic and cytotoxic effects in pre-adolescent and young adult Wistar rats exposed to the landfill leachate from an outdoor solid waste storage system, with animals in an earlier stage of development being more susceptible to toxic molecules.Conceptualization, E.P.-C. and J.M.F.-F.; methodology, O.R.T.-G., I.M.S.-H., E.P.-C. and J.M.F.-F.; experiment design, M.E.F.-S., V.C.-H., E.P.-C. and J.M.F.-F., formal analysis, O.R.T.-G., I.M.S.-H. and C.S.-F.; investigation, O.R.T.-G., C.S.-F., E.P.-C. and J.M.F.-F.; data curation, O.R.T.-G., I.M.S.-H., M.E.F.-S. and V.C.-H.; writing—original draft preparation, O.R.T.-G.; writing—review and editing, L.H.-G., E.P.-C. and J.M.F.-F.; visualization, V.C.-H. and L.H.-G.; supervision, E.P.-C. and J.M.F.-F. All authors have read and agreed to the published version of the manuscript.This study was funded by the National Council of Science and Technology (CONACyT) Mexico for postgraduate studies in Environmental Health Sciences (Reg. 926677).Animals were handled following the animal care guidelines by Federal Government of Mexico (NOM-062-ZOO-1999) and the study was approved by the Committee for Ethics and Health Research of CIBO-IMSS with registration number R-2019-1305-6.Not applicable.The data underlying this article will be shared on reasonable request from the corresponding author.The authors thank everyone involved in this study.The authors declare no conflict of interest.Genotoxicity effect by 30 exposure consecutive days of landfill leachate at 5% and 25% to (a) 60-day old pre-adolescent Wistar rats, (b) 90-day-old and (c) 120-day-old young adult rats at the end of the exposure. Micronucleated polychromatic erythrocytes per 2000 polychromatic erythrocytes (MNPE/2000 PE) in peripheral blood are represented as mean ± SE. The superscripts are significantly ((b) p < 0.01; (c) p < 0.001) different from the negative control group (distilled water). Positive control: cyclophosphamide (20 mg/kg).Cytotoxic effect as a result of 30 consecutive days of landfill leachate exposure at 5% and 20% to (a) 60-day-old pre-adolescent Wistar rats, (b) 90-day-old, and (c) 120-day-old young adult rats when the experiment concluded. Polychromatic erythrocytes per 1000 total erythrocytes (PE/1000 TE) in peripheral blood are represented as mean ± SE. The superscripts are significantly ((a) p < 0.05; (b) p < 0.01; (c) p < 0.001) different from the negative control. Positive control: cyclophosphamide (20 mg/kg).Heavy metal content in landfill leachate and maximum permissible limits (mg/L).* Values were expressed as mean ± SE of triplicate analyses. ** NOM: Mexican Official Standards, NOM-002-SEMARNAT-1996 [13]. *** EPA: United State Environmental Protection Agency (https://www.epa.gov/ground-water-and-drinking-water/national-primary-drinking-water-regulations#Inorganic (accessed on 23 August 2021)).Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11030.txt

@@ -0,0 +1,9 @@
+The aim of this study was to redesign an emergency department [ED] data management system to improve the availability of, and access to, data to facilitate patient flow. A pre-/post-intervention design was employed using Lean Six Sigma methodology with a focus on the voice of the customer, Gemba, and 5S to identify areas for improvement in ED data management processes and to inform solutions for improved ED patient flow processes. A multidisciplinary ED team includes medical consultants and registrars, nurses, patient service staff, radiology staff, as well as information technology and hospital management staff. Lean Six Sigma [LSS] diagnostic tools identified areas for improvement in the current process for data availability and access. A set of improvements were implemented to redesign the pathway for data collection in the ED to improve data availability and access. We achieved a reduction in the time taken to access ED patient flow data from a mean of 9 min per patient pre-intervention to immediate post-intervention. This enabled faster decision-making by the ED team related to patient assessment and treatment and informed improvements in patient flow. Optimizing patient flow through a hospital’s ED is a complex task involving collaboration and participation from multiple disciplines. Through the use of LSS methodology, we improved the availability of, and fast access to, accurate, current information regarding ED patient flow. This allows ED and hospital management teams to identify and rapidly respond to actions impacting patient flow.The National Emergency Medicine Programme [1] advises all emergency departments [ED] to implement a six-hour standard for ED attendances so that 95% of patients are admitted or discharged within six hours of attending an ED. This target is to ensure ED patients receive timely assessment and intervention as required by their clinical presentation. This indicator aims to reduce the delays without compromising the quality of care. Inpatient boarder is a phrase used to describe a patient who has been assessed in the ED as requiring hospital admission and is waiting in the ED until an inpatient bed becomes available [1]. Prolonged delays for inpatient boarders in EDs have been shown to be associated with poorer outcomes [1]. Achieving the six-hour National Emergency Medicine Programme target requires each step of the process from the point of patient arrival and registration at the ED to their eventual discharge from the department, or their admission to an inpatient bed, to work seamlessly to ensure the highest quality care is delivered as efficiently as possible [1]. By definition, ED care is unscheduled and of varying acuity [2]. Delivery of high-quality care in an efficient manner requires clinical expertise, adequate space, and appropriate equipment, as well as timely access to meaningful data on care delivery. Clinical data supports and assists an ED team in their clinical decision making, but also, importantly, in tracking the progress of their patients’ care journey and maintaining a smooth workflow. It facilitates a better understanding of the flow of patients, bottlenecks, and patient–staff interactions [3]. The Health Information and Quality Authority states on page 4 of their 2017 “Information management standards for national health and social care data collection” document that “health information has an important role to play in healthcare planning decisions” [4]. The National Emergency Medicine Programme requires that ED information systems should be developed to facilitate measurement for ED processing times and support the delivery of high-quality care [1]. The individual time periods between each specific point of the patient’s journey are termed “turnaround time” [TAT].The study site is an ED in a private hospital in South Dublin, Ireland. Private hospital indicates that the organization operates independently of state health services, and receives no state funding. Care is funded through private health insurance. Public health services in Ireland are provided in Health Service Executive [HSE] hospitals and public voluntary hospitals and in practice, there is very little difference between these two types of hospital [5]. Of note, many of these hospitals also provide private health care but they must clearly distinguish between public and private beds.All Irish residents are eligible for public healthcare; however, there are noted variations in coverage, access, and cost, depending on a person’s income, geographic location, and the length of time it takes to receive care. Currently, 45% of the population has voluntary health insurance [6]. In Ireland, a ten-year plan to achieve universal healthcare “Sláintecare” was published in 2017, with the action plan launched in 2019, and this is currently ongoing.In the study site, the patient journey through the ED commences with registering for the service with patient service staff. This triggers a process by which the patient interacts with multiple healthcare staff across all grades and disciplines and the process includes:◦completion of a nurse-led triage assessment;◦completion of physician and nursing assessments;◦completion of diagnostics such as radiology and pathology that inform clinical decisions regarding admission as an inpatient or discharge to outpatient care.completion of a nurse-led triage assessment;completion of physician and nursing assessments;completion of diagnostics such as radiology and pathology that inform clinical decisions regarding admission as an inpatient or discharge to outpatient care.In the study site, the patient may have direct or indirect contact with up to fifteen different staff members across up to five different hospital departments. Target turnaround times for triage and completion of assessment as demonstrated in Table 1 are based on the Manchester Triage System [7], a system of clinical risk management employed in EDs worldwide to manage patient flow safely when clinical need exceeds capacity. It sets the target times by which patients assessed in different categories of severity should be seen [7].Other aspects of the ED pathway are context-specific and subject to local arrangements; for example, the TAT from referral to specific diagnostics to receipt of results, and the TAT from the decision to admit to actual patient admissions to an inpatient bed. Therefore, optimizing patient flow through ED is multifactorial [8] with many points of data entry, access, and collection by staff. Optimizing patient flow is dependent on the ability of multidisciplinary teams to meet the needs of the acutely unwell patient and on the capacity within a hospital to triage, assess, admit, and discharge patients. Importantly, given the multiple steps in the process, it is critical to success to meet the ED staff requirement for easily accessible, relevant data to monitor a patient’s journey, comply with targets, and identify areas for improvement in patient flow [9]. This study discusses a process redesign to improve the availability of, and ED staff access to, relevant data to facilitate patient flow.The setting for this study was a consultant-led ED in a private hospital in Dublin with a capacity to see and treat 55 patients per day. Private hospital indicates the organization operates independently of state health services and receives no state funding. Care is funded through private health insurance. Patients may self-present or be referred to the ED by their general practitioner. The ED does not accept ambulance admissions and it does not offer 24-h cover, operating between 10:00 and 19:00. This is an important difference in the arrangements of EDs in public hospitals in Ireland. Patients must be admitted or discharged each evening; there is no option for patients to remain in the ED overnight. An initial verbal inquiry to the ED staff as to issues that affected patient flow indicated that access to data was considered to be the primary factor and they found the current process to be slow and cumbersome. This access to data to inform patient flow was, and is, important as ED patient flow impacts the entire hospital system. Longer lengths of stay [LOS] in the ED affects many stakeholders as follows: the patient, in terms of potential delays in care and treatment; admitting teams, as admission assessments are required later in the evening or overnight when staffing is reduced, and ED staff, as overtime is required if the patient’s care in the ED has not been completed by the end of the scheduled shift [10]. There were 180 h of ED nursing overtime in January and February 2020, leading to concerns within the ED and wider management team about staff wellbeing, which included occupation-related fatigue [11]. Finally, any delay in ED patient flow and LOS has a corresponding effect on bed requirements for scheduled surgery within the operating room [OR]. Patient feedback on their ED experience was also considered. However, the main suggestions for change from patients were related to invoicing for their ED visits and did not reference their actual care needs. In December 2019, the hospital executive management team [EMT] identified improving patient flow as an area for targeted improvement. As the first step in this improvement process, and reflective of staff feedback, consistent accurate data regarding ED patient flow would be required. In keeping with the hospital’s strategic approach to process improvement, Lean Six Sigma [LSS] was the improvement methodology of choice. LSS methodology has been effective in reducing LOS in a hospital ED [9,10]; Futera and colleagues [12] highlighted access to information technology and data-driven improvement as key facilitators for change. This process improvement, therefore, aimed to improve the availability of, and ease of access to, ED patient flow data management through the application of LSS methodology.LSS has been used in healthcare since the early 2000s to improve efficiency and achieve quality and operational excellence [13]. Since healthcare providers worldwide, whether publicly or privately funded, are faced with similar challenges of caring for an aging population with a limited pool of financial and personnel resources, the need to seek efficiencies while continuing to provide quality services has become more and more acute [14]. LSS has been implemented in many healthcare organizations with improvements achieved across many clinical and administrative pathways and processes, including medication management, specific patient conditions such as stroke [15], OR organization and efficiency [16], and appointment and clinic management [17]. Lean or LSS has been utilized in EDs to improve waiting times and patient flow [18].The hospital adopted LSS as a methodology for process improvement in 2017. By 2020 the LSS programme in the organization had matured to a team of 13 advanced improvement practitioners from disciplines including nursing, physiotherapy, speech and language therapy, administration, and patient services who had completed a post-graduate certificate or diploma training in LSS Process Improvement in Healthcare. These practitioners had previously delivered process improvement projects across a wide variety of topics including streamlining of booking of elective surgeries, reducing LOS in elective orthopedic surgery, as well as procurement and theatre stock management. The University training programme carried out with the study site’s academic partner University College Dublin [UCD], aims to give staff an appreciation of systems and to avoid using LSS as a decontextualized toolkit [19].A pre-post study design measures a variable of interest before and after an intervention with the same location, setting, and participants [20]. For this study, a pre-/post-intervention design was employed using Lean Six Sigma methodology to measure variables related to the availability of and access to ED data within the ED setting and with specified participants [ED doctor, nurse, and patient service team members]. The design enabled us to measure the impact of a LSS redesign of existing processes for data access and retrieval within the ED. The LSS Define, Measure, Analyze, Improve, Control [DMAIC] framework was utilized to structure the improvement. The LSS tools used throughout the improvement process are set out in Table 2.A LSS project team, convened by a graduate of the University education and training programme, was established to carry out this process improvement, as outlined in Table 3. The support of the EMT facilitated engagement from the emergency department and supporting teams. Having a data analyst/information technology specialist on the team was crucial to providing detailed statistics behind ED processes, which was a challenge to the completion of the project as it ran during the organization’s response to the first wave of COVID-19 in Ireland. IT resources were reassigned to implement urgent technical solutions as part of the organization’s response to the COVID-19 pandemic.DMAIC provided a model for the structured approach set out in a project charter [21]. A team project charter was agreed upon with the support of the CEO and deputy CEO with a project goal of having timely access to accurate information about ED patient flow for both the immediate ED team and associated departments, as well as for the EMT. The project goal was SMART [22], i.e., there was a specific goal, the achievement of which would be measurable through the hospital’s electronic patient record [EPR] system and business intelligence system. The target was deemed to be achievable with engagement from all relevant stakeholders and was considered to be relevant as it aligned to hospital data management strategy. The project timeline was set to start November 2019 for completion July 2020.Baseline departmental data at the point of commencing the project indicated:◦In 2019, the ED received 45–55 presentations each day.◦ED visits increased from 8773 in 2017 to 10,100 in 2018, and 11,186 in 2019.◦The percentage of patients admitted to an inpatient bed was 32% in 2017, 29% in 2018, and 27% in 2019.◦The median LOS in the ED in the period 2017–2019 was 3 h 35 min; however, 16% of patients had a LOS of which exceeded 6 h in that period.◦There was no single source for data regarding ED patient flow. The ED management team accessed three different data reports and four different sections of the electronic patient record for information regarding patient flow.In 2019, the ED received 45–55 presentations each day.ED visits increased from 8773 in 2017 to 10,100 in 2018, and 11,186 in 2019.The percentage of patients admitted to an inpatient bed was 32% in 2017, 29% in 2018, and 27% in 2019.The median LOS in the ED in the period 2017–2019 was 3 h 35 min; however, 16% of patients had a LOS of which exceeded 6 h in that period.There was no single source for data regarding ED patient flow. The ED management team accessed three different data reports and four different sections of the electronic patient record for information regarding patient flow.Stakeholder engagement was informed by person-centered collaborative, inclusive, and participatory [CIP] principles [31,32] which have been shown to be synergistic with LSS use in healthcare [15,33,34]. In practice, the project group sought active participation and input from stakeholders in defining a minimum dataset required to make visible the ED patient flow. This dataset was informed by the experts on the ground involved in delivering care rather than defined by a remote management team. The stakeholders were also instrumental in discussing the ideal process for mining and presenting data. Having the IT analyst participate as a key stakeholder from the outset ensured that when decisions were taken concerning data required, the process for extracting this data was assessed and confirmed as achievable/not achievable from the outset. A communication plan was implemented with ongoing stakeholder engagement sessions with all key participants including ED teams, radiology, the EMT, inpatient admitting teams, and with information technology staff included from the outset. The first output from the stakeholder engagement sessions was a high-level process map or SIPOC [21] of visits to the ED [Figure 1]. The SIPOC enabled us to visualize the variables involved in the process and facilitated all stakeholders having a clear, visual reference for what inputs were required in the process to facilitate required outputs to satisfy both patients and staff.The terminology “voice of the customer” [VOC] is used in Six Sigma to denote the expectations of the customer; in healthcare, the customer can be staff or patients or any participant in the delivery or receipt of care [26]. Following five VOC sessions, and building on the SIPOC, a more detailed process map [Figure 2] was developed. The process map was used to highlight areas where data access and availability had a particular impact on patient care and correspondingly patient flow. Six critical areas, highlighted in orange in Figure 2, indicated these areas.The process map was validated by the ED team who worked in and with the process on a day-to-day basis. The process map further facilitated discussion and led to a more detailed capture of the VOC. Bertels [25] discussed how VOC data is gathered, and then mapped onto a LSS tool known as a critical to quality [CTQ] tool. A CTQ tool is designed to capture the key measurable characteristics of a process or service whose performance standards must be met to satisfy the customer. A CTQ tool was completed. The CTQ characteristics of the ED patient flow process were identified as follows:Ease of access to and time taken to access ED patient flow data;ED LOS;TAT for completion of nurse triage, physician assessment, and completion of radiology.Ease of access to and time taken to access ED patient flow data;ED LOS;TAT for completion of nurse triage, physician assessment, and completion of radiology.It was clear from our discussion with the full project team that LOS and TAT [characteristics 2 and 3] were influenced by staff access to real-time data [characteristic 1]. Combining knowledge gained from the SIPOC exercise and CTQ allowed the creation of a data collection plan. Ease of, and timely, access to data was identified as a primary outcome. Processes impacting on patient flow that were influenced by data availability were broken down to achievement of ED triage time and achievement of assessment targets.A Gemba walk is an observation/understanding of where and how the work is done and is an important component of the LSS approach [27]. Gemba were completed by observing the journey of the patient through the ED. Accompanying audits were completed through data mining from the hospital’s EPR system enabling data such as registration time, triage time, physician assessment time, and LOS to be recorded. ED operations were altered dramatically during the COVID-19 lockdown period as the department changed from a walk-in service to telephone triage and appointment-only service for the period 16 March through to 1 August 2020. Gemba and audits were repeated during this period [June 2020] and again in August 2020.Data availability: To source complete datasets regarding the patient flow for one patient required access to four separate clinical records on the EPR. The staff user had to click in and out of four sections of the patient record, including consultant records, medical records reports [which include non-consultant hospital doctors reports], diagnostic imaging reports, and assessment forms [which include nursing records]. Three different data reports were involved, an ED nursing report, ED doctor report, and ED patient experience time report. Data access to inform decision-making was observed as taking an average of 9 min per patient (n = 45). With the ED averaging 45 patients per day, collating this data takes between 5 and 6 hours of ED staff time.Table 4 shows the ED patient volumes and acuity, as well as key turnaround times.Achievement of targets: In August 2020, a Gemba of patient flow within the ED confirmed the following median times:◦From arrival in the ED to triage was 17 min;◦To physician assessment was 38 min;◦TAT to completion of radiology was 2 h, 5 min;◦Median LOS was 3 h, 43 min [well below the NEMP 6 h target].From arrival in the ED to triage was 17 min;To physician assessment was 38 min;TAT to completion of radiology was 2 h, 5 min;Median LOS was 3 h, 43 min [well below the NEMP 6 h target].Special cause variation: Lessons learned during the period of the COVID-19 lockdown included:◦LOS targets were not achieved for 10% of patients attending the ED.◦Access to inpatient beds was also not a limiting factor during the COVID-19 lockdown period as bed occupancy was at 54%.◦Access to radiology reports was not a limiting factor in this period as radiology was completed immediately on request. The instantaneous availability of radiology during COVID-19 lockdown was due to general outpatient radiology activity being significantly reduced resulting in the ED having almost exclusive access to radiology resources, in effect special cause variation [35].LOS targets were not achieved for 10% of patients attending the ED.Access to inpatient beds was also not a limiting factor during the COVID-19 lockdown period as bed occupancy was at 54%.Access to radiology reports was not a limiting factor in this period as radiology was completed immediately on request. The instantaneous availability of radiology during COVID-19 lockdown was due to general outpatient radiology activity being significantly reduced resulting in the ED having almost exclusive access to radiology resources, in effect special cause variation [35].The COVID-19 lockdown period taught us some valuable lessons–improving bed availability and access to radiology in isolation would not guarantee a reduction in ED LOS without the ability to improve patient flow within the ED itself, which as outlined was impacted by the availability of data.A failure mode effect analysis [FMEA] and fishbone cause-effect analysis were completed. An FMEA is a product risk assessment that analytically approaches the prevention of defects by prioritizing potential problems and their resolution [28]. The FMEA identified the completion of the ED-based processes including triage and physician assessment, as well as completion of radiology processes as a high risk both in occurrence and detection with scores of 225, 360, and 360, respectively. These scores indicate that there is a high chance of target times for these processes not being met as well as a high chance that deficiencies in these TAT will be undetected.Doggett [36] (2005) wrote that cause-effect analysis diagrams illustrate the possible causes of a particular problem by sorting and relating them using a classification scheme. In this project, the fishbone cause-effect analysis as demonstrated in Figure 3 supported further insight into probable causes of ED TAT not being detected which then had an impact on patient flow.Following analysis, a brainstorming session with the project team agreed to first focus improvements on the measurement of the process. It was agreed that without immediate access to accurate data the impact of further improvements would be difficult to assess. Once accurate, timely data was available, processes such as admission to inpatient bed could be examined and targeted improvements implemented.Stakeholder engagement sessions were conducted to co-create the required data set for the ED and method for presentation. As described earlier, the ED data was available in various forms and reports. It was agreed to utilize a 5s approach to conclude the final dataset. A 5s is a popular tool within the lean paradigm, for organizing spaces so work can be performed efficiently, effectively, and safely. While 5s is intended for a physical work environment, its central function is to organize, standardize, and maintain through visual management [21] which we translated for use in analyzing the current process for data access [Table 5].The 5s exercise (Table 5) was used to illustrate the current state and target states and led to agreement on the following improvements:◦Daily ED patient flow report.◦Reduce from 7 ED patient data sources to 1 report.◦Reduce from 73 general to 37 data points specific to patient flow.◦Available at a set time each day, no data mining is required.◦On the advice of the IT analyst, it was agreed to complete the daily tracker as a first step. The IT build required for a “live tracker” would be extensive. The project team agreed to assess the impact of the daily tracker, re-confirm the minimum dataset, and then proceed to the live tracker.◦Governance structure agreed–report available to ED team including the ED clinical nurse manager, ED nurse coordinator, ED consultant on duty, and ED patient services lead. Availability of this data is important to the ED team in order to monitor patient flow daily and identify and guide improvements. If targets are not met, the ED management is aware immediately and implement timely interventions. The ED management also has data available to share with wider stakeholders to inform wider process change and improvements, for example, negotiate increased access to radiology and guide improvements in the admission process. The hospital EMT utilizes data to inform strategic planning for the ED, for example, data availability regarding patient acuity [demonstrated by Manchester score] informs the need for increased senior decisionmaker staffing; referrals to radiology inform decisions regarding expanding of radiology support and services to the ED.◦Obstacles to the achievement of targets can be managed in a proactive rather than reactive manner.Daily ED patient flow report.Reduce from 7 ED patient data sources to 1 report.Reduce from 73 general to 37 data points specific to patient flow.Available at a set time each day, no data mining is required.On the advice of the IT analyst, it was agreed to complete the daily tracker as a first step. The IT build required for a “live tracker” would be extensive. The project team agreed to assess the impact of the daily tracker, re-confirm the minimum dataset, and then proceed to the live tracker.Governance structure agreed–report available to ED team including the ED clinical nurse manager, ED nurse coordinator, ED consultant on duty, and ED patient services lead. Availability of this data is important to the ED team in order to monitor patient flow daily and identify and guide improvements. If targets are not met, the ED management is aware immediately and implement timely interventions. The ED management also has data available to share with wider stakeholders to inform wider process change and improvements, for example, negotiate increased access to radiology and guide improvements in the admission process. The hospital EMT utilizes data to inform strategic planning for the ED, for example, data availability regarding patient acuity [demonstrated by Manchester score] informs the need for increased senior decisionmaker staffing; referrals to radiology inform decisions regarding expanding of radiology support and services to the ED.Obstacles to the achievement of targets can be managed in a proactive rather than reactive manner.In addition to the above key improvements, some quick wins were also identified, for example, more bedside computers were purchased to allow for bedside completion of electronic radiology referrals/radiology reports, and ED/radiology operations team meet weekly to identify and agree on the need for extra ED specific slots.A control plan was devised to support and monitor continued improvements. The impact of the availability of ED patient flow data was monitored through stakeholder feedback, monitoring compliance with turnaround targets as well as informing strategic decisions. Achievement of the ED patient flow targets was reassessed in March, May, and August 2021.One organization-wide ED activity report is now circulated to the EMT and the ED team each morning at 9 am as per sample in Figure 4. Key metrics including patient volumes, Manchester score, LOS exceeding 6 and 9 h, achievement of triage and assessment targets, and radiology volumes are captured in this report. The report has reduced time spent compiling patient flow reports from 9 min per patient to 0 min. At the commencement of this project, the ED department saw from 45 to 55 patients per day. This equated to 405–495 min of nursing time occupied with compiling reports which now have been reduced to zero minutes. The report is immediately available. Reducing time spent on data management releases the ED nurse manager time for other duties including patient care, staff support, as well as service improvement and development. Availability of accurate relevant data allows the ED team to identify areas for improvement in patient flow.In terms of activity, the number of ED presentations increased monthly during the control period from 929 presentations in March 2021 to 1154 presentations in August 2021.Patient acuity was largely unchanged, the majority of patient presentations categorized into Manchester score 3, requiring urgent but not immediate care.Median time to completion of triage, assessment, and LOS increased during the control phase as a consequence of increased patient presentations [Table 6]. Time to completion of assessment and length of stay remains within the National Emergency Medicine Programme targets of 3 h to completion of assessment and 6 h for LOS. Time to triage falls outside the National Emergency Medicine Programme target of 15 min.By definition, an ED is unpredictable. Optimizing patient flow is dependent on processes within and outside the ED operating efficiently and effectively. The main learnings from this improvement are as follows:There is no single factor upon which ED LOS will succeed or fail.Proactive/on-the-spot visibility over the process, through data, information, and knowledge sharing, is essential for optimizing patient flow.There is no single factor upon which ED LOS will succeed or fail.Proactive/on-the-spot visibility over the process, through data, information, and knowledge sharing, is essential for optimizing patient flow.Availability of the ED daily report allowed the stakeholders to gain an accurate insight into the needs and challenges which confront the ED daily. The ED clinical nurse manager no longer has to spend 9 min per patient data mining to establish bottlenecks in patient flow. This information is immediately available to the ED team and also to members of the EMT. VOC feedback from improved reporting include:“The new report is great, when I see triage and assessment times are slipping, I can follow it up immediately with the team member” [ED clinical nurse manager].“Tell me about triage scoring–can we use it to better predict admission requirements for ED patients” [commercial director].Availability of this information has helped inform decisions regarding inpatient bed allocation including earmarking specific beds to the medical admissions unit.Healthcare is frequently described as fragmented or siloed, and this is reflected in how data is captured, managed, and shared throughout the system. Ward et al. [37] noted that data relating to business performance, quality, and patient safety is extracted from different systems, and its primary use is to inform senior decision-makers about organizational-level performance rather than to support those at the front line in understanding and improving their daily performance. It is estimated that up to 30% of the total health budget may be spent on handling data and information, i.e., collecting it, looking for it, and storing it [38]. This study identified significant resources dedicated to handling data; however, the process for translating that data to meaningful information regarding patient flow and making that information available to frontline staff is onerous and time-consuming, limiting the knowledge gained related to the process.Data regarding ED patient flow must be translated into accessible knowledge and ultimately wisdom, as Ackoff outlined in his classic data, information, knowledge, understanding, wisdom hierarchy [39]. This knowledge and wisdom can drive performance improvement at the team/unit level. At the commencement of this process improvement, we collected data which was occasionally processed by an individual and disseminated at the time of crisis. As the process improvement enters the control phase, we now use the data in an organized team manner to create meaningful knowledge about the ED processes. Following Ackoffs theory, this will further evolve into a shared understanding between the ED team and the EMT which will guide further ED process improvements. Strome (2013) described the challenge of information overload and the need to harness data to improve clinical and organizational performance [40]. This process improvement is a first step in harnessing data re ED patient flow. The availability of timely, relevant, accurate, complete, valid data [4] regarding ED patient flow has given both the local ED and the EMT knowledge regarding ED patient flow and has helped inform decision making regarding ED operations; a key function of data gathering in Ackoff’s theory. The next step is to make this data available in real-time which will allow immediate interventions when challenges to ED patient flow arise.At the commencement of this process improvement, VOC and stakeholder engagement sessions pointed at perceived bottlenecks in the system such as access to radiology as well as inpatient beds as main areas to address. Lessons learned during the COVID-19 lockdown prompted the project team to investigate further and gain more knowledge from the data available. We avoided the temptation to jump to immediate conclusions [such as add radiology slots or reserve inpatient beds for patients awaiting admission from ED]. Instead, we took an “outside-in” perspective. We recognized the need to see the data regarding ED patient flow from an external perspective and utilize knowledge gained to co-create solutions to challenges identified [41].Future areas to focus on, include the following:Process:
+          ◦Access to data regarding referrals for admission per specialty will give insight into the potential benefits of the system-wide “fast track” admission process to specialties, for example, assess the impact of suggested medical admission pathway prior to reporting of radiology [42,43].◦Access to data regarding outpatient follow-up requirements per specialty will inform the requirement to reserve appointments for outpatient follow-ups as an alternative to admission.◦In this area of improvement, we will employ a more complex interdepartmental application of lean, as we utilize the voice of the customer across admitting and outpatient teams, observe the process for inpatient admissions, as well as outpatient follow-ups and work with stakeholders to implement change [43].
+		  Staff/team working
+          ◦A second-generation project will examine the potential role of advanced nurse practitioners in EDs. Advanced nurse practitioners have been established in public EDs with a proven impact on improving patient flow and delivery of care [44]. The knowledge regarding ED patient flow made available from this project will give the second-generation team a platform to examine what tasks could be shared from the ED consultant and non-consultant hospital doctor team to advanced nurse practitioner, potentially improving the TAT to completion of the ED assessment, and therefore improving LOS.
+		  Process:
+          ◦Access to data regarding referrals for admission per specialty will give insight into the potential benefits of the system-wide “fast track” admission process to specialties, for example, assess the impact of suggested medical admission pathway prior to reporting of radiology [42,43].◦Access to data regarding outpatient follow-up requirements per specialty will inform the requirement to reserve appointments for outpatient follow-ups as an alternative to admission.◦In this area of improvement, we will employ a more complex interdepartmental application of lean, as we utilize the voice of the customer across admitting and outpatient teams, observe the process for inpatient admissions, as well as outpatient follow-ups and work with stakeholders to implement change [43].
+		  Access to data regarding referrals for admission per specialty will give insight into the potential benefits of the system-wide “fast track” admission process to specialties, for example, assess the impact of suggested medical admission pathway prior to reporting of radiology [42,43].Access to data regarding outpatient follow-up requirements per specialty will inform the requirement to reserve appointments for outpatient follow-ups as an alternative to admission.In this area of improvement, we will employ a more complex interdepartmental application of lean, as we utilize the voice of the customer across admitting and outpatient teams, observe the process for inpatient admissions, as well as outpatient follow-ups and work with stakeholders to implement change [43].Staff/team working
+          ◦A second-generation project will examine the potential role of advanced nurse practitioners in EDs. Advanced nurse practitioners have been established in public EDs with a proven impact on improving patient flow and delivery of care [44]. The knowledge regarding ED patient flow made available from this project will give the second-generation team a platform to examine what tasks could be shared from the ED consultant and non-consultant hospital doctor team to advanced nurse practitioner, potentially improving the TAT to completion of the ED assessment, and therefore improving LOS.
+		  A second-generation project will examine the potential role of advanced nurse practitioners in EDs. Advanced nurse practitioners have been established in public EDs with a proven impact on improving patient flow and delivery of care [44]. The knowledge regarding ED patient flow made available from this project will give the second-generation team a platform to examine what tasks could be shared from the ED consultant and non-consultant hospital doctor team to advanced nurse practitioner, potentially improving the TAT to completion of the ED assessment, and therefore improving LOS.An appreciation of the system [43], in which inquiries are conducted and improvements implemented, was critical to the combined and effective use of the person-centred and LSS improvement approaches we undertook. As well as contributing to the use of LSS in access to and use of LSS, we feel this study contributes to the wider body of knowledge in the use of LSS and person-centred approaches, an under-researched area [45,46].The evolving LSS culture in the hospital supported a system-wide approach to improving access to ED data. Rather than working in isolation, the ED team worked across silos involving patient services, radiology, information technology, and EMT, as well as the lean practitioner in analyzing the process and formulating solutions.It is recognized that busy hospital staffs often work in departmental silos and do not see the entire service [46,47]. However, LSS can facilitate breaking down these barriers to facilitate a system vision or perspective. According to Graban (2012, p. 1).“Lean is an approach that can support employees and physicians, eliminating roadblocks and allowing them to focus on providing care. Lean helps break down barriers between disconnected departmental ‘silos,’ allowing different hospital departments to better work together for the benefit of patients”.The strengths of this project were the stakeholder involvement. Changing from what could be construed as a silo approach to analyzing ED performance to involving all stakeholders in the process. The EMT sponsorship supported the project team in suggesting change. The EMT participated actively in contributing to stakeholder engagement sessions regarding minimum dataset/data requirements.This process improvement is one part of a wider improvement plan to reduce ED length of stay which is ongoing. This study, however, has given a solid platform to understanding ED patient flow. We can now analyze the system factors and relationships around ED patient flow and implement informed solutions.Optimizing patient flow through EDs is a key target for any healthcare organization. Identifying and addressing challenges in isolation is unlikely to lead to success. ED patient flow is multifactorial. An understanding of the challenges and opportunities for improvement at each stage of the process is essential. Through adapting a LSS approach including cross-disciplinary stakeholder engagement, rigorous data analysis, and person/process centered improvements, we have begun the process for improving LOS in hospital ED. System vision and awareness and person-centered approaches contributed to a wider understanding of the factors involved. Ongoing control and monitoring of this improvement will be required which will also identify further avenues for improvement within and outside the ED. This will contribute to the ongoing development of a lean culture in the organization.All authors made a proportional contribution. Conceptualization, A.D., C.R. and S.P.T.; methodology, A.D. and S.P.T.; formal analysis, A.D.; investigation, A.D.; resources, A.D. and C.R.; data curation, A.D.; writing—original draft preparation, A.D.; writing—review and editing, A.D., S.P.T., M.W. and M.M.; visualization, S.P.T.; supervision, S.P.T.; project administration, A.D. All authors have read and agreed to the published version of the manuscript.This project received no funding.This work took place as part of an ongoing organizational quality improvement.Informed consent was not required.The authors acknowledge all staff members immediately involved in this project; Richard Clerkin, Niamh Moffatt, and all ED and IT staff members. We thank other Beacon Hospital Green Belts for their invaluable support and advice. We would also like to thank the UCD Mater Lean Academy; Sean Paul Teeling, Maria Ward, Martin McNamara, and Beacon Hospital.The authors declare no conflict of interest.SIPOC.Process map of emergency department patient flow.Fishbone cause-effect analysis.Sample of the ED daily report.Manchester Triage.LSS Tools.Project team.Activity, patient acuity, patient flow data for the emergency department January, June, and August 2020.5S.Activity, patient acuity, patient flow data for the emergency department control phase–March, May, and August 2021.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11031.txt

@@ -0,0 +1 @@
+Artisanal and small-scale miners (ASMs) labour under archaic working conditions and are exposed to high levels of silica dust. Exposure to silica dust has been associated with an increased risk of tuberculosis and silicosis. ASMs are highly mobile and operate in remote areas with near absent access to health services. The main purpose of this study was to evaluate the prevalence of tuberculosis, silicosis and silico-tuberculosis among ASMs in Zimbabwe. A cross-sectional study was conducted from 1 October to 31 January 2021 on a convenient sample of 514 self-selected ASMs. We report the results from among those ASMs who attended an outreach medical facility and an occupational health clinic. Data were collected from clinical records using a precoded data proforma. Data variables included demographic (age, sex), clinical details (HIV status, GeneXpert results, outcomes of chest radiographs, history of tuberculosis) and perceived exposure to mine dust. Of the 464 miners screened for silicosis, 52 (11.2%) were diagnosed with silicosis, while 17 (4.0%) of 422 ASMs were diagnosed with tuberculosis (TB). Of the 373 ASMs tested for HIV, 90 (23.5%) were sero-positive. An HIV infection was associated with a diagnosis of silicosis. There is need for a comprehensive occupational health service package, including TB and silicosis surveillance, for ASMs in Zimbabwe. These are preliminary and limited findings, needing confirmation by more comprehensive studies.Artisanal and small-scale gold mining (ASGM) occurs in about 80 countries, and there are more than 100 million artisanal and small-scale gold miners globally [1]. Within sub-Saharan Africa, informal employment is the main source of employment in central Africa (91.0%), eastern Africa (91.6%) and western Africa (92.4%) [2].Worldwide, the incidence of silicosis is declining due to a combination of better engineering controls and other interventions to reduce industrial dust levels [3]. On the contrary, epidemiological studies show that 30–50% of workers may be suffering from silicosis and other pneumoconiosis in low- and middle-income countries [3]. The Southern African Development Community (SADC) recognised the triple epidemic of silicosis, tuberculosis (TB) and human immunodeficiency virus (HIV) infection in the large population of miners and ex-miners in southern Africa as needing urgent control across all member states [4]. Countries like Lesotho have extremely high prevalences of silicosis and silico-tuberculosis, 42.5% and 25.7%, respectively, among ex-miners [5]. For many years, Zimbabwe was among the top 30 high TB burdened countries [6].Zimbabwe has an estimated population of more than 500,000 artisanal and small-scale miners (ASMs), with over one and a half million people depending on artisanal and small-scale mining [7]. In 2016, ASGM contributed 45% of the national gold production, contributing substantially to Zimbabwe’s revenue generation [8]. Artisanal and small-scale mining is characterised by excessive exposure to occupational hazards such as mercury- and silica-containing dust [9,10,11]. It is also associated with multiple health conditions such as respiratory, cardiovascular and musculoskeletal disorders, malaria, anaemia, diabetes mellitus and physical injuries [12]. Exposure to silica-containing dust is associated with an increased risk of developing silicosis and/or TB [13]. There is a well-documented increased risk of TB and fungal lung infections in patients with silicosis [13,14,15,16] and the risk of developing TB increases with an increasing severity of silicosis [17]. Silica dust has been classified as a group I carcinogen by the International Agency for Research on Cancer and is associated with lung cancer [18].Silicosis is a permanent and irreversible parenchymal lung disease that results from the inhalation of silicon dioxide or silica in crystalline form [13]. Crystalline silica is a major component of rock and sand, and it is also found in soil, concrete, mortar, granite, other minerals and artificial stone [13,19]. Silica dust is intensely fibrotic, producing a characteristic lesion called the silicotic nodule in the lungs [20]. Inhaled silica dust is ingested by alveolar macrophages, which produce cytokines which stimulate fibroblasts to proliferate and produce increased amounts of collagen, resulting in lung fibrosis. However, the pathological changes in silica-injured lungs are complex and not completely understood [21]. In the early stages of silicosis, a chest radiograph shows isolated opacities against a normal lung parenchyma that can progress over years to a reticulonodular infiltrate. Simple silicosis can progress over several months or years to progressive massive fibrosis [20].Artisanal miners labour under archaic and difficult working conditions and live in extreme poverty [22]. They also have poor health-seeking behaviours and live in overcrowded and poor shelters. Overcrowding and poorly ventilated shafts predispose them to both silicosis and TB. The highly nomadic life of ASMs complicates the provision of comprehensive health services. It may be difficult for this group to adhere to TB treatment since they must attend scheduled clinic visits for at least six months. The providers are not without challenges. One of the challenges is lack of occupational health expertise and services in most primary health care facilities in Zimbabwe [23,24]. It is very possible that many ASMs with silicosis and silica dust exposures, which are risk factors of TB, are being missed. In 2018, TB yields in ASMs attending an occupational health clinic in Zimbabwe were extremely high at 13% [25]. It is quite possible that the proportion of ASMs with silicosis is high. This is plausible because ASMs work in dusty environments with poor ventilation and without adequate personal protective equipment (PPE). There is currently a knowledge gap regarding the proportion of ASMs with silicosis, TB and silico-tuberculosis. The purpose of this study was to evaluate the burden of TB and silicosis among ASMs in Zimbabwe. The specific objectives were to: (i) Determine the proportion and characteristics of ASMs screened for and diagnosed with TB and silicosis; (ii) Assess the relationship between respiratory symptoms, chest X-ray and diagnostic yields of TB and silicosis; (iii) Evaluate the relationship between the duration of exposure to silica-containing dust and the diagnosis of silicosis.This was a cross-sectional retrospective review of occupational health records of ASMs who volunteered to be screened for TB and silicosis at the occupational health clinic at Gweru Provincial Hospital and those consulted during a medical outreach facility at artisanal and small-scale mining sites in the Midlands and Matabeleland South provinces.Zimbabwe is a landlocked country with 10 provinces and 64 administrative districts. In 2019, 60% of the notified TB cases in Zimbabwe were coinfected with HIV [26]. The Matabeleland South and Midlands provinces have high artisanal mining activities. Screening for silicosis and TB was conducted within the context of Zimbabwe’s Kunda-Nqob’ iTB project, funded through the United States Agency for International Development (USAID)’s TB Local Organization Network funding mechanism. The intervention targeted ASMs in the selected project districts of Zvishavane, Insiza, Gwanda and Gweru. Both males and females engage in the same gold mining activities. Therefore, a miner was defined as any person who was involved in the extraction and/or processing of gold. The outreach medical facility was provided at the ASGM mining sites, and all willing miners were free to attend. There was no prescreening of ASMs at either medical facility.All ASMs were assessed for TB using a symptoms screen tool and chest radiographs. The diagnosis of TB was based on a positive Cepheid GeneXpert test (Cepheid, SunnyVale, CA, USA) and/or clinical findings as per national TB Guidelines [26]. Screening for silicosis was done using chest radiographs. Interpretations of the chest radiographs were performed by medical officers trained in the diagnosis of occupational lung diseases (OLD) and TB, and were further checked by a specialist occupational physician trained and experienced in the International Labor Organisation (ILO) classification of chest radiographs. The diagnostic criteria for silicosis were a bilateral multinodular pattern with or without progressive massive fibrosis on a chest radiograph, a positive occupational history of exposure to silica-containing dust and having or not having symptoms of a subtle and progressive shortness of breath or a dry cough, in the absence of any other identifiable disease/s [13]. A diagnosis of silicosis was based on a threshold profusion of ≥1/0 on the ILO classification of chest radiographs [27]. The diagnosis of silico-TB was based on the presence of both silicosis and TB. Since the study was conducted during the COVID-19 pandemic, spirometry testing was not conducted as it is an aerosol-generating procedure. ASMs were captured in the presumptive TB register and the occupational health record. For the mobile outreaches, the presumptive TB registers used were from the nearest clinics within the catchment area where the outreaches were conducted. ASMs diagnosed with TB and/or silicosis were entered in the TB and TB-preventive therapy registers, respectively, and were linked to care through the local health facility.The source population consisted of ASMs working in the Midlands and Matabeleland South provinces in Zimbabwe. The study population consisted of ASMs working in the Midlands and Matabeleland South provinces from selected sites in Zvishavane, Gweru, Insiza and Gwanda who attended an outreach medical facility or the occupational health clinic at Gweru Provincial Hospital between 1 October 2020 and 31 January 2021. The ASMs included self-employed men and women working on an individual basis as well as those working in family groups, in partnerships, or as members of cooperatives.All complete occupational health records of ASMs who were attended to at the occupational health clinic and medical outreach facility were included in the study. The records containing complete demographic details, routine observations and tests such as chest X-ray, HIV and GeneXpert test results were included in the study.Occupational health records with missing chest radiograph results and demographic data were excluded from the study. The records of ASMs seen from outside the stipulated project districts were excluded from the study.All 514 occupational health records of the ASMs that were available during the study period were included in the study. This sample size was higher than the minimum sample size of 385 that was calculated using the OpenEpi software v3.03 (Dean and Sullivan, Atlanta, GA, USA) [28]. The calculation was based on the following assumptions: a prevalence of 50%, 95% confidence interval and a precision of 0.05.Individual level data were extracted from occupational health records of ASMs screened at the two medical facilities. The data variables included age, gender, HIV status, history of TB, respiratory symptoms, comorbidities (any of the following self-reported conditions: diabetes mellitus, chronic obstructive pulmonary disease, hypertension, asthma, cancer and chronic cough), exposure to silica dust, duration of mining, substance and alcohol use. The assessment of exposure to silica dust was subjective and based on whether the ASMs perceived themselves as being exposed to dust or not during mining. The sources of data were occupational health clinical records for ASMs. Data were collected using a pre-coded data proforma.Data were single entered in MS Excel and were exported to Stata v 13.0 (Stata Corporation, College Station, TX, USA) for cleaning and analysis [29]. Categorical variables were summarised using proportions, while continuous variables were summarised using means and standard deviations. The key outcome variables, silicosis diagnosis, silico-TB diagnosis and TB diagnosis, were dichotomous. The Chi-square test was used to test for associations between each of the independent and outcome variables. The modified Poisson regression model with robust variance estimators was used to assess the independent association of each characteristic with key outcome variables after adjusting for the following confounders: age, sex, duration of exposure to silica, comorbidities, respiratory symptoms, HIV status, previous history of TB, substance use and use of personal protective equipment. The associations were expressed as prevalence ratios (PRs) and adjusted PRs. The level of significance was set at 5% (p < 0.05).A total of 641 ASMs were reached during the study period. Of these, 127 records were excluded from the study. The final sample size was 514. The demographic and clinical characteristics of the 514 ASMs who were enrolled in the study are presented in Table 1. There was a predominance of males (85%). Of the 373 ASMs who were tested for HIV, 90 (23.5%) were HIV positive. The modal age category was 25–34 years, which constituted a third of the study population. The mean age (SD) was 37 years (12.7). Almost all ASMs were exposed to silica dust (95%), and just above a quarter (27%) had a duration of employment in artisanal and small-scale mining of at least 10 years. Just under two-thirds (61%) of ASMs did not report any respiratory symptoms.The factors associated with silicosis diagnosis are presented in Table 2. There were 52 (11.2%) patients who were diagnosed with silicosis (95% CI: 8.6–14.4). ASMs who tested positive for HIV were 2.8 times more likely to be diagnosed with silicosis compared to those who tested negative for HIV, adjusted prevalence ratio—aPR = 2.79, 95% CI: (1.01–7.66). Those who reported comorbidities were two times more likely to be diagnosed with silicosis.The factors associated with the development of silico-TB are presented in Table 3. Ten patients from the 52 who were diagnosed with silicosis had developed silico-TB. The prevalence of silico-TB was 2.2% (95% CI: 1.2–3.9). Having a positive HIV status, a previous history of TB and presence of any respiratory symptoms was associated with a silico-TB diagnosis. However, the associations were not significant after adjusting for the other variables.The factors associated with the development of TB in ASMs are presented in Table 4. The prevalence of TB among ASMs was 4% (95% CI: 2.5–6.4). Of the 17 ASMs diagnosed with TB, nine were bacteriologically confirmed, two had pleural effusions and six were diagnosed on clinical grounds. The median [interquartile range (IQR)] duration of employment in ASMs who had abnormal chest radiographs was 8 years (IQR: 3.5–15.0). This was significantly higher than the median duration of 5 years (IQR: 1.0–8.0) of employment among ASMs who had normal chest radiographs, p < 0.001.We have studied a group of 514 ASMs in Zimbabwe during the period from 1 October 2020 to 31 January 2021. We report the results from those attending an outreach medical facility and an occupational health clinic which searched for TB and silicosis, and measured patients’ HIV status. We found a high prevalence of silicosis and TB, despite short periods of exposure to silica-containing dust. An HIV infection increased the risk of silicosis by three-fold.The burden of silicosis and TB is well documented in formally employed miners and ex-miners, but there is a dearth of research and published data on the burden of TB, silicosis, or silico-TB in ASMs [30]. This is the first study to be conducted in Zimbabwe and in Africa among ASMs describing the burden of silicosis, silico-TB and TB. Apart from a study conducted in southern Brazil describing the prevalence of silicosis, we did not find comparable studies focusing on ASMs [31].The prevalence of TB in ASMs in our study was almost 15 times higher than the national TB point prevalence of 275 per 100,000 in Zimbabwe [32]. Studies in Ghana and Malawi reported TB prevalences of 0.9% and 12% [33,34], respectively. However, both studies were not comparable to our study since they focused on formally employed miners and used random sampling techniques. The prevalence of silicosis observed in our study was higher than the prevalence ratio of 3.08 reported in ASMs in southern Brazil [31]. However, our study employed convenience sampling, whereas the Brazil study employed random sampling. This poses a challenge in making valid comparisons between the two studies. Most studies reporting the prevalence of silicosis and TB in Africa were on formally employed miners and/or ex-miners utilising random sampling methods. These studies, though not comparable to our study, show prevalences of silicosis ranging from 3.8% in South Africa [35] to 42.5% and 24.6% reported in former gold miners from Lesotho, 26.6–31% in Botswana and 22–36% in former miners in South Africa [5,36,37,38]. The quoted prevalences for miners in formal employment serve to highlight the differences between the unique and disadvantaged ASMs and miners in formal employment with better occupational health and safety services and working conditions. Our study differs from other studies conducted on formally employed miners and ex-miners reporting silicosis and TB prevalence in many aspects. Firstly, our study population was relatively young. Median ages reported by other studies ranged from 42 to 62 years. Secondly, the median duration of 4 years exposure to silica dust in our study was shorter than durations ranging from 19 to 42 years reported in previous studies [5,36,37,38]. Furthermore, unlike other studies which focused on miners and ex-miners who were mainly underground miners, our study focused on informal ASMs who were actively involved in predominantly surface mining activities.The association between HIV infection and silicosis has not been described previously. This is one of the first studies that focuses on the burden of HIV in ASMs with silicosis. Here we found that just under a quarter of ASMs with silicosis were HIV positive. This is an important finding, especially in the context of high HIV prevalence in Zimbabwe and much higher sero-prevalence amongst ASMs in this study. HIV burden among ASMs is compounded by the fact that ASMs are hypermobile and operate in hard-to-reach areas where access to healthcare services is poor. However, the causal pathway between HIV and silicosis is not fully understood, and this finding needs further exploration.The lack of association between respiratory symptoms and silicosis has been previously described [5]. This study confirms that silicosis can be diagnosed even in asymptomatic individuals who do not present with any respiratory symptoms.Our study was not without limitations. Firstly, some variables had missing data and this could have reduced the power of the study to reach statistical significance in some variables. Secondly, our study focused on active ASMs and this might have inadvertently introduced survival bias since ASMs who died of silicosis were excluded from the study. Thirdly, we employed convenience sampling unlike most studies reporting the prevalence of silicosis, which use mostly random sampling for selecting study participants. Moreover, the study may have suffered from “healthy worker effect” as ASMs who were either sick during study periods or exited the ASMs pool due to ill health were excluded from the study. Sicker ASMs are more likely to have a higher burden of OLD than those who were finally enrolled in the study. Thus, the prevalence of TB and silicosis reported here is likely to be an underestimate. It would have been important to assess how use of respiratory PPE differs among previously treated TB patients and those who were never treated for TB, however, correct and consistent use of respiratory PPE was not captured as a variable in this study. Moreover, it would have been important to assess the pack-years of cigarette smoking in patients, however tobacco smoking was also not captured as a variable in the study. Furthermore, the ASMs selected in this study only underwent a symptom and temperature screen for COVID-19. This could have affected the number of TB cases diagnosed on clinical grounds as COVID-19 can present with similar clinical and pulmonary manifestations. We did not analyse the prevalence of TB by the different grades of silicosis (acute, accelerated, chronic simple silicosis or complicated silicosis).Despite these limitations, our study serves as the first baseline study focusing on the burden of silicosis and TB in ASMs in Zimbabwe, and paves the way for future research in the artisanal and small-scale mining population. Our study has important policy considerations. Since silicosis can be diagnosed among asymptomatic and symptomatic ASMs, enhanced access to radiology services as a basic screening tool for ASMs is crucial. Given that ASMs visit public health institutions when they feel unwell, there is need to integrate occupational health into primary health care. This can be achieved through: (i) Developing guidelines for OLD (or through integrating OLD into TB and HIV guidelines); (ii) Capacitating health care workers to provide health education and screening services to this group.This study has shown that silicosis and TB in ASMs are a huge problem in Zimbabwe and are affecting young people. People aged 25–34 years have a high prevalence of TB in Zimbabwe [6]. The synergistic relationship between HIV and silicosis may fuel the TB epidemic in Zimbabwe since both increase the risk of TB. ASMs who get sick usually travel to their hometowns where they may act as foci for community transmission of TB.Although our cross-sectional findings must be considered preliminary, the prevalence of TB, silicosis and silico-TB among ASMs in Zimbabwe is very high. There is need to provide a comprehensive occupational health service package, including TB and silicosis surveillance, to ASMs in Zimbabwe. Interventions to reduce exposure to silica-containing dust through raising awareness on safer mining methods are needed. This calls for enhanced collaboration between the Zimbabwe Miners Federation, Ministry of Mines, Ministry of Labour and Social Services and the Ministry of Health and Child Care.D.M. and C.T. designed the study. All authors read and approved the protocol. B.C., O.M., F.M., P.N. and A.N. collected the data. C.T. analysed the data. D.M., C.T., C.S., R.M., G.M. and N.S. discussed the results. D.M., R.N. and C.T. wrote the manuscript with input from G.M., F.M., O.M., P.N., A.N., C.S., N.S., R.M., C.Z. and F.K. All authors have read and agreed to the published version of the manuscript.This research was funded by the United States Agency for International Development (USAID), Cooperative Agreement number 72061319CA00003—Kunda-Nqob’iTB Project.Ethical clearance for the study was obtained from the Medical Research Council of Zimbabwe (MRCZ/E/283).Not applicable.The data presented in this study are available on request from the corresponding author. The data are not publicly available due to authorisations that may be required by the Ministry of Health and Child Care in Zimbabwe.This work was done in collaboration with several individuals and institutions. We are grateful to the nurses working in the Gwanda, Insiza and Zvishavane districts and the staff at the occupational health clinic at Gweru Provincial Hospital. The staff from the Union Zimbabwe Trust, and Baines Occupational Health Services are also acknowledged for their cooperation and participation in data collection/provision which made it possible to write up this project. We are saddened to announce the passing on of Christopher Zishiri on the 14th of August 2021 prior to the publication of our manuscript. We value his immense contribution as one of the co-authors.The authors declare no conflict of interest.Demographic and characteristics of participants who were enrolled in a large occupational health outreach programme in Zimbabwe, 2020–2021.‡ = Column percentages. † = Any of the following self-reported conditions: diabetes mellitus, hypertension, chronic obstructive pulmonary disease, cancer, asthma and chronic chest pains. TB = tuberculosis.Factors associated with silicosis diagnosis among ASMs who were enrolled in a large occupational health outreach programme in Zimbabwe, 2020–2021.PR = Prevalence ratio; aPR = Adjusted prevalence ratio; CI = Confidence interval; ‡ = Row percentages. * Records with missing data either in the outcome variable or in the predictor variable were not included in the analysis. ** = statistically significant; † = Any of the following self-reported conditions: diabetes mellitus, hypertension, chronic obstructive pulmonary disease, cancer, asthma and chronic chest pains.Factors associated with silico-TB diagnosis among ASMs who were enrolled in a large occupational health outreach programme in Zimbabwe, 2020–2021.PR = Prevalence ratio; aPR = Adjusted prevalence ratio; CI = Confidence interval; ‡ = Row percentages. * Records with missing data either in the outcome variable or in the predictor variable were not included in the analysis. ** = statistically significant † = Any of the following self-reported conditions: diabetes mellitus, hypertension, chronic obstructive pulmonary disease, cancer, asthma and chronic chest pains.Factors associated with TB diagnosis among ASMs who were enrolled in a large occupational health outreach programme in Zimbabwe, 2020.* Records with missing data either in the outcome variable or in the predictor variable were not included in the analysis. ** = statistically significant; PR = Prevalence ratio; aPR = Adjusted prevalence ratio; CI = Confidence interval; ‡ = Row percentages. † = Any of the following self-reported conditions: diabetes mellitus, hypertension, chronic obstructive pulmonary disease, cancer, asthma and chronic chest pains.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11032.txt

@@ -0,0 +1,3 @@
+This study aimed to assess the relationship between the landscape patterns and non-point source (NPS) pollution distribution in Qixia County, China. The sub-basin classification was conducted based on a digital elevation model and Landsat8 satellite images. Water samples were collected from each sub-basin, andtheir water quality during the wet and dry seasons was estimated. The correlation between the landscape indices and water pollution indicators was determined by Pearson analysis. The location-weighted landscape contrast index (LWLCI) was calculated based on the “source-sink” theory. Qixia was further divided into five sections based on the LWLCI score to illustrate the potential risk of NPS pollution. The results showed that the water quality in Qixia County was generally good. Cultivated land, orchards, construction areas, and unused land were positively correlated with the water pollution index and weredesignated as the “source” landscape categories, while forests, grasslands, and water bodies, which were negatively correlated with water pollution, were the “sink” landscapes; the LWCI was high in 36.94% of the study area. In these areas, measures such as increasing vegetation buffer zones are necessary to decrease the sediment and nutrient loads carried by precipitation.Depending on the source, water pollution can be categorized as point and non-point source pollution [1]. Point source pollution originates from a single source, while non-point source (NPS) pollution is caused by various undefined pollutants, such as soil sediments, livestock fecal sewage, and solid waste [2]. For example, the movement of rainwater or irrigation water can carry pollutants, such as fertilizers, herbicides, and pesticides, into rivers, lakes, reservoirs, and underground waters [3]. With the rapid economic development and high population growth in recent years, NPS pollution has become the main cause of water pollution [4], which mostly carries large amounts of N and P (NPS-N and NPS-P, respectively). The increase in NPS water pollution has caused serious hydrological and environmental problems worldwide and become a major challenge in environmental protection [5,6]. Although NPS has attracted much attention for its effects on human and ecosystem health, it is almost impossible to regulate due to the uncertainty of its sources and the complexity of its transmission process [7]. Accurate monitoring of the changing patterns of NPS pollution is essential for combating, controlling, and reducing water pollution. Studies on NPS pollution have mainly focused on source identification, characterization, hazard assessment, mechanism discussion, and the factors influencing NPS pollution, and mainly involved field surveys, monitoring, and simulation [8]. Research shows that NPS pollution is the main cause of surface water pollution, with agricultural NPS pollution making the greatest contribution [9,10,11]. Landscape patterns can determine the amount of pollutants reaching water bodies through different ecological and hydrological factors [4]. Therefore, landscape pattern change has long been considered as a major factor influencing the extent and transmission of NPS pollution [12]. Numerous researchers have explored this topic through the application of hydrological models and geographic information systems (GIS). Ouyang et al. used the SWAT model to illustrate the diffuse pollution dynamics and their response to land development in the Yangtze River [13]. The Slurp model was utilized by Jain et al. (1998) and Chen (2013) to analyze the load estimation and source apportionment of NPS pollution in the Satluj catchment and Jinjiang River, respectively [14,15]. The sparrow model was incorporated in a study of fluvial suspended sediment transport in the non-tidal streams of the Chesapeake Bay water body and its vicinity [16]. Such modelscan both quantify the amount of NPS pollutants produced and simulate the attenuation and transformation of these pollutants during transport. Although they have been extensively used, they require massive data input for simulation and a large amount of time for manipulation [10].China has pledged to fight water pollution caused by NPSs. The Action Plan for Prevention and Control of Water Pollution (2015) published by the Ministry of Ecology and Environment of the People’s Republic of China includes specific chapters related to combating NPS pollution [17,18]. However, more technologies for NPS pollution control and monitoring in accordance with local landscape features are yet to be developed [19].In 2003, Chen proposed that the “source-sink” theory should be used to better illustrate the correlation between landscape patterns and water quality [20,21]. The location-weighted landscape contrast index (LWLCI) was also proposed by Chen based on the “source-sink” theory [20]. Chen suggested using the Lorentz curve to determine the spatial distribution pattern of “source” and “sink”. The Lorentz curve was originally used to describe the distribution of wealth among different groups of people and demonstrates the accumulation of one factor in an increasing process, as well as other related variables. This theory has been widely used to study many ecological processes, such as NPS pollution, soil erosion, and carbon loss. Jiang et al. analyzed NPS pollution in the Jiulongjiang Estuary by constructing a grid landscape contrast index based on the spatial load contrast index [2]. Wu studied the relationship between soil carbon loss and landscape indicators based on the LWLCI [22]. Chen et al. used the LWLCI to study the interactions between urban landscape patterns and land surface temperatures [8]. Furthermore, Wang examined the effects of land cover change on the soil erosion process [23], and Olivera used the Land-Cover Pollution Index (LCPI) to evaluate the relationship between land use and cover change (LUCC) and water quality [6]. In most of these studies, only LWLCI was calculated based on the land configuration of the research area, and few studies have determined the relationship between the LWLCI and water quality indicators.With technological advancement and economic development, human activities are increasingly affecting the environment. NPS pollution is more likely to be influenced by watershed landscape patterns than point source pollution. This study aimed to identify the key areas for the prevention and control of NPS pollution in the study area based on the “Source and Sink” theory. Mitigation measures based on our results would aid in managing NPS pollution in the watershed, creating a spatial pattern conducive to regional human health, and promoting the sustainable development of the region. The objective of this study was to determine the relationship between water quality and the landscape and provide a scientific basis for authorities to implement measures to ensure the health of residents and local sustainability. To better illustrate the relationship between landscape patterns and NPS pollution, Pearson correlation analysis was conducted between the landscape and water quality indicators, and the LWLCI was calculated to illustrate the quantitative relationship between NPS and landscape configuration. Future mitigation measures were also proposed based on a literature review and expert opinions.Qixia County (120°33′–121°15′ E, 37°05′–37°32′ N) was selected as the study region, which is characterized by mountainous and hilly terrain with a total area of 2017 km2. The altitude in the study region ranges from 53 to 830 m (Figure 1). The main soil types are mountain yellow-brown earth, alluvial soil, and aquic brown soil [24]. Qixia receives 743.1 mm of annual precipitation, four distinct seasons, large amount of sunshine, and a moderate amount of rainfall. The mean annual temperature of Qixia is 11.3 °C, with a frost-free period of 207 d and annual actual sunshine duration of 2760 h. It experiences a warm, temperate monsoon climate [24]. Qixia consists of six main tributaries with two dams, with a total length of approximately 184 km. The water area is approximately 1685 km2. According to the Qixia Yearbook 2018, the average daily water usage per person is approximately 117.82 L, and the annual water usage of the whole county is 4.19 million m3 [25].Qixia County is located in a hilly, mountainous area. The soils of Qixia can be divided into two main categories: brown loam and alluvial soil. Brown loam, a significantly alkaline soil, covers approximately 70% of the city’s arable land area, accounting for 88.6% of the total sown area. Alluvial soils are mostly found in the alluvial plains of river valleys [25]. The intrusive rocks of the valleys and plains in Qixia are mainly fine and medium-grained diorite granites from the Early Yuan Dynasty Luliang [26]. However, groundwater is shallow and vulnerable to environmental change. Therefore, unified management and protection should be strengthened to avoid groundwater contamination.Based on data from the website of the Qixia Government, Longmenkou Reservoir provides the city with drinking water. The main pollutants affecting the water quality are chemical oxygen demand (COD) and total nitrogen, and the overall pH is weakly alkaline [27]. Water quality data are provided in Table 1.As of 2017, Qixia has a total population of 485,400, birth rate of 0.809%, mortality rate of 1.044%, and natural population growth rate of −0.234% [25]. The death rates for men and women were 961.74 per 100,000 and 795.40 per 100,000 in 2017, respectively. In 2014, a total of 1876 confirmed cases of tumors were reported in Qixia, with an incidence rate of 3022.19 per 100,000; for men, the total number of cases and incidence rate were 1123 and 3504.34 per 100,000, respectively, while the number of cases and incidence rate for women were 753 and 2477.80 per 100,000, respectively. The top 10 malignant tumors in order of incidence rate were caused by stomach, lung, liver, colorectal, breast, bladder, uterine body, esophageal, and pancreatic cancer, accounting for a total of 15,911 cases and 84.81% of the total [28]. Qixia County is part of Yantai City. According to Wang, Yantai has the highest incidence and mortality rates of liver cancer among all cities in Shandong Province [29].Water quality is correlated with the incidence of malignant tumors and cancer [30]. A previous study reported that COD and biological oxygen demand (BOD) in drinking water are closely, positively related to the liver cancer death rate in southern China [31]. Forman et al. found that drinking water containing high levels of nitrates could result in the enlargement of the thyroid gland, and increased incidence of 15 types of cancer and two types of birth defects in humans [32]. Several studies have also demonstrated that excessive (NH4)2SO4 and NH4NO3 in drinking water could exert detrimental effects on human reproduction, causing abnormal decreases in body weight [33].A Landsat8 remote sensing image of Qixia County with a resolution of 30 m was acquired from the China Geospatial Data Cloud website. Geometric precision correction, spectral processing, and atmospheric correction of the acquired image were conducted. Based on the Chinese Land Use Classification Standard, the landscape in the study region was classified into seven types: cultivated land, forests, orchards, construction land, water bodies, grasslands, and unused land (Figure 2). The classification was conducted based on visual interpretation to ensure classification accuracy. A digital elevation model (DEM) with a 30-m resolution was acquired from the National Fundamental Science Data Sharing Platform. Annual precipitation and fertilizer application data were obtained from the Chinese meteorological data website and a local yearbook, respectively. Approximately 126,501 t of fertilizer was applied over 875 km2 of total sown land in Qixia2017. Fertilizer usage in China, Shandong Province, and Qixia County is compared in Table 2. The usage data for Shandong and China in 2018 were acquired from the Chinese National Bureau of Statistics inquiry website [34].Both the average fertilizer and pesticide usage in Shandong slightly exceeded the national average, while the usage in Qixia was over three times higher than the national average. The large area of apple orchards in Qixia could be an explanation for this, as the average fertilizer usage for orchards in China during 2014 was approximately 93.19 t/km2 [34]. Shandong applied 1,306,700, 421,300, and 356,400 t of N-fertilizer, P-fertilizer, and K-fertilizer during 2018, accounting for 31%, 10%, and 8% of the total fertilizer usage, respectively. Based on these percentages, we calculated that 39,324.11, 12,678.69, and 10,725 t of N-fertilizer, P-fertilizer, and K-fertilizer was applied in Qixia during 2017, respectively. As the fertilizer utilization rate in China is 30–50% [35], approximately 23,594.46, 7607.2, and 6435.34 t of N-fertilizer, P-fertilizer, and K-fertilizer remained in the soil and river systems of Qixia, posing a great hazard to local public health.Water samples were collected from 61 sites within the study area to better understand the correlation between water quality and landscape distribution (Figure 3). These 61 sites were selected to represent various land-use types and morphologies. According to the Qixia Yearbook, the average precipitation in May is only 7.7 mm, while precipitation in July and August altogether (490mm) accounts for more than 60% of Qixia annual total precipitation [25]. Samples (1000 mL) were collected from each site weekly during May and early September 2016 to reflect the water quality in the dry and wet seasons. The samples were immediately sealed in an incubator after collection and processed for 24 h. The water quality indicators analyzed in this study included the total N (TN), ammonium N (NH4+-N), total P (TP), and chemical oxygen demand (COD). The NH4+-N, and TN, COD, and TP contents were determined using a continuous-flow injection, and following the potassium persulfate oxidation spectrophotometric, potassium chlorinate, and ammonium molybdate spectrophotometric methods, respectively. The tests were repeated thrice for each sample, and the final results for each indicator were recorded after averaging the experimental results.The water quality was determined following the comprehensive pollution index method, which is based on the evaluation of the functional areas of a water environment. The index was constructed via the following steps: (1) measuring individual water quality indicators and determining corresponding water quality standards in each water environment category; (2) applying arithmetic average, multiplication, weighted average, and other mathematical methods to the individual water quality indicators and generating a comprehensive pollution index that reflects water conditions and quality [36]. The weighted average was used to calculate the composite pollution index and quantify the water quality indicators of each sub-basin using the Formula (1) [36]:(1)P=1n∑i=1nPi=1n∑i=1ncisi
+        where P is the comprehensive pollution index, n is the number of river water quality indicators involved in the evaluation, Pi is the pollution index of pollutant category i, Ci is the average value of pollutant category i (mg/L), and Si is the evaluation standard of pollutant category i (mg/L). The higher the value of P, the more severe the pollution condition of the water body [37].The independent-samples t-test is the analysis of whether there is a significant difference in the sample data and is conducted by testing the means of two samples from two independent subjects [22]. In this study, by calculating the comprehensive pollution index of water quality in the dry and tributary streams of different small water bodies, an independent-samples t-test was conducted using river water quality indicators to further elucidate the differences in water quality between the dry and tributary streams of Qixia City.This study used the Hydrology module in ArcGIS 10.2 (ESRI, Redlands, CA, USA) to extract the water body boundaries by combining the DEM of the study area to extract the flow direction of the surface water runoff model, calculate the sink flow accumulation, and generate the river network to partition the water body in the study area.Landscape metrics are a quantitative characterization of landscape pattern characteristics. Landscape spatial pattern variables based on landscape indices are commonly used in the literature [38]. Using Fragstats 4.2, five landscape metrics, i.e., patch density (PD), largest patch index (LPI), landscape shape index (LSI), interspersion and juxtaposition index, contagion index (CONTAG), and Shannon’s diversity index (SHDI), were calculated to illustrate the heterogeneity of the study area. The correlation between landscape patterns and water quality was explored using Pearson analysis in SPSS 19.0.The “source-sink” theory posits that a certain landscape type can be categorized as either a “source” or “sink” based on its corresponding “source” or “sink” function in ecological processes. In this study, for NPS pollution, the landscape types that can provide or increase the volume of pollutants, such as farmland, construction fields, and orchards, were categorized as “source” types, while those that can absorb or alleviate the pollution load, namely forests, water, and grasslands, were categorized as “sink” types. This categorization was verified by the Pearson analysis results [39]. The total area and percentage of “source” and “sink” areas are listed in Table 3.The Lorentz curve diagram is shown in Figure 4 [21]. The O-E-B line represents the absolute even distribution curve of landscape types in the water body; if the “source” and “sink” landscapes are evenly distributed in the basin, the OEB distribution curve would appear. In this case, the effects on NPS pollution from the “source” and “sink” landscapes were the same, and the landscape pattern was theoretically in equilibrium. If the “source-sink” landscape is unevenly distributed in space, as shown by the O-D-B and O-F-B lines (assuming that O-D-B and O-F-B represent the “source” and “sink” landscape types, respectively), their contribution to the monitoring points at the river basin exit could be determined by the area accumulation curve of each landscape type and the area of the irregular triangle composed of the O-C and C-B straight lines. Thus, the LWLI′ can be represented as Formula (2) [20,21]:(2)LWLI′=∑i=1mAsourcei×Wi×APi÷[∑i=1mAsourcei×Wi×APi+∑j=1nAsinkj×Wj×APj]Considering that the relative height, slope, and distance between landscape elements and surface water are the major influencing factors, the LWLCI for these landscape elements can be calculated as Formula (3) [20,21]:(3)LWLCI=LWLCIdistance×LWLCIelevation÷LWLCIslope 
+        where LWLCI represents the comprehensive landscape spatial load comparison index, sink j represents the area of an irregular triangle composed of “source” and “sink” landscape types, m and n represent the number of “source” and “sink” landscapes, respectively, Wi and Wj represent the weight of various “source” and “sink” landscapes, and APi and APj represent the percentage of the area of various “source” and “sink” landscapes in each small water body, respectively.The weight allocation process is mainly based on empirical studies and expert opinions. Wang determined that the weight of a given landscape is related to its significance in the soil erosion process [11]. In light of this, in this study, farmland was assigned a weight value of 0.8 as it is a key pollution source, whereas orchards and construction land were assigned a weight value of 0.6 due to their potential contributions to pollution. The weight values of “sink” landscapes were determined by comparing their effects in mitigating pollution (Table 4). It is important to note that the LWLCI of NPS pollution zoning is a comprehensive index based on elevation, slope, distance, and landscape type, and can better reflect all types of water bodies.The test results showed that 94.73% of the cross-section water quality in Qixia at least met the Class-III standard at a minimum, according to the Environmental Quality Standards for Surface Water established by the Ministry of Ecology and Environment of the People’s Republic of China [40] (Table 5).Combining the distribution characteristics of water bodies and river paths in the satellite image data, Qixia was initially divided into 300 mini-basins. After merging and manually modifying the boundaries of the water bodies in Qixia, they were finally divided into 21 sub-basins, as shown in Figure 5.Spatial correlation analysis of the landscape spatial load contrast index and its correlation with the spatial distribution of the water body water quality index involves a collection of a series of spatial data analysis methods and techniques. Our Pearson analysis results showed that the proportion of land use areas and each water quality indicator were correlated (Figure 6). The areas characterized as cultivated land, orchards, construction land, and unused land were positively correlated with water quality indicators and were the main “source” types of the studied landscape. Forests, grasslands, and water areas had a negative correlation with water quality. The positive correlation of cultivated land and orchards with water quality was greater than that of construction land and water quality. The negative correlation between forests and water quality indicators was greater than that between grasslands and water quality indicators.To better demonstrate the distribution of the LWLCI in Qixia, the “source-sink” landscape spatial index map of Qixia County was created using SPSS 19.0(IBM, Armonk, NY, USA), Excel 2010 (Microsoft, Redmond, WA, USA), and ArcGIS 10.2 (ESRI, Redlands, CA, USA) (Figure 7). Overall, the landscape spatial index was higher in hilly, comprehensive grain-fruit utilization areas in southwestern Qixia, southern Taocun Town, and most of Zangjiazhuang Town, and lower in the low mountain forest protection area in eastern Qixia. The proportion of areas with moderately high LWLCI was the highest (37.47%), followed by areas with moderately low LWLCI, whereas the proportion of areas with high LWLCI was the lowest (4%). The global Moran index of LWLCI was 0.701 (p < 0.01) (Table 6). The correlations between LWLCI and water quality indicators during the wet and dry seasons are shown in Table 7.Our results confirmed that landscape configuration can significantly affect the load and distribution of NPS pollution. Cultivated land, orchards, and construction areas can enhance the transmission of pollutants, while forests, grasslands, and water bodies have the opposite effect. Our differentiation of “source” and “sink” landscapes was mainly based on experience and experimental results. It should be noted that a certain land-use type can shift from a “source” or “sink” or vice versa in the same ecological process depending on the research subjects. For example, in this study, forests acted as “sink” landscapes in evaluating NPS-T or NPS-P, but they may become “source” landscapes in terms of organic matter NPS pollution. This is because the organic matter in forest soil can be transmitted into rivers under heavy rainfall [41]. In this study, organic matter pollution was not considered in this study. Scientifically categorizing “source” and “sink” landscapes and how “source” or “sink” land-use types are interrelated could be explored in future studies [23].Our research showed that cultivated land, orchards, construction land, and unused land were positively correlated with each water quality indicator. The correlation coefficients of orchards were lower than those of most coefficients for cultivated land, which could be due to the higher application of fertilizers and pesticides in cultivated land than in orchards. Furthermore, the looser soil texture and lower amount of vegetation in cultivated land than in orchards can enhance the transition of soil nutrients [28]. Previous studies have shown that cultivated land and orchards have the highest correlation with the water pollution index [10]. This indicates that agricultural pollution is currently an important source of non-point source pollution.The transmission of nutrients from soil to rivers is a process during which the concentration of nutrients before they reach water bodies can be controlled by adjusting the combination of different landscape types in a given spatial pattern, thus reducing the risk of NPS pollution [17]. The landscape pattern index at the landscape level in each sub-basin exhibited some correlation with nutrient indicators in the basin [42]. In this study, patch density (PD), which reflects the degree of landscape fragmentation, exhibited a significant positive correlation with TN. The measurements of COD and electric conductivity indicated that the more complex the landscape type, the better the function of the fixation and retention of nutrients in the water body. LPI characterizes landscape dominance, and LSI shows the landscape shape complexity; the latter had a positive correlation with all nutrient indicators, among which TP and electric conductivity showed a highly significant positive correlation, indicating that the lower the complexity of the landscape composition, the lower the influence on nutrient concentration in the water body. The fractal dimension index area-weighted mean (FRAC-AM) had little correlation with water quality indicators in general, while CONTAG was negatively correlated with all water quality indicators. The greater the CONTAG value, the better the water quality of the river, indicating that there are dominant patches of aggregation and connectivity in the landscape; that is, good aggregation and connectivity of woodland lead to pollution retention, which shows a highly significant negative correlation with TN and electric conductivity. This indicates that woodland has a better TN retention capacity than other land-use types. Furthermore, SHDI was positively correlated with all five water quality indicators, including highly significant positive correlations with TN and TP, and significant positive correlations with conductivity, indicating that the dominant role of woodland as a “sink” landscape decreases with an increase in landscape heterogeneity [22].A comparative landscape spatial load index (LCI) greater than 0 indicates that the pollution output of the “source” landscape exceeds the pollution retention of the “sink” landscape and that there is a risk of non-point source pollution within the sub-basin; the larger the LCI value, the higher the output risk. The opposite is true if the “sink” landscape is dominant and the risk of non-point source pollution is low [8,43]. The value of the global Moran index of LWLCI was 0.701 (p < 0.01), indicating that the LWLCI in the study area and its spatial distribution were intrinsically, positively linked. As can be seen from the results, there was a significant, positive correlation between LWLCI and the water quality indicators of the basin. TN, COD, and conductivity were highly significantly correlated with the LWLCI in both study periods. The correlation between LWLCI and TN was greatest in September at 0.729, and exceeded that in May, which was consistent with the change in NH4+-N and was related to the application of large amounts of N fertilizers during this time period, as well as the transmission of N elements from the soil into the water with the increase in summer precipitation. The correlation between COD in the two time periods and that of LWLCI in the two time periods was not very different. Both exhibited highly significant correlations, indicating that each small trend of COD in the two time periods was more consistent. However, TP and electric conductivity were more correlated with LWLCI during May than in September, and the correlation coefficients decreased over time, indicating that the LWLCI in May reflected the changes in TP and conductivity somewhat better than that in September.The distribution of NPS pollution in the different watersheds of Qixia varied between May and September. According to the correlation between the distribution of land-use types and water quality indicators during May and September, surface runoff and soil erosion caused by precipitation played a role in surface pollution. The fertilization of agricultural land is the main cause of N eutrophication and the decrease in water quality. The usage of land for construction, industrial, and domestic water-use purposes in towns and cities is a source of the total P pollution in rivers. Additionally, the COD and EC indicators in Fengping, where watersheds contain a high proportion of orchards, were found to exceed more than those in other regions, which was likely related to the application of organic fertilizers. Therefore, effective control of all types of land-use practices can be used to control and decrease surface source pollution.The landscape spatial load comparison index constructed in this study did not consider other factors affecting the formation of non-point source pollution, such as rainfall and soil, and our evaluation model is more suitable for areas with similar soil and rainfall conditions. When studying areas with large differences in rainfall conditions and soil properties, researchers should consider landscape patterns and ecological processes. Corresponding technical treatments, such as appropriate weight assignment for spatial variability in rainfall and soil, should be conducted. Moreover, the samples used in this study were collected within one year. In future studies, the dispersion and accumulation processes of surface pollution caused by changes in landscape patterns over long periods of time should be explored.Local authorities should pay more attention and actively respond to the water pollution problem in Qixia County. Many studies have already discussed the mitigation of agricultural NPS pollution. Relevant measures include increasing the application of organic manure fertilizers and improving the awareness of local farmers of the severity of water pollution. Authorities should also establish standards for the application of agricultural fertilizers and pesticides based on local agricultural practices to improve their utilization rate [12].In this study, the weighted contrast index was constructed based on the “source-sink” landscape theory, and the correlation between landscape patterns and water quality was analyzed. Considering the factors affecting non-point source pollution, the risk of non-point source pollution in a mountainous river basin was assessed. The main conclusions were as follows:(1)The water quality indicators of the six main rivers in Qixia County varied across different periods. The TN indicator was the highest, greatly exceeding the surface water V standard; the concentration of pollutants in different sections of the same river showed some temporal and spatial differences. Based on the comprehensive pollution index, the water quality of Qixia County was found to generally be good, and most of the water body belonged to the third category or higher. The water quality of the main stream and its tributaries was not significantly different between May and September. The average values of various water quality indicators of the main stream were higher than those of the tributaries during May but lower during September.(2)The proportion of each land-use type in the 21 sub-basins of Qixia County was moderately different. The proportion of woodland areas was 38.43%, and these areas covered the highest proportion of land in Qixia County. The proportions of different land-use areas had certain correlations with the water quality indicators. Cultivated land and orchards were positively correlated with NPS pollution indicators, while forest land was negatively correlated with water quality indicators. Cultivated land, orchards, construction land, and unused land were the “source” landscapes in the basin. The areas of woodlands, grasslands, and water were negatively correlated with the quality of nutrient salts and were the main “sink” landscapes in the basin.(3)The spatial load contrast index of the “source” and “sink” landscapes showed a significant spatial correlation in the region and had a significant positive correlation with all NPS pollution indicators in the basin. TN, COD, and electric conductivity were significantly correlated during both periods and can be used as indicators of non-point source pollution.(4)The heavily polluted area was located in the hilly comprehensive grain-fruit utilization area in the southwest, covering one-seventh of Qixia County. The potential pollution areas were located in the low mountain forest protection areas of Tingkou Town, Miaohou Town, and Taocun Town. The moderately polluted district area accounted for 80% of the total area of Qixia County, indicating that the NPS pollution situation of Qixia County is generally light. Future mitigation measures that comply with local topographic features should be considered in high-LWLCI regions.(5)Whether the high incidence of liver cancer in the Yantai City region, where Qixia is located, is related to the dietary habits or water quality conditions in the region requires further exploration. The large quantities of fertilizers and pesticides applied in Qixia have greatly affected water quality and the local environment. The high level of nitrogen-related contaminants in the water quality of Qixia also requires the attention of the local authorities. This study provides a scientific basis for further potential measures from the relevant sectors.The water quality indicators of the six main rivers in Qixia County varied across different periods. The TN indicator was the highest, greatly exceeding the surface water V standard; the concentration of pollutants in different sections of the same river showed some temporal and spatial differences. Based on the comprehensive pollution index, the water quality of Qixia County was found to generally be good, and most of the water body belonged to the third category or higher. The water quality of the main stream and its tributaries was not significantly different between May and September. The average values of various water quality indicators of the main stream were higher than those of the tributaries during May but lower during September.The proportion of each land-use type in the 21 sub-basins of Qixia County was moderately different. The proportion of woodland areas was 38.43%, and these areas covered the highest proportion of land in Qixia County. The proportions of different land-use areas had certain correlations with the water quality indicators. Cultivated land and orchards were positively correlated with NPS pollution indicators, while forest land was negatively correlated with water quality indicators. Cultivated land, orchards, construction land, and unused land were the “source” landscapes in the basin. The areas of woodlands, grasslands, and water were negatively correlated with the quality of nutrient salts and were the main “sink” landscapes in the basin.The spatial load contrast index of the “source” and “sink” landscapes showed a significant spatial correlation in the region and had a significant positive correlation with all NPS pollution indicators in the basin. TN, COD, and electric conductivity were significantly correlated during both periods and can be used as indicators of non-point source pollution.The heavily polluted area was located in the hilly comprehensive grain-fruit utilization area in the southwest, covering one-seventh of Qixia County. The potential pollution areas were located in the low mountain forest protection areas of Tingkou Town, Miaohou Town, and Taocun Town. The moderately polluted district area accounted for 80% of the total area of Qixia County, indicating that the NPS pollution situation of Qixia County is generally light. Future mitigation measures that comply with local topographic features should be considered in high-LWLCI regions.Whether the high incidence of liver cancer in the Yantai City region, where Qixia is located, is related to the dietary habits or water quality conditions in the region requires further exploration. The large quantities of fertilizers and pesticides applied in Qixia have greatly affected water quality and the local environment. The high level of nitrogen-related contaminants in the water quality of Qixia also requires the attention of the local authorities. This study provides a scientific basis for further potential measures from the relevant sectors.Conceptualization, W.Q. and C.Y.; methodology, C.Y.; software, M.W.; validation, Y.L. and M.W.; formal analysis, X.Y.; investigation, C.Y.; resources, Y.L.; data curation, Y.L.; writing—original draft preparation, C.Y.; writing—review and editing, Y.L. All authors have read and agreed to the published version of the manuscript.This research received no external funding.Not applicable.Not applicable.Data available on request due to restrictions. The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the field survey and lab processing.The authors declare no conflict of interest.Above to below: location of Shandong Province in China, location of Qixia in Shandong Province, and Digital Elevation Model (DEM) of Qixia.Land-use classification of Qixia.The sixty-one sampling sites selected in this study.Distribution of “source” and “sink” landscape types. Adapted from [21].Sub-basin partitioning of Qixia.Heat maps of the correlation between land use and water quality factors.Location-weighted landscape contrast index (LWLCI) zoning of Qixia.Water quality of Qixia in June 2020 (mg/L).Comparison of fertilizer and pesticide usage in Qixia County, Shandong Province, and China.Area and percentage of each “source” and “sink” land-use type.Weight of each “source” and “sink” land use type.Water quality test results in Qixia.LWLCI classification and percentage of area covered.Correlation between the LWLCI and water quality indicators.* θ < 0.05, significance level Sig(θ) (significant correlation); ** θ < 0.01, significance level Sig(θ) (highly significant correlation); TN, total N, NH4+-N, ammonium N, TP, total P, COD, chemical oxygen demand.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11033.txt

@@ -0,0 +1 @@
+During the COVID-19 pandemic, adolescents could not leave their house freely, meet up with friends, or attend school; previous literature showed that youths under enforced confinement or quarantine were five times more likely to suffer from psychopathological symptoms and use social networks sites (SNs) greatly. This study aimed to verify whether the quality of the parent-adolescent relationship could predict youths’ psychopathological symptoms and their SN use during the pandemic, and to evaluate the possible moderator role of their the capacity to be alone. Seven hundred and thirty-nine (n = 739) adolescents were recruited from the general population during the COVID-19 lockdown in Italy, and they were administered The Capacity to be Alone Scale, The BSMAS, the YSR, and the Perceived Filial Self-efficacy Scale. Our results confirmed a direct effect of the perceived filial self-efficacy on the psychopathological symptoms so that a poorer perceived quality of the relationship with the caregivers predicted higher psychopathological symptoms in youths. Moreover, greater social networks use was predictive of psychopathological symptoms in adolescents. Our results also showed a significant interaction effect between adolescents’ perceived filial efficacy and the capacity to be alone on SN use and on psychopathological symptoms. These results suggest that youths’ response to the confinement during the pandemic is influenced both by individual characteristics (the capacity to be alone) and by relational variables (the perceived filial self-efficacy).The Sars-Cov-2 virus has been spreading throughout the world since February 2020, causing more than four million deaths and impacting people’s interpersonal and social interactions, freedom of movement and travel, as well as work, school, and family habits [1]. In response to the COVID-19 pandemic, governments have implemented disease containment measures such as school closures, social distancing, and home confinement. Youths experienced separation from their classmates, instructors, extended families, and community networks for long periods of time [2]. Although lockdown periods and quarantines have mostly been documented to be associated with psychopathological symptoms in adults [3], recent literature proposed these difficulties to be especially problematic for adolescents, probably due to the particular importance of the peer group for identification and support in this period of development [4,5]. It has been noted that the COVID-19 pandemic can be defined as a potentially traumatic environmental experience challenging individuals’ resilience capacities to cope with distress, uncertainty and subversion of previous and consolidated habits [6]; however, only a few studies have so far specifically investigated the psychological outcomes of the pandemic in adolescents experiencing solitude from their peers and every-day familiar environments (schools, gyms, etc.) during lockdowns and/or quarantines (see for example [7,8,9]). It must be acknowledged that in most cases adolescents had not been completely isolated and alone during the pandemic because they could nonetheless count on their parents’ and/or relatives’ presence. Although adolescents predominantly use peers as references and support, in times of extreme difficulty, youths (especially during the first years of adolescence) might return to their family as a source of comfort, security, and emotional regulation. However, as Goodman and Gotlib have posited, the quality of the relationship between caregivers and adolescents may vary [10], and if the quality of the relationship between youths and their caregivers is low and the environmental context within which the youth lives is not sensitive and positive, affect regulation processes in offspring may be hindered. Thus, youths could not find support in the family and may experience a number of psychopathological symptoms both in the internalizing and externalizing areas [11]. The capacity to be alone, which is one of the most important indicators of emotional maturity [12], has been shown to be a significant buffer that may help individuals cope with the negative impacts of risk variables and distressing situations [13] and to moderate the effect of the low quality of parent-youths’ relationship on adolescents’ psychopathology [14]. The capacity to be alone has also been shown to mitigate the effects of social media overuse on psychological distress in adolescents [15]. Individuals with a high capacity to be alone, who are usually characterized by adaptive emotion regulation processes, seem to be more able to cope with the adverse consequences of problematic use of social network sites (SNs); conversely, those with low capacity to be alone are suggested to suffer more depressive symptoms, anxiety and negative emotional states related to high use of SN, due to a reduced ability to regulating their emotional states. Given the importance of this capacity, however, no study (apart from [16]) has so far focused on the potential (null, positive or negative) effect of this variable on the psychopathological outcomes of adolescents during the pandemic.Adolescents are renowned for using social media to connect with peers (e.g. Instagram, TikTok) and several studies have shown that youths turn to social network sites to cope with negativity [17]. The COVID-19 lockdowns brought both distress (high emotional activation) and boredom (sense of void and confinement with emotional deactivation) in the life of adolescents and cut out their interactions with peers. As both hyperarousal and hypo-arousal are unpleasant degrees of activation and social interactions are central in adolescents’ lives, they may have used SN during the lockdowns to regulate their inner states to reach an adaptive equilibrium and to maintain exchanges with peers. Importantly, previous literature not focused on the pandemic period has shown that the overuse of SN may be associated with a number of psychopathological symptoms in adolescence [18]. However, no study, to the best of our knowledge, has investigated the effect of massive SN use by adolescents during the COVID-19 pandemic. In fact, in a condition where adolescents were forced to stay indoors and have SN as the only form of communication with peers, it could be posited that frequent and prolonged use of social network sites cannot be defined as excessive and potentially problematic. However, it has been demonstrated that the potentially problematic effect of the intense use of social networks is not only linked to the motivation driving adolescents to stay online, so that although the motive to use SN could be considered as adaptive (keep the contacts with their peers), the use itself of such technology could lead to negative outcomes. Indeed, high use of SN may cause over-stimulation, with screen-time putting the nervous system into fight-or-flight mode, causing dysregulation, disorganization, and distress [19]. Moreover, previous literature has not addressed the role of the capacity to be alone in moderating the effect of social networks use on adolescents’ psychological distress during this period. In this study we hypothesized that the perceived quality of the relationship with their caregivers predicted social networks sites use and psychopathological symptoms in adolescents. Moreover, we hypothesized that the capacity to be alone played a moderating role.Seven hundred and thirty-nine (n = 739) adolescents from the general population (Mage = 13.4; SD = 1.2; 51% males) were contacted through social network ads during the COVID-19 lockdown in Italy (from March to May 2020). A consecutive sampling method was used to form a convenience sample, and the inclusion criteria were: (1) no referred psychiatric diagnosis in the subjects and/or in their parents; (2) no medical condition present in the subjects at the moment of the recruitment; (3) no medical and/or psychological treatment pursued; and (4) no COVID-19 contagion in any member of the family and no death of any close relative associated with COVID-19. All recruited adolescents were students. All contacted subjects agreed to participate in this study and their parents or guardian signed the written informant consent, consistent with the Declaration of Helsinki. Before its start, the present study was authorized by the Ethical Committee of Sapienza (N. 0000809-2020). All measures (described below) were administered via an online platform and adolescents filled out all questionnaires remotely. Our study was therefore consistent with the indications of the Horizon Programme 2020 (H2020) that recommended focusing on assessment, prevention and intervention, also via technology-mediated tools and with the COVID-19-related guidelines suggesting interpersonal distancing and remote-administered research.The Capacity to be Alone scale [20] is made up of two inter-correlated l0-item Likert subscales (solitary coping scale and solitary comfort scale) ranging from 1 (never) to 4 (always). This scale is concerned with the specific use of confinement to deal with stress (e.g., “Being alone is not healing for me”). The solitary comfort scale assesses an individual’s emotional comfort or discomfort while alone (for example, “I can’t have pleasure until I’m with someone”). In the present study we chose to use a unidimensional score with higher scores indicating higher capacity to be alone. Cronbach’s alpha for the entire scale in this study was 0.85. The reliability coefficients for solitary coping and solitary comfort were 0.75 and 0.76, respectively.The Bergen Social Media Addiction Scale (BSMAS) [21] evaluates experiences in the use of social media within a 12-month period. It comprises six items rated on a 5-point Likert scale (from 1 = Very rarely to 5 = Very often) and is related to core addiction elements (salience, mood modification, tolerance, withdrawal, conflict, and relapse). The items of this measure usually refer to the last year (e.g., “How often during the last year have you used social media so much that it has had a negative impact on your job/studies?” and “How often during the last year have you felt an urge to use social media more and more?”). In this study, the items were modified to tap the three-month period from March 2020 to May 2020. Higher scores in this scale indicate higher problematic use of SN. The clinical cut-off for this measure is 24 [22]. As in previous studies on the general population [23], in this research, no subject exceeded the clinical cut-off for problematic use and the internal consistency of the scale was good. In this study the Cronbach’s alpha was 0.80.The Youth self-report (YSR) [24] is a self-report questionnaire that covers behavioral and emotional problems. It contains 112 problem items, which are scored on a three-point scale (0 = not true, 1 = somewhat or sometimes true, 2 = very or often true). The YSR total problem scale can be divided into nine syndrome subscales: Withdrawn, Somatic complaints, Anxious/depressed, Social problems, Thought problems, Attention problems, Delinquent behavior, Aggressive behavior, Self-destruct Identity. Withdrawn, Somatic complaints and Anxious/Depressed together comprise a broad ‘‘Internalizing” dimension (31 items), whereas Delinquent and Aggressive behaviors together constitute an ‘‘Externalizing” dimension (32 items). Higher scores on these scales indicate more maladaptive functioning. Some YSR items are included in the ‘‘Other problems” subscale (32 items). For the purpose of this study we only used the Total Problem score, which taps all the symptoms and is a measure of perceived general maladjustment level. In this study Cronbach’s alphas was 0.81.The perceived filial self-efficacy questionnaire measures perceived filial self-efficacy, i.e., youths’ perceptions of their parents’ accessibility, sensitivity, and support in everyday settings as well as in a hypothetical important point in their lives. Higher scores on this tool suggest that parents are regarded to be more supportive [25]. Adolescents’ perceived filial self-efficacy (PFSE) was assessed using 16 questions ranging from “strongly disagree” (ranked as 1) to “strongly agree” (scoring as 7) on a seven -point scale, assessing belief in their ability to discuss personal concerns with their parents even under challenging conditions; nurture positive affective relationships and control negative emotional reactions to them; persuade parents to see their point of view on controversial topics; manage stress caused by parental marital problems; and impact parental views and social behaviors in a favourable way. ‘‘I can persuade my parents to see my point of view when it varies from theirs” is a metric for efficacy in dealing with potentially difficult topics. The measure was developed using knowledge of prototypical scenarios that teenagers face with their parents [26]. In this study, the Cronbach’s alpha was 0.87.Gender and age were included as control variables in this study since previous research indicated that they were strongly connected to the key variables in this study [27].First, we examined the means, standard deviations, and bivariate correlations for all research variables using descriptive statistics and Pearson correlations. Second, we utilized Hayes’ suggested SPSS (IBM SPSS, Version 24.0. Armonk, NY, USA) macro PROCESS (model 8) to evaluate the proposed moderated mediation model [28]. This SPSS macro was used to evaluate mediating and moderating models in numerous studies, and it demonstrated greater statistical testability [29].All of the observed variables’ means, standard deviations, and correlations are listed in Table 1. Social network site use was shown to be negatively linked with the perceived filial self-efficacy and the capacity to be alone; SN use was moreover positively correlated with psychopathological symptoms. The capacity to be alone was linked to a lower level of psychopathological symptoms. Gender and age showed no significant correlation with all of the study variables.The suggested moderated mediation model effect was tested using Hayes’ [28] SPSS macro PROCESS. The major findings are given in Table 2. The total effect model (F (1,738) = 48.12, R2 = 0.45, p < 0.001), the mediator variable model (F (1,738) = 22.15, R2 = 0.35, p < 0.001), and the dependent variable model (F(1,738) = 34.23, R2 = 0.41, p < 0.001) were all significant after controlling for adolescents’ gender and age. Social network use (β = 0.22, p < 0.001) and perceived filial efficacy (β = 0.25, p < 0.001) were shown to be (respectively positively and negatively) associated with psychopathological symptoms. The Sobel test was used to evaluate the relevance of the indirect effect of perceived filial efficacy on psychopathological symptoms via SN use. The findings revealed that SN use significantly moderated the association between perceived filial efficacy and psychopathological symptoms (z = 3.63, p < 0.001).Interaction effects were investigated using the PROCESS macro (Model 8) [28]. There was a significant interaction effect between perceived filial efficacy and the capacity to be alone on SN use in the mediator variable model (B = 0.24, p < 0.001). There was a significant perceived filial efficacy x capacity to be alone interaction impact on psychopathological symptoms in the dependent variable model (B = 0.16, p < 0.05). The ability to be alone affected both the relationship between perceived filial efficacy and psychological difficulties as well as the link between perceived filial efficacy and SN use.The restrictions used to contain the COVID-19 pandemic have resulted in a prolonged period of stress for adults, adolescents, and children [30]. It has been posited that adolescents could use social networks to keep in touch with peers and relatives, and rely on the support of parents and caregivers to cope with these psychological difficulties [31,32]. It has also been suggested that the capacity to be alone is a protection factor from the psychological distress caused by the pandemic [27]. However, to our knowledge, no studies have investigated the links between all of these variables (the quality of parent-adolescent relationship as perceived by the adolescent, SN use, the capacity to be alone, and psychopathological symptoms) in youths during COVID-19 lockdowns.This study aimed to fill this gap in literature and proposed that the quality of parent-adolescent relationships is proposed as the predictor of the psychopathological symptoms and SN use, and the capacity to be alone is suggested as a moderator. Our results confirmed a direct effect of the perceived filial self-efficacy on the psychopathological symptoms so that a poorer perceived quality of the relationship with the caregivers predicted higher psychopathological symptoms in youths. This was an expected result, given the large bulk of literature that had already demonstrated how the low perceived quality of caregiving was a strong predictor of the psychological difficulties in children and adolescents [33,34]. The added value of this study, however, is that no other research had yet shown this effect in adolescents during the COVID-19 pandemic and the conditions of social confinement in which the effects of the quality of the relationship on psychological distress were tested. Therefore, although several previous studies focused on this effect, investigations during these unique times were necessary to confirm former research. Moreover, our results showed that greater social network use was predictive of psychopathological symptoms in adolescents. Numerous previous studies have demonstrated that SN overuse can be associated with psychopathology in adolescents [35,36]. On the other hand, with specific regard to the lockdowns, some authors have posited that the use of social networks constituted a protective factor from the development of psychological difficulties in adolescents during the forced confinement, as they allowed contacts with peers and relatives [37]. In our sample this protective effect has not proven to be present; on the contrary, in our study SN use fostered psychological problems in youths. Notably, no subject exceeded the clinical cut-off for problematic use of the SN at the BSMAS questionnaire; therefore, one cannot assume that psychopathological symptoms were predicted by problematic use of the social networks. Rather, our results showed that significant (yet not clinically problematic) use of SN by adolescents during the lockdown was predictive of higher psychopathological risk. This result has potential important implications for the implementation of social policies for the management of possible future lockdowns because it suggests that encouraging the use of SN during times of social confinement is not necessarily positive and could (partially counter intuitively) lead to poorer psychological well-being in youths. The mechanism undergoing this effect remains unclear and further studies should be performed to disentangle it. However, we can speculate that while during everyday life SN constitutes a valid means for keeping in touch with peers and allows them to get into contact with other adolescents because it is also supported by physical encounters in shared physical environments [38], the imposed social confinement during the lockdown impeded these exchanges and hindered the supportive effect of face-to-face meetings. This hypothesis is potentially confirmed by the further result of our study that showed that SN use in adolescents significantly moderated the association between perceived filial efficacy and psychopathological symptoms. In our sample, adolescents who perceived less sensitivity and support from their parents and used social networks more intensely were more likely to develop more serious psychopathological symptoms. It can be hypothesized that those youths who do not perceive to find emotional comfort from their caregivers during the lockdowns used SN to find support and/or cope with negative emotions. But social network use seems to fail in this task and, on the contrary, fosters more severe symptoms. It is widely known that psychopathology can be prevented by the protective effect of strong relationships (both with parents and with peers), which are effective at buffering distress, leading to better social and psychological outcomes [39]. But strong bonds are usually supported by physical proximity [40]. If social confinement is protracted in time (as in the case of the lockdowns during the pandemic), even those relationships continuing online after starting face-to-face may weaken, therefore diminishing their protective effect towards psychological suffering, and paradoxically adding distress and frustration to adolescents’ experience, eventually leading to higher symptoms.Moreover, our results showed a significant interaction effect between adolescents’ perceived filial efficacy and the capacity to be alone on SN use and on psychopathological symptoms. In sum, adolescents with (perceived) supportive and sensitive parents who also could rely on their capacity to be alone were less likely to suffer significant psychological distress and used SN in a less problematic fashion. These results confirm previous research which posited how the capacity to be alone may make adolescents more able to handle risk factors and distressing situations [16], especially when they can rely on supportive and sensitive caregivers [41,42]. Being capable of coping with negative emotions, we can speculate that these adolescents were less likely to use social networks as a means of affective regulation, therefore using them in a less problematic way.This study has limitations. First, all used measures were self-reported questionnaires that could suffer social desirability, and it might have been informative to have an objective assessment administered by professional psychologists. Second, we could not assess possible parental psychopathology, whereas caregivers’ psychopathological risk is widely recognized as one of the main risk factors for negative outcomes in children and adolescents. In fact, family has a key role in the intergenerational transmission of psychopathological behavior [43,44]. Third, although we proposed a predictive model to describe the links between the study variables, this was a non-longitudinal study, and the interpretation of the effects remains exclusively speculative. Finally, the size of the study was relatively small, considering that it is a community sample and that lockdowns during the COVID-19 pandemic affected millions of adolescents.The results of this study suggest that youths’ response to the confinement during the pandemic in terms of psychopathological symptoms and social network use can be influenced both by individual characteristics (the capacity to be alone) and by relational variables (perceived filial self-efficacy). In fact, adolescents with (perceived) supportive and sensitive parents who also could rely on their ability to be alone were less likely to suffer high psychological distress and used SN in a less problematic fashion.Conceptualization, S.C., L.C.; methodology, L.C.; data curation, L.C.; writing—original draft preparation, S.C., L.C.; writing—review and editing, S.C., L.C. All authors have read and agreed to the published version of the manuscript.This research received no external funding. The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of the Department of Dynamic and Clinical Psychology at Sapienza University of Rome (protocol code N. 809/2020, date of approval: 10 September 2020).Informed consent was obtained from all subjects involved in the study.The data presented in this study are openly available in FigShare at doi:10.6084/m9.figshare.14402444.We thank all late adolescents who agreed to participate in this study. The authors declare no conflict of interest.Descriptive statistics and correlations among all of the study variables.Note. BSMAS = Bergen Social Media Addiction Scale; PFSE = Perceived Filial Self-Efficacy; YSR = Youth Self Report; p < 0.01 **, p < 0.05 *.Regression and moderated mediation results.Note. Unstandardized regression coefficients are reported; p < 0.001 ***.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11034.txt

@@ -0,0 +1,15 @@
+The objective of this paper was to gain novel insights into the complex relationships among Sustainable Development Goals (SDGs) in shaping productivity (GDP/capita) growth. Using dynamic panel regressions on data collected in 138 countries between 2000 and 2017, we found that rising temperatures negatively affect growth and mitigate the impact of other SDGs on growth. We also found that CO2 emissions have a U-shaped relationship with growth; life expectancy negatively influences growth (positively moderated by rising temperatures), and food security positively impacts growth (negatively moderated by rising temperatures). This study highlights the difficulty of simultaneously implementing SDGs and elucidates novel research perspectives and policies to decrease the negative impacts of climate change on socio-economic and environmental well-being.Following the Millennium Development Goals (MDGs), the Sustainable Development Goals (SDGs) were introduced in 2015 to emphasise global efforts toward reconciling economic and social with ecological aspirations [1]. The 17 SDGs comprise 169 targets related to poverty, hunger, health, education, gender equality, water, energy, work and growth, industries, inequality, communities, consumption, climate, oceans, biodiversity, institutions, and international partnerships [2]. Since development goals and targets depend on and influence one another [3], implementing them simultaneously in a coherent manner is a daunting task policymakers have to face. It is unclear how these interlinkages work, or how progress on one goal or target influences other goals and targets through causal relationships and feedback loops [4].Mitigating the impact of global warming is vital for the future of humanity. In 2019, the earth’s surface temperature was around 0.95 degrees Celsius (°C) warmer than the 20th-century average [5]. Temperatures have consistently been among the hottest for years, increasing sea levels and decreasing Arctic ice [6]. As a result of increasing global surface temperatures, weather-related disasters have become much more frequent, and the number of extreme events is increasing yearly [7]. Droughts, storms, and floods caused catastrophic damages worldwide and resulted in almost $129 billion of economic loss in 2016 [8].Rising temperature is likely to have a continued negative effect on societies and economies. Shared socio-economic pathways explore possible paths for climate change projections that could affect a wide range of future trends [9]. Xu et al. [10] forecast that areas inhabited by one-third of the human population could become the hottest parts of the world in 50 years unless greenhouse gas emissions (GHGs) are reduced. Climate change and global warming will lead to the permanent loss of critical resources, droughts and floods, imbalances in ecosystems, extinction of species, and threats to human life [11].Global warming has direct impacts on human productivity, i.e., output per capita (productivity). Roson and Mensbrugghe [12] assessed various climate change effects (e.g., rising sea levels, variations in crop yields, water availability, human health, tourism, and energy demand) and found that the effect of rising temperatures on real GDP is significant and impacts are especially severe for developing countries. Recent studies have focused on the extent to which temperature change and CO2 emissions contribute to per capita growth or total factor productivity (TFP) [13].This paper aims to better understand the impact of rising temperature on productivity growth by examining its direct and moderating role. How rising temperature interacts with other sustainability challenges in shaping growth has received very scant attention. After examining the impact of food security (SDG2), life expectancy (SDG3), and GHG emissions (SDG13) on productivity growth, we tested novel hypotheses vis-à-vis the moderating effect of rising temperature.We estimated two-step dynamic panel regressions built on a Cobb Douglas production function, using a sample of 138 United Nations (UN) member states. The advantage of the dynamic approach is to eliminate the deeper lags of the dependent variable, which reduces the number of observations available while also taking endogenous economic growth into account [14].The remainder of the paper proceeds as follows. In Section 2, we elaborate six hypotheses based on the literature. In Section 3, we present the variables and data analysis method. The results of the regression analyses are presented in Section 4. The final section (Section 5) suggests conclusions and implications for future research and policy.
+          Global warming (rising surface temperature) has a negative impact on productivity growth.
+        SDG13 aims to combat climate change and its impacts by regulating emissions and promoting developments in renewable energy [15]. Our preliminary hypothesis assesses the direct effect of climate change (temperature rise) on growth. It is generally accepted that climate change impacts output and growth substantially, especially in emerging countries and in the long run [16]. Rising temperature influences economic growth directly by reducing agricultural output and crop yields [17], industrial output, and labour [18]. The sensitivity of productivity to climate change could be much higher than predicted by direct damage functions, estimating a 23 per cent decline in global GDP by 2100 [19]. The effects would be overwhelmingly adverse at the end of the century and significantly higher in developing countries [20].
+          Increasing carbon dioxide (CO2) emissions have a negative effect on productivity growth.
+        While the direct impact of CO2 on temperature is well-documented, the overall effect of emissions is not apparent because of feedbacks and complicated interconnections in the ecosystem [21]. The atmospheric CO2 concentration has been seen as a significant contributing factor that causes global warming [22]. It is also widely accepted that economic growth is coupled with increased levels of CO2 emissions. However, the exact nature of the relationship between growth and environmental degradation is not straightforward. While much of the literature deals with the effect of economic output on CO2 emissions, some scholars found reverse causality running from carbon emissions to growth [23]. Developing economies are even more vulnerable, as they use more emission-intensive technologies, ultimately decreasing their economic growth [24].The environmental Kuznets curve suggests an inverted U-shaped relationship between income per capita and environmental quality [25]. The environment gradually degrades as countries increase production, but after a certain level of growth and standard of living, societies begin to improve their relationship with the environment. Some other studies [26,27,28] also indicated a robust non-linear relationship between CO2 and economic growth as captured by TFP growth. However, the relationship between emissions and economic growth is likely to be affected by the myopia of societies, the ability to implement intergenerational transfers, and the externalisation of pollution over borders [29]. Chavaillaz et al. [30] predicted an additional annual loss of labour productivity of about two per cent of total GDP per unit of trillion tons of carbon emitted. However, some individual countries (e.g., China, Japan, and the USA) show a significantly positive relationship between economic growth and carbon emissions [31].
+          Increasing life expectancy positively impacts productivity growth.
+        SDG3 envisions healthy lives and well-being for all people of all ages. It is generally accepted that higher life expectancy is a good proxy of health associated with economic growth. The subject of academic debate is whether an improvement in life expectancy causes an increase in per capita income. On the one hand, improvements in life expectancy generally lead to faster economic growth [32]. On the other hand, Acemoglu and Johnson [33] found that better health conditions trigger faster population growth, which is expected to have a negative impact on productivity. The direction of the effect may depend on the stage of “demographic transition”, after which individuals’ education and fertility decisions start to depend on life expectancy reducing population growth and increasing productivity [34]. The authors argue that most countries today are close to or have passed the demographic transition. Therefore, the effect of increasing life expectancy on per capita income are positive, on average.
+          Higher food security (measured by MDER) has a positive impact on productivity growth.
+        SDG2 aspires to end hunger, achieve food security and improved nutrition, and promote sustainable agriculture. Global hunger estimates are generally based on such indicators as minimum dietary energy requirement (MDER). MDER is the minimum amount of dietary energy (kcal/capita/day) that can be considered adequate to meet the minimum energy needs with low physical activity. The modernisation of industries has improved food supply, and food intake plays a vital role in increasing labour productivity [35]. At the same time, the demand for certain food products sharply increases with economic growth, posing many challenges for food supply chains and food safety [36]. Proper food intake improves health, and better childhood nutrition raises educational attainment, improving productivity through human capital [37].
+          Global warming alters (negatively moderates) the impact of life expectancy on productivity growth.
+        The first four hypotheses evaluate the links between key SDGs (climate, health, and food) and economic growth. In the following two hypotheses, we propose that rising temperature alters the effect of other sustainability goals on growth. The interaction of global warming with health and food safety can manifest through various factors, such as mortality [38], extreme climatic events [39], crime and unrest [40], damage to infrastructure [41], as well as adaptation efforts and the production of more expensive carbon-free energy technologies [42]. The causal link between health (life expectancy) and growth may also depend on the ecological consequences of rising temperature [43]. For example, global warming changes the abundance and habitats of organisms that transmit diseases, i.e., vectors, which can shift the seasonal occurrence of several infectious diseases (e.g., malaria, dengue fever, West Nile virus) and cause them to spread. The adverse health effect of rising temperature is exacerbated by crowding, food, and water scarcity [44], and a much depends on the adaptation skills of public health systems, including vaccines and therapies [45].
+          Global warming alters (negatively moderates) the impact of food security on productivity growth.
+        Rising temperatures can alleviate the positive effect of food security, and the impacts fall disproportionately on the poor [46]. Warmer temperatures can increase the speed of insect proliferation, increasing the need for food security measures and crop protection [47]. The stability of whole food systems may be at risk under global warming because of short-term variability in supply, which aggravates food insecurity in areas vulnerable to hunger and undernutrition [18]. Access to food and drinking water can indirectly affect household incomes through damage to health [48].Table 1 presents the descriptions and sources of the variables. Our dependent variable, productivity growth, is calculated by dividing the natural logarithm of real GDP at constant (2011) prices by labour force. Explanatory variables were collected from various sources such as the Penn World Table (9.1) [49], World Bank Databank [50], and Food and Agricultural Organization (FAO) Database [51].The examined variables’ descriptive statistics (mean, standard deviation, minimum and maximum values, skewness, and kurtosis), and the pairwise correlation matrix of dependent and independent variables can be found in Appendix A (Table A1 and Table A2). The (LLC) test of stationarity statistics was also applied to a subset of the panel data to examine whether the series contained unit root [52].We employed an unbalanced panel dataset of 138 UN countries (see Figure 1) for the period between 2000–2017. The selected countries cover about 71.5 per cent of the countries in the world, giving the study overall global representativeness. Time-invariant features of static regression methods in the panel data can cause bias across countries [53]. Generalised method of moment (GMM) approaches are better than fixed-effect regression estimates for analysing panel databases, and dynamic methods used to quantify the impact of climate change on economic growth have recently emerged [54]Similarly to that employed by Mankiw et al. [55], we assumed a Cobb-Douglas production function (Equation (1)). The notations are standard:(1) Yt=Kt∝(AtLt)1−∝The sum of total income [Yt], physical capital [Kt], and labour [Lt], determined by past accumulation in period t. α and (1 − α) are the elasticities of capital and labour, and constant returns to scale (0 < α < 1) are assumed in this model. The technical efficiency of production is denoted by [At], as the residual output, which is not explained by expenditures of labour and capital used in production. Both sides of Equation (1) should be divided by [Lt] to determine output per worker (productivity) [yt] Equation (2):(2)yt=ktAt1−∝The term [k] represents the capital/labour ratio as the amount of capital available per unit of labour input. The dynamics of capital intensity [k*] is equal to the amount of [sk] and the unit of effective labour (n + δ + g) that needs to be invested in preventing [k] from falling [56]. Thus, depreciation [δ] and technological change [g] are assumed to have little effect on the estimates, resulting in a constant (0.05) increase in employment growth [n]. Likewise, the steady-state level of productivity [yt*] corresponds with [k*] Equation (3):(3)yt*=At(skn+δ+g)∝/(1−∝)In our model specification, the economy will tend to return to long-term equilibrium. The steady-state prediction takes logs (ln) from both sides of Equation (3), and the relationship between the explanatory variables is now linear Equation (4):(4)lnyt*=lnAt+∝(1−∝)[lnsk−(n+δ+g)]Transform and arrange Equation (4) into a linear formula at country i, and time t Equation (5):(5)lnyi,t*=βo+β1lnski,t−β2ln(n+δ+g)i,t+β3lnAi,t+εi,t
+      where [lnyi,t] is the dependent variable of GDP per capita at constant prices, lnski,t is the ratio of gross capital formation per GDP, and [ln(n + δ + g)] is calculated as the sum of the growth rate of employment. [lnAi,t] denotes the exogenous rate of TFP (total factor productivity), the model remaining after the capital accumulation. TFP can capture the impact of climate change, emissions, life expectancy, and food security on productivity.The frequent misunderstandings about the neoclassical model is that it fails to explain the catching-up countries. However, the real explanation for the economic growth needs to be derived from the model, which can be understood as the changes that the economy itself (endogenously) forms [57]. Arellano and Bond [58] proposed a generalised method of moments (GMM) model that uses instrumental variables to resolve the endogeneity problem of inconsistencies. Following this dynamic approach, lagged dependent and predetermined variables are used as exceptional instruments. The number of instruments and the maximum lag of the independent variables are limited to avoid overestimating [59]. The two-step (2SGMM) estimators are preferred to the less efficient one-step ones such as least square (LS) and maximum likelihood (ML) [14]. 2SGMM is less likely to be mis specified, and it is more flexible as it does not impose any restrictions on data distribution [60].After taking the first differences of the dependent variable, the above Equation (5) was transformed as follows Equation (6):(6)Δlnyi,t∗=βo+β1Δlnyi,t−1+β2lnski,t−β3ln(n+δ+g)i,t+β4Tempi,t                                              +β5CO2i,t+β6CO2sqi,t+β7+lnLifei,t+β8lnMDERi,t+εi,t
+      where the dependent variable [yi,t] is the growth ratio of real GDP per capita of the country [i] in the period [t]. The first independent variable refers to the lagged dependent variable. The second concerns the share of investment in output. [n] is the average growth rate of employment, and [δ] + [g] is assumed to be constant (0.05). [Temp] denotes the average temperature change. [CO2] refers to carbon dioxide emissions. [CO2sq] is included to test the potential quadratic relationship between productivity and emissions. [Life] is life expectancy at birth, and [MDER] is the minimum dietary energy requirement.Table 2 contains the results of the dynamic regression estimations based on Equation (6). The significant Wald-tests validate the dynamic approaches’ exact choice. Wald-tests imply that a GMM estimator is appropriate in all models, and empirical results can be relied upon for statistical inference [61]. Autocorrelation tests are performed by AR(2) for second-order serial correlations. In all models (1–8), the results demonstrate that all estimators are free from serial correlations and are well-specified. The Sargan tests demonstrate the lack of over-identifying restrictions, and instruments are lower than the number of countries. Therefore, such violations from mean stationarity are not detectable [62]. Assuming economic growth theories, an increase in GFCF as the proxy of the investment rate (sk) positively impacts productivity growth, and employment growth (n + δ + g) is negatively related to the dependent variable in both models.The coefficients of temperature change (Temp) are relatively small and range from −0.005 to −0.072, negatively affecting productivity growth. If the average surface temperature rises from 0 to 1.5 °C, productivity decreases by 0.008 units (0.022 to 0.014), keeping all other variables constant, which is approximately 64 per cent lower than without a temperature rise.Results also show a U-shaped relationship between productivity growth and CO2 emission: a negative relationship between CO2 and growth, but a positive one between CO2 square and growth (Models 3 and 4). The overall t-test (value = 2.11**) also supports the presence of a curvilinear U-shaped relationship. First, as pollution increases, productivity growth decreases (negative relationship) until a local minimum, and afterwards, growth starts to increase again (positive relationship). Models (5–8) indicate that the life expectancy coefficient is significant; however, its sign is negative in all regression models. If life expectancy increases by one unit, GDP per capita will decrease by 0.119–1.197.Models 7 and 8 show that food security (MDER) positively contributes to GDP per capita growth, while rising temperature negatively affects it. We also found significant two-way interaction effects between temperature change and MDER. Figure 2 and Figure 3 plot these interaction effects; solid and dashed lines indicate significant differences between slopes, based on Dawson [63]. The influence of MDER is more substantial (steeper) at low-temperature change than high-temperature change; hence rising temperature negatively moderates (decreases) the impact of MDER on productivity. Food security has a weaker (positive) impact on growth if global warming increases.Similarly, we found significant interaction effects between temperature change and life expectancy. Figure 3 shows that both increasing life expectancy and higher temperature change negatively affect productivity growth, i.e., growth is the smallest in countries where both temperature increases and life expectancy are relatively high. More interestingly, global warming seems to mitigate the negative effect of life expectancy on growth, indicated by the difference in slopes. Life expectancy has a weaker (negative) effect on growth if global warming increases.This study examined how critical sustainable development goals (SDGs) interact in shaping economic growth. We tested the effects of global warming, CO2 emissions, life expectancy, and food security on productivity growth and the interaction of increasing temperature with life expectancy and food security. A dynamic panel regression (Arellano and Bond) model estimates multidimensional data with longitudinal properties. This method eliminates the problem of adding deeper lags of the dependent variable, reducing the number of observations available. Contrary to previous approaches, we also considered the moderating effects of global warming, which is necessary for exploring the underestimated relations between socio-economic and environmental challenges.Our results indicate that global warming negatively affects GDP per capita growth (H1). Carbon dioxide emissions have a U-shaped relationship with productivity growth (H2). In addition to its direct negative impact, global warming also mitigates the effects of other SDGs on growth. While life expectancy negatively influences growth (H3), it is positively moderated by global warming (H5a). Food security positively impacts growth (H4), which is negatively moderated by global warming (H5b). Hence, our data and analysis support H1, H4, and H5b, partly support H2, and reject H3 and H5a.The adverse effect of increasing temperature on living standards urges policymakers to combat climate change and its devastating impacts worldwide. However, Hasegawa et al. [64] claim that in vulnerable regions such as sub-Saharan Africa and South Asia, implementing stringent climate mitigation policies impacts global hunger and food consumption more adversely than the direct adverse effects of climate change.The Paris Agreement of 2016 aims to strengthen the global response to the threat of increasing temperature by keeping its increase below 1.5 °C compared to pre-industrial levels [65]. According to our results, if the temperature rises from zero to 1.5 °C, productivity will decrease by 64 per cent. Moreover, the thresholds of heat exposure that will lead to declined labour productivity are likely to be exceeded in warmer parts of the world, which are often developing countries [30]. The most severely affected regions are tropical areas, such as Southeast Asia, North Central Africa, and northern South America.Contrary to Rigas and Kounetas [24], we found that the relationship between CO2 emissions and GDP per capita growth is robustly curvilinear at a global scale. Lower CO2 emissions and higher productivity growth are typical of developing (African) countries, while higher pollution couples with higher growth in China, India, and the United States. Several developed countries (e.g., Western European and Scandinavian countries) tend to implement innovative green technologies to accelerate sustainable growth with decreasing levels of GHGs. Meanwhile, countries rich in fossil fuel (e.g., in the Middle East) will also show faster per capita growth with higher environmental degradation [66]. Lower-income countries are more vulnerable to climate change and endure more substantial economic losses than higher-income ones [67].We found that food security positively influences growth, but life expectancy has a negative impact on it. It appears that most countries have not achieved the demographic transition yet when population growth decreases because of increasing life expectancy. In the long run, increasing life expectancy will contribute to productivity growth due to accelerated human capital accumulation [68].Results also suggest that besides its direct (negative) effect, rising temperature moderates the economic impact of other SDGs. Global warming can exert influence on other SDGs in many ways as it impacts health, food and water scarcity, weather conditions, heat exposure, and diseases [42]. SDGs should not be studied in isolation as there are complex interdependencies among them. For example, global warming exacerbates poverty, especially in developing countries, where the incidence of agriculture and other outdoor activities is relatively higher [12]. Climate change negatively affects food quality due to rising temperatures and declining plant growth periods [69]. Global warming also impacts precipitation, which influences soil moisture content and groundwater balance [70]. The effects of long-term climate change, including extreme frosts and sub-optimal temperatures, on the earlier occurrence of flowering and the phenology of (potato) vegetables, transform food distribution and waste undesirably [71]. Food security problems mainly occur in land-based developing countries due to unsustainable arable land usage and irrigation systems. In contrast, land degradation is enhanced by extreme weather conditions such as drought, environmental pollution caused by human activities, and deteriorating soil quality [72].There are several potential policy implications of the findings. Countries have decided to rebuild their economies; they can only become cleaner, greener, healthier, safer, more resilient, and sustainable by adhering to recovery plans [73]. The post-pandemic recovery requires nations to discover innovative solutions and complex scientific approaches for a more profound, systematic shift towards a more sustainable economy [74]. Climate-positive actions need to trigger the trajectory of atmospheric CO2 levels; for instance, green investments accelerate the decarburisation of all aspects of the economy [75]. The availability and accessibility of food, clean water, and better sanitation and hygiene services (WASH) are keys to preserving health and well-being [76]. Rapid progress in reducing hunger and malnutrition over the next decade could pave the way for eradicating extreme poverty through other SDGs [77]. Responses should include making economic recovery packages more resilient to future crises and updating global environmental governance to reverse the degradation of ecosystems worldwide [2].This study is mainly limited by omitted variable bias, as the variables in the models reflect only a few SDGs (climate, health, food) that we consider vital for the future of humanity. We urge researchers to test further interactions between SDGs (e.g., education, water, energy, innovation, consumption, production, institutions) to help policymakers implement them simultaneously by minimising trade-offs. The Cobb–Douglas production function can also be replaced by a constant elasticity of substitution function or an alternative green growth approach.Future research also needs to consider theories from various disciplines (e.g., green growth, climate theories) to obtain results and develop global indicators that reflect the complexity of SDGs and their potential future trajectories. Finally, growth should only be supported if it ensures that the people and our planet will continue to provide resources and environmental services for the well-being of all.Conceptualisation, D.M. and A.N.; methodology, D.M.; software, D.F.M.; validation, D.M., A.N. and D.F.M.; formal analysis, D.M.; investigation, D.F.M.; resources, D.M.; data curation, D.M.; writing—original draft preparation, D.M.; writing—review and editing, A.N.; visualisation, D.M.; supervision, A.N.; project administration, D.F.M.; funding acquisition, D.M. All authors have read and agreed to the published version of the manuscript.This research was funded by Technology, National Research, Development and Innovation Fund of Hungary, the K_19 funding scheme and was supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences.Not applicable.Not applicable.Publicly available datasets were analysed in this study. Data can be downloaded from: https://www.rug.nl/ggdc/productivity/pwt/, accessed on 27 September 2021; http://databank.worldbank.org/data/, accessed on 27 September 2021; http://www.fao.org/faostat/en/#data, accessed on 27 September 2021.This research was supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences.The authors declare no conflict of interest.Descriptive statistics of examined variables.Notes: *** p < 0.01. LLC denotes Levin–Lin–Chu test.Pairwise correlation matrix of dependent and independent variables.Notes: *** p < 0.01, ** p < 0.05, * p < 0.1.Countries in the sample (selected UN member states are marked with red).The two-way interaction effects for MDER and moderator (temperature change).The two-way interaction effects for life expectancy and moderator (temperature change).Description, Abbreviations, and Sources of Examined Variables.Sources: based on [49,50,51].Dynamic panel regression results of Equation (6).Note: z statistics are in parenthesis, *** p < 0.01, ** p < 0.05, * p < 0.1.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11035.txt

@@ -0,0 +1 @@
+Objective: The purposes of this paper were to (a) develop a new short, theory-driven, version of the physical activity enjoyment scale (PACES-S) using content analysis; and (b) subsequently to measure the psychometric properties (construct validity, internal consistency, test–retest reliability, and concurrent validity) of the PACES-S for adolescents. Methods: Six experts used a four-point Likert scale to assess the content validity of each of the 16 items of the physical activity enjoyment scale according to a provided definition of physical activity enjoyment. Based on the results, exploratory factor analysis was used to analyze survey data from a longitudinal study of 182 individuals (Measure 1 of Study 1: 15.75 ± 3.39 yrs; 56.6% boys, 43.4% girls), and confirmatory factor analysis (Measure 2 of Study 1: 15.69 ± 3.44 yrs; 56.3% boys, 43.7% girls) was used to analyze the survey data from a cross-sectional study of 3219 individuals (Study 2; 15.99 ± 3.10 yrs; 47.8% boys, 52.2% girls) to assess the construct validity of the new measure. To assess the reliability, test–retest reliability was assessed in Study 1 and internal consistency in Study 1 and 2. For the concurrent validity, correlations with self-reported and device-based physical activity behavior were assessed in both studies. Results: Four out of sixteen items were selected for PACES-S. Exploratory factor analysis and confirmatory factor analyses identified and supported its factorial validity (χ2 = 53.62, df = 2, p < 0.001; RMSEA = 0.073; CFI = 0.99; RFI = 0.96; NFI = 0.99; TLI = 0.96; IFI = 0.99). Results showed good test–retest reliability (r = 0.76) and internal consistency (a = 0.82 to 0.88). Regarding concurrent validity, the results showed positive correlations with a physical activity questionnaire (Study 1: r = 0.36), with a physical activity diary (Study 1: r = 0.44), and with accelerometer-recorded data (Study 1: r = 0.32; Study 2: r = 0.21). Conclusions: The results indicate that PACES-S is a reliable and valid instrument that may be particularly useful to measure physical activity enjoyment in large-scale studies. It shows comparable measurement properties as the long version of PACES.The scientific evidence allows the conclusion that physical activity (PA) during adolescence contributes to developing a healthier lifestyle in later life, reducing the prevalence of non-communicable diseases and improving psychological well-being [1,2,3,4]. According to the WHO recommendations on the health benefits of PA, adolescents should accumulate at least 60 min of moderate to vigorous PA per day [5]. However, only a minority of adolescents report engaging in PA at a level compatible with the health guidelines [6,7,8]. Moreover, while many adolescents start PA programs to improve their health and lose weight, the rate of dropouts is high [9]. Specifically, regarding the maintenance of PA, researchers emphasize the role of affective processes [10,11,12]. Notably, there is a large volume of studies describing the critical role of enjoyment in PA [13,14,15,16,17,18,19]. Despite extensive research demonstrating the importance of PA enjoyment, to date, however, there has been little consensus on what PA enjoyment actually is [20].In general, enjoyment can be regarded as an emotion. There has been a long debate about how emotion might be defined. One study has collected a long list of definitions of emotion, none of which have been able to gain general acceptance [21]. One well-known study stated that a distinction should be made between an automatic affect and a full-blown emotion [22]. While automatic affect represents a simple and rapid appraisal that something is good or bad, positive or negative, emotions are more deliberate, slow, and involve cognitive processes. Although there is currently no universal definition of emotion, most scientists agree that emotions always represent a valenced state of relatively short duration and are related to an object, person, or activity [23]. Based on the component process model [24], five emotion components can be distinguished: cognitive appraisal, physiological responses, action tendencies, motor expressions, and feelings (also called subjective experience). The components are recursively influenced by appraisal processes, contributing to their consistency and synchronization [24,25]. All these changes are then integrated and centrally represented as feelings [25], which are then further categorized and labeled as emotional terms (e.g., enjoyment). That is, the feeling component is considered the most central component of emotion, which differentiates it from other psychological states [26]. Based on these theoretical considerations, we define PA enjoyment as a positively valenced emotion directed towards PA associated with feelings such as pleasure, joy, and fun [27].In measuring PA enjoyment, the physical activity enjoyment scale (PACES) is the most prominent instrument. While the original version was developed by Kendzierski and DeCarlo [28], several alternative forms have been developed (see Table 1 for a comparison). In detail, the original unidimensional 18-item PACES [28] was validated for its validity and reliability (α = 0.93) in college students (aged between 18–24 years). However, a factor analysis of the PACES in the youth sports population(aged between 10–17 years) showed that the scale was not unidimensional [29]. After evaluating by a focus group, two items were removed [30], one of which (“I was very absorbed in the activity”) was removed because it was considered to be irrelevant to PA enjoyment, the other (“It is very invigorating”) was removed because it was considered redundant. However, the study also reported that the 16-item PACES fitted a unidimensional model with methodologic effects behind positively worded items [30]. Given this, Dishman et al. [31] eliminated the positively worded items reducing the scale items to seven and identified sufficient construct validity of the seven-item scale in a sample of US adolescents. However, one study argued that many scale shortening studies do not start from a conceptual point of view but place excessive credit on statistical techniques [32]. Then PA enjoyment was defined as a positive response to the movement experience or an optimal psychological state that leads to performing PA [33]. Raedeke noted that the 18-item PACES appears to tap not only into PA enjoyment (i.e., PA enjoyment reflects feelings about exercise and is a psychological state directly connected to an eliciting stimulus—the exercise experience) itself but also the potential antecedents and consequences of PA enjoyment. Therefore, content analysis with four experts was implemented to shorten the 18-item PACES, and ten items were removed because they were considered not to be the generalized state of enjoying PA or the experience itself. However, the inclusion of an item, “I was very absorbed in the activity,” conflicts with Motl et al.’s [30] results (“I was very absorbed in the activity” was removed because the content was considered not relevant to enjoyment). Furthermore, Raedeke [33] only reported the item-total correlation and did not attempt to identify other psychometric properties (e.g., construct validity, test–retest reliability, and concurrent validity). In summary, various forms of PACES have been developed for which different limitations have been identified (e.g., the inadequate conceptualization of the PA enjoyment, the methodological effect of positively and negatively worded items). It can be assumed that the methodologic effect is based on an inadequate conceptualization of the construct enjoyment and that the items of PACES might contain contents of further similar constructs [27]. To address these limitations, we argued that it might be helpful to use the definition mentioned above of PA enjoyment as a starting point to develop a new, shortened scale based on the long versions of PACES.The purpose of this article was to provide a new form of PACES, using those items that are in line with the definition of PA enjoyment as “PA enjoyment as a positively valenced emotion directed toward the PA associated with feelings such as pleasure, joy, and fun.” This implies a reduction of items since we are only interested in those items that truly reflect the subjective experience of PA enjoyment. We believe it could be further beneficial because it can reduce the burden on participants and be more easily used in large-scale studies [34,35]. Hence, the first aim of this paper was to use content analysis to preliminary develop a new short scale. Based on the results of this procedure, the second aim was to measure the psychometric properties of the shortened scale. These include (a) construct validity, (b) internal consistency, (c) test–retest reliability, and (d) concurrent validity. To achieve these aims, first, experts were asked to evaluate the content validity of the individual items of PACES based on the definition of the provided PA enjoyment. Subsequently, the data collected in two studies [36,37] (the original authors and project director were contacted to obtain the original PA and PA enjoyment measurement data) were used to determine the psychometric properties of the new PACES.According to Lynn [38], at least five experts were required to provide sufficient control over the chance agreement. Therefore, six experts were selected. Four of these six experts held doctoral degrees in sports science, three of which hold professorships in sports psychology (based in Germany or Switzerland), and one held a research fellowship in sports management in Germany. The other two experts were a Ph.D. student in sports psychology and a master student in sports science in Germany, respectively. To determine the content validity index, the definition of PA enjoyment (i.e., PA enjoyment is a positively valenced emotion directed toward PA associated with feelings such as pleasure, joy, and fun) was provided based on the component process model [24]. Experts were explicitly asked to consider whether negatively worded items (e.g., it is not fun at all) could also measure PA enjoyment. A modified four-point Likert scale (1 = “does not match the definition”; 2 = “matches the definition somewhat well”; 3 = “matches the definition quite well”; 4 = “matches the definition very well”) [39] was used to assess the content validity of each of the 16 items [30,40]. By calculating the results of the experts’ evaluation, a new short version of the German PACES would then be preliminary developed, subsequently referred to as PACES-S.The statistical analyses of content validity were performed in Microsoft Excel [41] using the formulas below.A four-point Likert scale, clearly labeled with the definition of PA enjoyment and the content of each item, was sent to each expert separately. They were invited to rate the relevance of each item according to the definition of PA enjoyment independently. Based on the experts’ evaluation results, ratings of 1 or 2 for each item were considered unacceptable, and 3 or 4 were considered acceptable [38]. Two types of content validity indices were used to assess and delete items: (a) item-level content validity index (I-CVI; i.e., the number of experts assigned Grade 3 or 4, divided by the total number of experts) [39]; (b) the scale-level content validity index calculated by the average method (S-CVI/ Ave; i.e., the average proportion of items assigned either Grade 3 or 4 across judges) [42].When N experts evaluated one item, of which n1 experts assigned it a rating of 1 or 2 and n2 assigned it a rating of 3 or 4 (N = n1 + n2), the I-CVI could be computed as:I-CVI=n2NHowever, the results derived from the above equations ignored the chance agreement. Therefore, Polit and Beck [42] and Wynd et al. [43] advocated adjusting I-CVI calculation and using k* to denote the adjusted I-CVI results. To compute k*, the probability of chance agreement (Pc) was calculated first. The formula was as follows:Pc=N!n2!N−n2!.5NNext, k* was computed using the I-CVI and Pc:k*=I-CVI−PC1−PcThen, if a scale had n items and the data value was I-CVIi (i = 1, 2, …, n), then we had:S-CVI/Ave=1n∑i=1nI-CVIiFinally, k* and S-CVI/Ave were employed to evaluate the acceptability of the scale in item level and overall level, respectively. With six experts, the evaluation criteria for k* were as follows: below 0.40 indicated “poor” validity, 0.40 to 0.59 indicated “fair” validity, 0.60 to 0.74 indicated “good” validity, and greater than 0.74 represented “excellent” validity [44,45]. Polit and Beck [42] recommended that a scale should be composed of items with k* of 0.74 or higher and S-CVI/Ave of 0.90 or higher.Based on the content validity evaluated by six experts, four out of sixteen items have been selected. All these four items showed k* higher than 0.74, and the S-CVI/Ave of the PACES-S was 0.96 (see Table 2). The items included in the PACES-S were: “I enjoy it”, “I find it pleasurable”, “It is very pleasant”, and “It feels good”.The data of two cohort studies [36,37] were used to determine internal consistency, test–retest reliability, construct validity, and concurrent validity of the PACES-S developed in Phase 1. The subjects’ PA enjoyment and PA data were measured in Study 1 (Measure 1, Measure 2) and Study 2, respectively.ParticipantsA total of 182 students (male, n = 103, female, n = 79) aged between 11–17 years were recruited for this study. All students came from a comprehensive secondary school in a German city, with all three types of the traditional German tripartite secondary school system: Hauptschule, Realschule, and Gymnasium. After the teachers had agreed, and according to the Helsinki Declaration, informed written consent was obtained from the participants and their parents or guardians before entering the study [46]. The study was approved by the ethics committee of the Charité Universitätsmedizin Berlin. Detailed information on the data collection techniques and quality of the sample are presented elsewhere [36].ProcedureParticipants provided their personal information (e.g., age, gender, school type). They also completed the MoMo physical activity questionnaire (MoMo-PAQ) and the PACES-S twice (Measure 1, Measure 2; Measure 1 and Measure 2 correspond to the PACES-S administered before and after seven days, respectively) at school, with a 7-day interval between the completions. During these seven days, participants wore accelerometers and completed Previous Day Physical Activity Recall (PDPAR; [47]) daily. This study was performed between April and July 2009.MeasurementPhysical activity enjoyment. The 16-item PACES was used in this study [30,40]. However, based on the results of the content analysis described above, we only included the four items of PACES-S (i.e., Item 1: I enjoy it; Item 2: I find it pleasurable; Item 3: It is very pleasant; Item 4: It feels good) [40]. The items were answered using a five-point Likert scale ranging from 1 = “strongly disagree” to 5 = “strongly agree”.PA questionnaire. Habitual PA was measured by MoMo-PAQ [36]. This questionnaire contained 28 items and measured PA in four distinct settings: daily PA, school PA, PA in and outside organized sports clubs. For each setting, the frequency, duration, intensity, and types of PA were measured. MoMo-PAQ has been shown to be a validated instrument with acceptable reliability (test–retest reliability = 0.68) and significant correlations with accelerometer-recorded data (r = 0.29) [36].PA diary. The PDPAR [47] is a self-reporting and time-based recall instrument designed to capture adolescents’ previous day’s PA. In the present study, certain hours of a day were divided into one-hour metric blocks. Participants were instructed to note their specific activities (38 activities were listed for participants to select from, which could be grouped into six main clusters: eating, sleep/bathing, transportation, work/school, spare time, PA) and the intensity of activity for each time block (light, moderate, vigorous, very vigorous). Finally, the metabolic equivalent (MET) levels were computed to determine each participant’s PA. The instrument has proven to be valid and reliable in measuring PA [47,48].Accelerometer. The Actigraph GT1M accelerometer (Pensacola, FL, USA) was also used to measure PA. It is a two-axis accelerometer with a solid-state sensor and micro-electro-mechanical system with a dynamic range of 0.05–2.5 G and frequency range of 0.25–2.5 Hz. The filtered acceleration signal was digitized, rectified, integrated (calculating the ‘activity count’), stored, and reset at user-specified intervals (10 s for the present study). Ultimately, we evaluated the participants’ daily PA based on the duration and intensity of PA (light < 3 METs, moderate 3–6 METs, vigorous 6–9 METs, very vigorous > 9 METs) measured and calculated by accelerometers. In particular, the duration of moderate, vigorous, and extreme vigorous PA per day was combined into a single variable as “accelerometer-recorded MVPA”. The accelerometers were worn around the participants’ waists via elastic waistbands. Participants were requested to wear the devices for seven consecutive days of waking hours (except for swimming and bathing). Measuring PA with the Actigraph GT1M has been proven valid and reliable for adolescents [49,50]. Eligible accelerometer data should meet the criteria that: (1) participants wore the accelerometer for at least 10 h per day over a minimum of 5 days, and (2) non-wearing was defined as at least 60 consecutive minutes of zero activity intensity (1–2 min of counts between 0 and 100 were allowed).To replicate the reliability and validity analyses of the PACES-S in Study 1, psychometric properties of the measure were also assessed using data from Study 2 [37].ParticipantsThe German Health Interview and Examination Survey for Children and Adolescents (KiGGS) is part of the Federal Health Monitoring System conducted by the Robert Koch Institute (RKI) and consists of regularly conducted nationwide surveys among children, adolescents, and young adults aged 0 to 29 years and living in Germany. KiGGS Wave 2 was conducted between 2014 and 2017. The Motorik-Modul Study (MoMo) is a submodule of the KiGGS study and aims to assess physical fitness, PA, as well as determinants of PA in children and adolescents [51]. The whole study sample was drawn from the German resident population aged 4 to 17 years (only subjects aged between 11 and 17 years were selected for this study) using a two-stage cluster sampling approach. Informed consent to participate in the study was obtained from the participants and their parents or guardians. In addition, participants from the baseline study (2003–2006) and Wave 1 (2009–2012) were reinvited. A detailed description of the study design and sampling procedure can be found elsewhere [37,52,53]. KiGGS and MoMo provide nationally representative data of PA and sedentary behavior of children, adolescents, and young adults living in Germany [52]. A favorable vote of the ethics committee of Karlsruhe Institute of Technology of 23 September 2014, is available for the study. A total of 3219 participants (male, n = 1538, female, n = 1681) aged between 11–17 years were recruited for this study.ProcedureParticipants provided their personal information (e.g., age, gender, school type) and completed the PACES-S after physical fitness tests [54]. After completing the scales, participants were assigned to wear accelerometers for eight days to record their PA data (data measured on the first day were discarded. This study was performed between 2014 and 2017.MeasurementEnjoyment. Enjoyment was measured using the PACES-S described in Study 1.Accelerometer. PA was measured using the Actigraph GT3X, the successor accelerometer model described in Study 1. The technical and methodological details of the accelerometer measurement of Study 2 can be found elsewhere [52]. In short, placement of the device was on the hip, sampling frequency was 30 Hz, the same filter as in Study 1 was used, epoch lengths was 1 s with the possibility to convert into 5 s, 10 s, 15 s, 30 s, and 60 s, non-wear time definition was the algorithm by Choi et al. [52], and the valid datasets needed eight hours of recordings on four weekdays and one further weekend day when wearing the device for seven days. Sedentary and physical activity intensity classification used algorithms by Evenson et al. [55] and Romanzini et al. [56]. In addition, the number of days that each participant met the WHO physical activity recommendation level (i.e., Daily MVPA greater than 60 min; [5] over seven days was combined into a new variable, “PA compliance days”.For psychometric properties, we evaluated the internal consistency, test–retest reliability, construct, and concurrent validity of the PACES-S.Internal consistency. The PACES-S data from Study 1 (Measure 1, Measure 2) and Study 2 were used to analyze internal consistency in SPSS 25 [57]. The internal consistency was assessed by examining Cronbach’s alpha coefficient [58]. An acceptable alpha value would be in the range of 0.70 to 0.90 [59,60].Test–retest reliability. The PACES-S scores measured twice a week apart in Study 1 were used to calculate Pearson correlation coefficients in SPSS 25. A 5% cut-off was taken for significance, whereby a value greater than 0.70 was deemed to be acceptable [61].Construct validity. Factor analyses were conducted to assess construct validity based on the results of the PACES-S from Study 1 (Measure 2) and Study 2. Data from Study 1 (Measure 2) were used for an exploratory factor analysis (EFA) to explore the underlying structure of the PACES-S in SPSS 25 [62]. Then, data from Study 2 were used for a confirmatory factor analysis (CFA) to validate the identified factor structure in AMOS 25 [63,64]. Firstly, the factors were extracted in EFA using the principal component method with varimax rotation. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were employed to test the appropriateness of the factor analysis [65]. Missing data ranged between 0.5–2.7% for the PACES items. Further, the specific evaluation criteria were as follows: (1) the factor loading of an item was not less than 0.6 [66]; (2) the number of factors was determined using a scree plot [67] and the following criteria: eigenvalue greater than 1 [68,69], an individual factor accounting for no less than 10% of the total variance, and a composite of the extracted factors accounting for no less than 70% of the total variance [70]. Secondly, CFA was used to validate the structure obtained in EFA using full-information maximum likelihood estimation. This method yields less biased estimates than classical missing data procedures, such as list-/pairwise deletion or means imputation [71]. Missing data ranged between 1.9–2.5% for the PACES items. Given the high sensitivity of Chi-square statistics in large samples [72], the following fit indices and criteria were used to examine the goodness of fit of the model (it was considered good if the following criteria were satisfied): root mean square error of approximation (RMSEA) between 0 and 0.08 [73]; comparative fit index (CFI), normed fit index (NFI), relative fit index (RFI) and Tucker-Lewis Index (TLI) between 0.95 and 1, and incremental fit index (IFI) over 0.90 [74,75,76].Concurrent validity. The concurrent validities for PACES-S were derived by computing Pearson correlation coefficients between PACES-S scores (Measure 2) and criterion scores for MoMo-PAQ [36], PDPAR [47], and accelerometer (“accelerometer-recorded MVPA”) in Study 1. Simultaneously, the correlations between results on accelerometers (“accelerometer-recorded MVPA” and “PA compliance days”) and PACES-S provided the estimate of concurrent validity in Study 2. A 5% cut-off was used for significance, with four levels of interpretation for correlation-based effect sizes: very small (r < 0.1), small (0.10 ≤ r ≤ 0.30), moderate (0.30 ≤ r < 0.50), large (0.50 ≤ r) [77].Descriptive StatisticsOf the 182 participants, 103 (56.6%) were males, and 79 (43.4%) were females. Regarding age distribution, 111 (61.0%) were between 11 and 13 years old, 71 (39.0%) were between 14 and 17 years old. Different types of schools accounted for the following percentages of participants: Hauptschule (14.8%), Realschule (30.8%), and Gymnasium (54.4%). As can be seen in Table 3, the overall data of 174 PACES-S data and participants were valid (missing or invalid data: PACES-S (time 1), n = 8, PACES-S (time 2), n = 8, accelerometer, n = 2; PA questionnaire, n = 0, PA diary, n =0). Concerning males only, 100 (97.1%, missing or invalid data, n = 3) and 98 (95.1%, missing or invalid data, n = 5) participants’ PACES-S data were valid for time 1 and time 2, respectively. All (n = 103, 100%) male participants’ accelerometer, PA questionnaire, and diary data were valid, 101 (98.1%) male participants’ accelerometer data were valid (missing or invalid data, n = 2). For females only, 74 (93.7%, missing or invalid data, n = 5) and 76 (96.2%, missing or invalid data, n = 3) participants’ PACES-S data were valid for time 1 and time 2, respectively. All (n = 79, 100%) female participants’ accelerometer, PA questionnaire, and diary data were valid.Internal ConsistencyAs seen in Table 3, for Measure 1 of Study 1, the overall Cronbach’s alpha for the PACES-S was 0.83, for male participants, the Cronbach’s alpha for the PACES-S was 0.82, and the Cronbach’s alpha for the PACES-S for female participants was 0.85.For Measure 2 of Study 1, the overall Cronbach’s alpha for the PACES-S was 0.86, for male participants, the Cronbach’s alpha for the PACES-S was 0.87, and the Cronbach’s alpha for the PACES-S for female participants was 0.83.Test–Retest ReliabilityThe stability coefficient of the PACES-S for a one-week interval was found to be significant and sufficiently high (r = 0.76, t = 15.14, df = 165, p < 0.01).Construct ValidityIn EFA, the results of Study 1 (Measure 2) showed KMO=0.80, Bartlett’s test of sphericity χ2 = 313.18, df = 6, p < 0.001, indicating that the data were suitable for the factor analysis. Following the principle of eigenvalues greater than 1 and the scree plot to assess the results of the principal component analysis, we identified one factor (eigenvalue = 2.82), which explained 70.38% of the total variance. The factor loadings for the items ranged from 0.79 to 0.86 (see Table 4).Concurrent ValidityWe found a moderate correlation between scores on the PACES-S and the MoMo-PAQ, r = 0.36, t = 4.98, df = 173, p < 0.001; a moderate correlation between the PACES total score and PDPAR (MVPA minutes) results, r = 0.44, t = 6.34, df = 173, p < 0.001; and a moderate correlation between the PACES-S scores and the accelerometer criterion (accelerometer-recorded MVPA), r = 0.32, t = 3.48, df = 109, p < 0.001.Descriptive StatisticsOf the 3219 participants, 1538 (47.8%) were males, and 1681 (52.2%) were females. In terms of age distribution, 1343 (41.7%) were between 11 and 13 years old, and 1876 (58.3%) were between 14 and 17 years old. Different types of schools accounted for the following percentages of participants: Grundschule (1.8%), Hauptschule (3.5%), Realschule (22.2%), Gymnasium (50.7%), Gesamtschule (9.1%), Förderschule (0.7%), and other types of schools or missing data (11.87%). As shown in Table 5, the overall data of 3118 PACES-S data were valid (missing or invalid data: PACES-S, n = 101, accelerometer, n = 1318). Concerning males only, 1493 (97.1%) participants’ PACES-S data were valid (45 missing or invalid data), 885 (57.5%) participants’ accelerometer data were valid (653 missing or invalid data). For females only, 1625 (96.9%) participants’ PACES-S data were valid (56 missing or invalid data), 1016 (60.4%) participants’ accelerometer data were valid (665 missing or invalid data).Internal ConsistencyAs seen in Table 5, for Study 2, the overall Cronbach’s alpha for the PACES-S was 0.87, for male participants, the Cronbach’s alpha for the PACES-S was 0.88, and the Cronbach’s alpha for the PACES-S for female participants was 0.87.Test–Retest ReliabilityThe stability coefficient of the PACES-S for a one-week interval was found to be significant and sufficiently high (r = 0.76, t = 15.14, df = 165, p < 0.01).Construct ValidityWe further used data from Study 2 to test the one-factor model (identified through EFA in Study 1) fit of the PACES-S in AMOS and the overall results indicated a good model fit (χ2 = 53.62, df = 2, p < 0.001; RMSEA = 0.073; CFI = 0.99; RFI = 0.96; NFI = 0.99; TLI = 0.96; IFI = 0.99).Concurrent ValidityWe found a small correlation between scores of PACES-S and the accelerometer-recorded MVPA, r (1840) = 0.21, t = 9.19, p < 0.001; and a small correlation between the PACES-S scores and the accelerometer criterion PA compliance days, r (1840) = 0.20, t = 8.78, p < 0.001.This study aimed to develop a new short, theory-based version of PACES, as there was no reliable version for German adolescents. To this end, first content validity was used to select items that matched the definition of PA enjoyment” PA enjoyment as a positively valenced emotion directed toward the PA associated with feelings such as pleasure, joy, and fun.” Subsequently, psychometric properties of the new measures were assessed (i.e., construct validity, internal consistency, test–retest reliability, concurrent validity). Based on the internal consistency and test–retest reliability, the results indicate the good reliability of the new measure. Moreover, both exploratory and confirmatory factor analyses showed a good construct validity of the measure. Finally, regarding the concurrent validity, the results showed that PACES-S positively correlated with self-reported and device-based measures of physical activity.Previous studies have pointed to the inappropriateness of the unidimensional factor and redundant items in the original 18-item PACES [28,30] and the methodological effect of negatively worded items in the 16-item PACES [30,40]. Thus, Dishman et al. [15] and Raedeke [33] shortened the scales and obtained a seven-item PACES and an eight-item PACES, respectively. However, the psychometric properties were not adequately validated for the 7-items PACES [15,78], and the theoretical conceptualization was missing for the 8-items PACES [33].To solve the issue of inadequate conceptualization, we conceptualized PA enjoyment based on the Component Process Model [24] and adopted the methodology of Davis and Polit and Beck [42] to select items. The analytical results found that only 4 of the 16 items achieved the benchmark value for retention (k* ≥ 0.74), and the S-CVI/Ave for the shortened scale was 0.96, indicating that the PACES-S had excellent item-level and scale-level content validity indices. Although the experts were explicitly asked to consider that some items are negatively worded with a higher number indicating a low level of PA enjoyment, the procedure resulted in only positively worded items. Including only positively worded items showed similarity to Raedeke’s [33] experts’ assessment.The results indicated good reliability with Cronbach’s alpha ranging from 0.82 to 0.88 and test–retest reliability of 0.76. These values were comparable to studies measuring the psychometric properties of other forms of PACES [30,40]. The values were a bit lower than Kendzierski and DeCarlo (α = 0.96) [28]. However, considering that Kendzierski and DeCarlo’s alpha value is greater than 0.9, as pointed out by Tavakol and Dennick [60], this might imply the presence of redundant items in the scale. Compared to the results of Jekauc et al. [36], the internal consistency is similar to the long version of the PACES.The exploratory factor analysis showed that all items were on a single factor. The CFA was then conducted to verify the one-factor solution. Overall, the fit indices indicated that the one-factor model did represent an acceptable fit. Thus, it represented that PACES-S was not suffered from method effects similar to the long version of PACES [30].The PACES-S presented adequate concurrent validities with total MoMo-PAQ (r = 0.36), PDPAR (r = 0.44), accelerometer-recorded MVPA (Study 1: r = 0.32; Study 2: r = 0.21), and accelerometer-recorded PA compliance days (r = 0.20). Taken together, the PACES-S displayed small to moderate significant correlations with both self-reported PA and accelerometer-measured PA. Similarly, Jekauc et al. [40] measured the predictive validity of the original German version of the 16-item PACES and showed that the scale significantly correlated with the MoMo-PAQ, PDPAR, and accelerometer-recorded MVPA results in German adolescents. Besides, the acceptable concurrent validity between PACES (16 items) and self-reported PA was also in line with the result (r = 0.16, p < 0.01) by Moore et al. [79] concerning American children and adolescents. The results of this investigation were also consistent with other studies [80,81] that identified PA enjoyment as an important motivating factor for adolescent participants in PA.Based on the component process model [24], the study provided a theory-based definition of PA enjoyment to develop a new version of PACES. This study utilized a reasonably large sample (Study 2) and a smaller sample (Study 1) to investigate the psychometric properties of the PACES-S. This procedure resulted in a new shortened version of PACES that may be particularly useful to reduce the burden of participants in large-scale studies, where a wide range of variables are measured. However, there were still some limitations. First, we did not measure PA enjoyment by more objective indicators (e.g., face expression). However, it is crucial to consider that the objective measure of discrete emotions is highly debated within the scientific community [82]. Moreover, the current results are based on studies with German-speaking participants. Therefore, future studies should try to replicate the findings in other languages. Besides, the research did not include children under 11 years old. We presume that children could benefit from this short version with graphical illustration. Further research could be refined and implemented among them. Finally, the technical development is a normal process, but we think that it should be mentioned in any case that Study 2 used the newer model of the accelerometer with three-dimensional accelerometer acquisition instead of one dimension in Study 1. On the other hand, Kaminsky and Ozemek [83] compared both models used in this investigation and concluded that the data are comparable with each other, whereby the comparability with our data should remain given as well.In conclusion, the four-item PACES-S offered a short and economical measure of PA enjoyment based on a comprehensive definition derived from the component process model. The investigations of the psychometric properties indicated good reliability and validity of the measure, which were comparable to the reliability and validity of the 16-item version of the PACES. The two studies showed that the method effect underlying the 16-item version of PACES could be eliminated. We hope that the use of PACES-S will contribute to a better understanding of the role of PA enjoyment in PA promotion and maintenance research.The manuscript was conceptualized by C.C., S.W., J.F. and D.J.; analyzed and interpreted by C.C. and D.J.; and written by C.C., A.W. and D.J., acquired funding. Research investigation and data collection were conducted by A.W., C.N., S.C.E.S., C.C., S.W., J.F., D.J. and A.B. revised and edited the manuscript. All authors have read and agreed to the published version of the manuscript.This work was carried out within the Motorik-Modul Longitudinal Study (MoMo) (2009–2022): Physical fitness and physical activity as determinants of health development in children and adolescents. MoMo is funded by the Federal Ministry of Education and Research (Funding Reference Number: 01ER1503) within the research program on long-term studies in public health research. The funding entity is public research funds and therefore had no input in the design of the study and the collection, analysis, and interpretation of the data or the writing of the manuscript. This study was supported by a Chinese Government scholarship—the Chinese Scholarship Council (CSC; No. 201608120055) Scholarship. We also acknowledge support from the KIT-Publication Fund of the Karlsruhe Institute of Technology.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committees of the Charité Universitätsmedizin Berlin (Baseline Study), the University of Konstanz (Wave 1), and the Karlsruhe Institute of Technology (KIT) (Wave 2 and 3, a positive ethics vote was given from on 23 September 2014 by the ethics committee of the KIT).Informed consent was obtained from all participants involved in the study.The datasets generated and analyzed during the current study are not publicly available due to the strict ethical standards required by the Federal Office for the Protection of Data with which study investigators are obliged to comply but are available from the corresponding author on reasonable request.The authors declare no conflict of interest.Characteristics of different versions of PACES and reasons for item deletions.Note: PACES = physical activity enjoyment scale; the blank cells are the items that were eliminated in the scales/studies.Experts’ rating of item relevance, item-level content validity index (I-CVI), and the Kappa designating agreement of relevance (k*) of the 16-item PACES.Note: A blank cell implies that the expert’s rating of the item relevance was 1 (not relevant) or 2 (somewhat relevant). “Y” = the expert’s rating of the item relevance was 3 (quite relevant) or 4 (highly relevant).Descriptive Statistics and Reliability of the PACES in Study 1.Factor loadings from exploratory factor analysis of each item in PACES-S.Descriptive Statistics and Reliability of the PACES in Study 2.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11036.txt

@@ -0,0 +1 @@
+North Korean refugees have not only endured traumatic experiences in North Korea and during defection but have also undergone an adaptation process after arrival in South Korea. Their quality of life (QoL) is likely to be affected by these traumatic life events, leading to subsequent posttraumatic stress disorder (PTSD), or postmigration adaptation-related stress, which involves a sense of dislocation with the culture, language, and people in South Korea. We investigated which aspects predicted the QoL of refugees from North Korea. Fifty-five participants currently living in South Korea completed a checklist about personal characteristics and traumatic experiences before, during and after migration. Diagnosis and symptom severity of PTSD, depressive mood, anxiety, and QoL were also assessed. A multiple regression analysis was performed to evaluate associations between QoL and other variables of interest. Overall, QoL was associated with previous economic status in North Korea, present occupation in South Korea, difficulty interacting with South Koreans, depressed mood, and state–trait anxiety. Finally, QoL was explained by having difficulty interacting with South Koreans, depressed mood, and state anxiety, with the model accounting for 51.3% of the variance. Our findings suggest that QoL among North Korean refugees in South Korea is influenced by the current level of their anxiety and depressed mood, and post-migration adaptation-related stress resulting from trying to integrate with South Koreans after settlement.Since the late 1980s, an increasing number of North Korean refugees have entered South Korea, and this group now comprises about 30,000 individuals [1]. These refugees left North Korea in fear for their lives, enduring considerable hardship [2]. Most studies of North Korean refugees have reported that they felt traumatized as a result of witnessing public executions and family members dying of starvation; additionally, they may have experienced beatings or been subjected to torture in North Korea [3,4,5]. During the process of defection, refugees had to remain hidden and were afraid of detection by the North Korean secret police or by border guards in other countries. They reported that they suffered from lack of food and water and poor treatment until they arrived in South Korea. Moreover, they faced the need to adapt to their new environment after arrival [3,6]. Their accumulated psychological distress, along with their traumatic experiences, led to a high prevalence of psychiatric conditions [7,8].They also suffered psychiatric symptoms. Above-threshold levels of anxiety and depressed mood were reported in 90% and 81%, respectively, of North Korean refugees in China, and 56% of this sample were suspected to have posttraumatic stress disorder (PTSD) [2]. Park et al. summarized the 56 studies in terms of North Korea refugees’ mental health status and identified that 10–48% of them were classified as having depression and anxiety symptoms [9]. These psychiatric symptoms can influence trauma survivors’ social behaviors and functioning. Recurrent and intrusive traumatic memories, for example, may lead to social withdrawal and isolation [10]. Thus, psychiatric symptoms are likely to be potential threats to refugees’ quality of life (QoL) in South Korea.QoL is a broad, multidimensional concept that usually includes subjective evaluations of both positive and negative aspects of life [11,12]. It is influenced by relationships between biological function, symptoms, functional status, and general health perceptions, which are influenced, in turn, by characteristics of both individuals and environments [13,14]. In cases of refugees in other countries, predictive factors for QoL, including traumatic events, psychiatric disturbances, and post-migration variables, have been studied for several decades [15,16]. Many studies have shown that unemployment and social relations predict refugees’ QoL, showing the importance of after-migration variables [4,15,16,17,18,19].The stress related with the language may impact QoL [20]. Even though South and North Koreans speak the same language, different words are used in North and South Korea to convey the same meaning. For example, “lettuce” is “sangchu” in South Korea and “puru” in North Korea. More than 60 years of separation has created a wide linguistic gap [21]. The importance of social relations for QoL were also reported in previous studies of Asian refugees in Canada [20] and Vietnamese refugees in Norway [22].PTSD symptoms, depressed mood, and anxiety have been reported to be associated with lower QoL in traumatized refugee studies [4,18,23,24]. However, the results of such studies vary depending on the type of trauma experienced, a wide range of variables, and unique political or social situations. In the case of North Korean refugees in South Korea, reduced QoL was related to PTSD within a relatively short period after arriving in South Korea (mean length of residence in South Korea = 4.02 years) [25]. However, whether the association between QoL and PTSD persisted at the time of a 3-year follow up study was not investigated [26]. Furthermore, a 7-year follow-up study in North Korean refugees found that PTSD had improved significantly during the intervening period, and current mental health was most significantly related to current culture-related stress [27]. The determinants of QoL among populations with diverse lengths of time since their arrival in South Korea remain under debate. Their unique political or social situations and the following sequent also should be considered as an influence on QoL. Based on several prior studies in refugee populations and North Korean refugees, the present study investigated refugees’ past lives in North Korea, experiences during the defection, present life after arrival in South Korea, psychiatric symptoms including diagnosed PTSD, and the associations of these variables with the level of QoL. We aimed to identify the related factors of QoL among North Korean refugees presently in South Korea. By focusing on refugees from North Korea, we tried to find specific variables to explain their QoL.We recruited North Korean refugees in South Korea from December 2012 to December 2013 through an advertisement on a board within the National Medical Center. The exclusion criteria were: any past history or current illness of psychotic or bipolar disorder, alcohol or substance dependence, organic mental disorder, dementia, severe head trauma, or neurologic illness. Fifty-six North Korean refugees were enrolled in this study. One participant was excluded from data analysis due to failure to complete the clinical measures. Of the 55 participants remaining, most were women (n = 51, 92.7%), and their ages ranged from 24 to 67 years (mean = 46.13, SD = 8.71). Two participants were taking antidepressants, and 15 participants had taken sedatives/hypnotics at a low dosage within 14 days of enrollment. The other participants were free of psychotropic medications.The study was approved by the Institutional Review Board of the National Medical Center, and all participants provided written informed consent. All procedures involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.Using a questionnaire, we asked participants about (1) sociodemographic characteristics while living in North Korea, (2) traumatic experiences during escape from North Korea, and (3) present life in South Korea. The first and second parts of the questionnaire were taken from a previous study [28], and the third part was taken from survey questions from an educational institution belonging to the Ministry of Unification, which educates North Korean refugees living in South Korea. The first part of the questionnaire, which asked about life in North Korea, education level in North Korea (‘less than high school graduate’ or ‘college/university or above’), socioeconomic status in North Korea (‘low’, ‘middle’, or ‘high’), occupation, and history of psychiatric symptoms and treatment. The second part, which dealt with experiences during defection, addressed the length of stay in a third country, arrest experiences, experiences of capture and return to North Korea, imprisonment experiences, whether refugees were accompanied by family or friends from North Korea, and family remaining in North Korea. The third part, which focused on present life in South Korea, addressed current occupation, current physical condition/medical treatment, and adaptation difficulties in South Korea. Items about adaptation-related stress addressed difficulty understanding the South Korean language, interacting with South Koreans, and being excluded by South Koreans.The Clinician-Administered PTSD Scale for DSM-IV (CAPS-DX) was used to diagnose PTSD by psychiatrists based on three elements: re-experiencing, avoidance, and arousal [29]. The frequency and intensity of each event was rated using a five-point scale ranging from never/none (0) to always/severe (4). Traumatic events were defined using two indices according to the DSM-IV. Participants were diagnosed with PTSD when they met three requirements: (1) more than one re-experiencing symptom, three avoidance symptoms, and two arousal symptoms having combined frequency and intensity scores of at least 1–1; (2) a period of more than one month in duration; and (3) presence of clinically significant distress or functional impairment [30]. Among the 55 participants, 32 individuals were diagnosed with PTSD by these criteria. The severity score for each criterion was calculated by summing the frequency and intensity scores, and the total score was calculated by summing the severity scores for all three criteria. The mean CAPS score for all participants was 35.66 (SD = 26.65). We used the Korean version of the CAPS-DX, which showed internal consistency in a previous validation study (Cronbach’s α = 0.95) [31]. The Minnesota Multiphasic Personality Inventory-PTSD (MMPI-PTSD) includes 45 questions drawn from the full MMPI questionnaire. The MMPI-PTSD was developed to identify PTSD symptoms in combat soldiers; higher scores indicate a higher likelihood of meeting the diagnostic criteria for PTSD [32]. The mean MMPI-PTSD score of participants was 25.95 (SD = 8.33). The Korean version was shown to be valid and internally consistent in a previous study (Cronbach’s α = 0.88) [33].The WHO Quality of Life Scale-Abbreviated Version (WHOQOL-BREF), a brief version of the WHOQOL, was administered to quantitatively evaluate QoL [29]. Participants answered two items for the overall QoL and questions addressing four domains, including physical health, psychological domain, social relationships, and environment, using five-point scales ranging from not at all (1) to very much (5). The total score was converted into scores 0–5 after summing the two overall items and scores for the four domains. Higher scores indicated better QoL. The Korean version of the WHOQOL-BREF demonstrated internal consistency in a validity study (Cronbach’s α = 0.90) [34]. The mean of total QoL score was 2.58 (SD = 0.56). The Beck Depression Inventory (BDI) was used to measure the severity of depression, including cognitive, emotional, motivational, and physiological symptoms [35]. In this self-report questionnaire, responses to each question used were provided on a four-point scale ranging from mild (0) to severe (3), with higher scores indicating more severe depression. The mean BDI score of participants was 28.49 (SD = 10.77). The internal consistency of the Korean version of the BDI was high, with Cronbach’s α = 0.85 in a previous study [36]. The State–Trait Anxiety Inventory (STAI) was administered to assess anxiety [37]. This inventory consists of two sets of 20 questions that measure temporary state (STAI-1) and long-lasting trait anxiety (STAI-2), respectively. Responses to questions used a four-point scale ranging from not at all (1) to very much (4), with higher scores indicating higher levels of anxiety (STAI-1, mean = 51.80, SD = 13.96; STAI-2, mean = 51.17, SD = 12.66). The Korean version was validated and internal consistency demonstrated in a previous study (STAI-1, Cronbach’s α = 0.90; STAI-2, Cronbach’s α = 0.92) [38].All measurement responses were screened to check missing values. To interpret clinical measurements with total scores, missing or uncompleted questionnaires were deleted in analysis. Analyses of associations involving WHOQOL-BREF total scores, demographic characteristics and other variables in questionnaires were performed using an independent t-test, one-way ANOVA, and Pearson’s correlation depending on the characteristics of variables. In consideration of multicollinearity, if a correlation between two significant variables in univariate screening was 0.8 or higher, one of the variables was excluded. Since the correlation coefficient between STAI-1 and STAI-2 was 0.86, STAI-2 was excluded from the following analysis because STAI-1 and STAI-2 measured different property of anxiety, and current state anxiety (STAI-1) has been mainly focused on in QoL rather than trait anxiety (STAI-2) in other previous studies [15,39]. Stepwise multiple linear regression analysis was performed to identify the best-associated factors of QoL in refugees from North Korea. In the regression analysis, the dependent variables were WHOQOL-BREF total scores, and the independent variables were selected through previous research findings. They included current occupation [20,22], difficulty in understanding the South Korea language [22,40], difficulty in interacting with South Korean people [16,40,41,42], PTSD-related scales (CAPS and MMPI-PTSD) [43,44,45,46], emotional state (BDI and STAI-1) [42,43,47]. SPSS Version 22.0 (SPSS, Inc., Chicago, IL, USA) was used for statistical analyses.As shown in Table 1, refugees with a higher economic status in North Korea had higher total QoL scores at the time of evaluation. In terms of current life in South Korea, participants working in South Korea and those reporting less difficulty interacting with South Koreans had higher total QoL scores than did others. There were no significant differences in QoL total scores with regard to sex, educational level, escape-related experiences, other variables addressing past life in North Korea or current condition in South Korea, or PTSD diagnosis. The total scores of clinical parameters are described in Table 2. We found negative associations between QoL total scores and BDI, STAI-1, and STAI-2 scores. The MMPI-PTSD score showed a marginal statistical significance. However, no significant correlation of QoL total scores with CAPS was observed. A linear regression analysis was performed to identify the factors that have an association with QoL (Table 3). Difficulty in interacting with South Koreans, depressed mood, and state anxiety were the related factors of QoL. These variables accounted for 53.1% of the total variance. Other variables, including current occupation, difficulty in understanding South Korean language, CAPS, and MMPI-PTSD did not significantly predict QoL scores.The level of QoL among refugees from North Korea was more closely associated with their current status than with past traumatic experiences or diagnosis of PTSD. The factors were difficulty interacting with South Koreans, and the level of depressed mood and state anxiety.The related factor for QoL among North Korean refugees was difficulty interacting with South Koreans. This is consistent with previous studies showing the importance of social relations for QoL [20,23]. A study of tortured refugees also showed that social relations were a predictive factor associated with QoL [19]. North Korean refugees are often treated as strangers in South Korea. The 60-year history of separation between North and South Korea has created many differences between the countries [3]. These include linguistic, cultural, political, and economic differences that interfere with the ability of North Korean refugees to understand and socialize with South Koreans. The refugees’ use of the Northern dialect, which differs considerably from the South Korean language [48], and the low status of North Korea can lead to their marginalization by South Koreans, difficulties with assimilation into South Korean society, psychological stress, and PTSD [3]. Unlike other refugees, North Korean refugees acquire the same citizenship as South Koreans according to the law. However, the stress experienced by North Korean in South Korea resembles that experienced by refugees to other countries [16]. There is a possibility that a hierarchy of citizenship exists [49]. Lazarus showed a possible deviation between citizenship in principle and in practice within the constructs of race, regarding the debate that immigrant citizens were largely perceived as second-class citizens in Austria [50]. Although both North and South Koreans belong to a single racial group, many North Korean refugees were confused about their identity as a citizen of South Korea [3,48]. Linguistic, cultural, political, and economic differences may prevent North Koreans from accepting them as genuine members of the community. As a result, some of them have difficulties in settling down in South Korea. North Koreans’ problematic post-migration experiences with South Koreans may lead to an identity crisis or sense of confusion, followed by adaptation-related stress and low QoL [51].PTSD, however, was not significantly associated with QoL in North Korean refugees in our results. Meanwhile, previous studies reported that health-related QoL was negatively associated with PTSD symptoms in child survivors of an earthquake, and lower QoL was predicted by the presence of PTSD symptoms in adults following physical assault [40,41]. It should be noted that these previous studies were conducted less than 15 months after the index traumatic events, and a prior finding suggests that having experienced a shorter period since immigration was associated with worse mental health status [44]. Most of our participants have been in South Korea for more than 3 years. In a refugee population with a mean length of post-migration duration of 16.7 years, post-traumatic growth explained more of the variance in QoL than did post-traumatic stress symptoms [18].By contrast, in the present study, the level of depressed mood and anxiety was associated with QoL. This finding was consistent with previous studies which showed psychiatric symptoms’ impact on life satisfaction or association with poor sociocultural adaptation [52,53]. Along with difficulty interacting with South Koreans, these emotions contribute to post-migration difficulties [43]. A previous study reported that severe post-migration difficulties significantly increased the risk of PTSD, independently of pre-migratory traumatic events [44]. In immigrant survivors of political violence, post-immigration experiences such as financial and legal insecurity significantly explained more variance in PTSD outcomes than did pre-migration variables alone [45]. It seemed that post-migration conditions had a greater impact on psychological state than PTSD itself. Considering that PTSD symptoms naturally decrease over time [27], it is considered that if these post-migration difficulties are not resolved after migration, this may result in more severe PTSD symptoms and a decrease in QoL.Although economic status and current working status were not significant in multi regression, they were significant in univariate analysis. Regarding economic status, a survey by the Korea Hana Foundation indicated that 92.5% of North Korean refugees were middle (58.3%) or lower (34.2%) economic class when they were in North Korea, and 97.6% of North Korean refugees were in either middle or lower class after migration into South Korea [46]. Furthermore, they needed to undergo job training in order to find a job in South Korea because most of refugees were unskilled laborers back in North Korea [46]. Many refugees from North Korea did not have jobs or earned much lower incomes than average in South Korea [46]. In previous findings, unemployment was an important factor in predicting mental health and QoL in refugees [16,19,47]. In the case of North Korean refugees, however, it seems that socioeconomic status and the types of occupation in North Korea not only affect their current occupation but also have a greater correlation with QoL in comparison to those of current working status and QoL.This study has several limitations. First, our dataset had limited properties. Because only about 30,000 North Korean residents live in South Korea, and they tended not to participate in research, being reluctant to reveal their origins, so there was recruitment restriction. Considering the sample size was calculated at 67 (effect size 0.35, power 0.95, number of predictors 6, alpha error probability 0.05), the sample size was relatively small, and participants were recruited from a single hospital and were skewed toward females. However, given that 71% of North Korean refugees currently consist of females [1] and PTSD is more prevalent among females than among males across the lifespan, the gender ratio of this study seems to be ecologically plausible in both epidemiological and clinical aspects. However, as it was a convenience sample with a population self-referring to the hospital, further studies are needed on larger populations and subgroup without mental and/or physical symptoms. Second, this study employed a cross-sectional perspective, and participants varied widely in terms of the duration of their residence in South Korea. To extend the present findings, further studies are needed of refugees who share a specific period of residence or who undergo longitudinal follow-up examinations.This research highlighted that state anxiety and difficulty interacting with South Koreans predicted the overall QoL of North Korean refugees. These results showed that post-migration difficulties are important contributors to QoL more than their past experience in North Korea, their traumatic experiences during the defection, and physical and medical status in South Korea. Additionally, although 58% of the participants suffered from PTSD symptoms, their QoL was not associated with the total score of CAPS but associated with their level of anxiety and depressed mood. This may be affected by the time the refugees spent in South Korea. These present findings suggest that more psychosocial attention to their psychiatric symptoms and social interaction is needed according to duration of living of in the new country in order to improve the QoL of North Korean refugees.Conceptualization, S.Y.Y.; methodology, S.Y.Y.; software, J.E.S.; validation, J.E.S., S.Y.Y. and S.-H.C.; formal analysis, J.E.S.; investigation, S.Y.Y.; resources, S.Y.Y.; data curation, J.E.S. and S.Y.Y.; writing—original draft preparation, J.E.S.; writing—review and editing, S.Y.Y. and S.-H.C.; visualization, J.E.S.; supervision, S.-H.C., J.-S.C. and S.Y.Y.; project administration, S.Y.Y.; funding acquisition, S.Y.Y. All authors have read and agreed to the published version of the manuscript.This research was funded by the National Medical Center, Research Institute, grant number NMC2012-MS-03.The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of National Medical Center (H-1208/022-008 and 31 January 2013).Informed consent was obtained from all subjects involved in the study.The data presented in this study are available on request from the corresponding author. The data are not publicly available due to participant confidentiality and privacy.We are grateful to our study participants.The authors declare no conflict of interest.Association of quality of life with demographics, past life history in North Korea, trauma experiences, current life condition in South Korea, and PTSD diagnosis.Independent t-test, one-way ANOVA, and Pearson’s correlation were used depending on each variable characteristic. Abbreviations: PTSD = posttraumatic stress disorder. a n = 54; b n = 48; c n = 53; d n = 47; e n = 51.Association of quality-of-life scores with clinical characteristics.Abbreviations: CI = confidence interval; CAPS = Clinician-Administered PTSD (posttraumatic stress disorder) Scale; MMPI-PTSD = Minnesota Multiphasic Personality Inventory-PTSD; BDI = Beck Depression Inventory; STAI = State-Trait Anxiety Inventory.Regression of variables to predict quality of life in refugees from North Korea (adjusted R2 = 0.603).Abbreviations: SE = standard error; CI = confidence interval; VIF = variance inflation factor; MMPI-PTSD = Minnesota Multiphasic Personality Inventory-PTSD; BDI = Beck Depression Inventory; STAI = State-Trait Anxiety Inventory.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11037.txt

@@ -0,0 +1,7 @@
+During the COVID-19 pandemic, there has been a lot of discussion about keeping interpersonal distance to prevent the virus from spreading. To keep this interpersonal distance, authorities at different levels have taken measures to reduce people’s interactions, such as reducing capacities, curfews, pop-up cycle lanes, temporary pedestrianisation, and lockdowns. Many of these temporary measures have been perceived from a static view. Nevertheless, in a scenario of “new normality” or in the face of a possible new pandemic, the amount of data (big data) generated by different sources, such as sensors, in large cities has extraordinary potential to be used together with tactical urbanism for quick adaptation. The aim of this study was to gain insight into the aforementioned issues by analysing spatio-temporal patterns of pedestrian mobility and developing a variation of the pedestrian level of service measure; the pandemic pedestrian level of service (P-PLOS). This measure provides a dynamic view of pavement capacities according to the interpersonal distance recommendations during the pandemic. P-PLOS was tested in the city of Madrid based on the pedestrian counter data that was provided by the local government through its open data website. We found that the application of P-PLOS, together with street design, allows for knowing where and when it is necessary to take tactical urbanism measures in order to maintain or improve the level of service, as well as where it is necessary to take measures to reduce pedestrian flow.With the arrival of COVID-19, the scientific community and public bodies began to ask people to practice social distancing as one of the main measures to prevent contagion [1]. These measures to reduce virus transmission have led to a reduction in mobility, to a greater or lesser extent, at the international level [2,3,4,5], as well as a change in daily mobility patterns [6]. However, the impact on mobility has not been the same for different modes of transport. In the case of bicycles, it has led to an increase in the use of bicycles as a mode of transport due to the distance it allows and the implementation of numerous temporary bicycle lanes by the public bodies [7]. Nevertheless, the use of private vehicles has increased, to the detriment of public transport [8].With the recommendations for social distancing and restrictions on mobility, it gradually became clear that it was difficult to maintain the distance in some situations and environments, in particular, when pavements are narrower than the minimum recommended distance between two or more people. In fact, the early months of the pandemic saw the appearance of a slew of websites (e.g., https://www.sidewalkwidths.nyc (accessed on 5 April 2021)), blog entries, and applications dedicated to the analysis of the width of pavements in relation to the recommended social distance.However, the lack of tools for monitoring and controlling pedestrian mobility has become evident. In this respect, it should be recalled that the scholarly research on the safety of human crowds has mainly been developed within the last few years [9]. This is due, among other factors, to technological advances related to the detection of moving objects in general and moving people in particular [10]. Within the context of COVID-19, this lack of tools is, in turn, due to the lack of data acquisition devices and uncertainty about the spread of the virus. Concerning data acquisition, the existing literature shows that pedestrian dynamics are mainly analysed through empirical methods and visual crowd analysis, including simulations [9]. The use of real data, e.g., from pedestrian counters, appears to be much lower, as this technology is quite recent and not many cities utilise them. Regarding the spread of the virus, this depends on different variables, including droplet size [11], air temperature, air humidity, breathing strength, and the use of a face mask. Greater dispersion occurs when people sneeze, cough, and when they speak or breath harder while walking or running [12]. This greater dispersion distance implies that social distancing should be greater when people walk along the pavements. At this point, there is a gap in the literature regarding virus transmission control within the context of pedestrian mobility.The goal of the present study was to evaluate the potential of pedestrian counters to monitor and control social distancing through a simple tool based on the classic measure of pedestrian level of service (PLOS). Pandemic pedestrian level of service (P-PLOS) has been designed to take more realistic measurements of pedestrian mobility in situations where specific recommendations on interpersonal or social distancing must be complied with. This tool could be of great use for decision making regarding tactical urbanism measures. This study, developed for the central district of the city of Madrid, had three main objectives, as follows:Evaluate the effects of social distancing on spatio-temporal patterns of pedestrian mobility using pedestrian counters.Propose a pedestrian level of service for pandemics that enables the use of tactical urban planning measures.Evaluate the pedestrian level of service with and without social distancing.Evaluate the effects of social distancing on spatio-temporal patterns of pedestrian mobility using pedestrian counters.Propose a pedestrian level of service for pandemics that enables the use of tactical urban planning measures.Evaluate the pedestrian level of service with and without social distancing.The main contributions made by this study are: (1) rethinking and adapting PLOS to a situation with social distancing; (2) providing an example (in a contemporary context) of the use of pedestrian counter data; and (3) showing the results for the city centre of Madrid, where pedestrian counters have only recently been installed.This article is structured as follows: Section 2 contains the literature review. Section 3 presents the data and methodology used to assess the spatio-temporal effects of the pandemic on pedestrian mobility. Next, in Section 4, we compare the spatio-temporal distribution of pedestrians in a year prior to the pandemic to the distribution in a year during the pandemic, and analyse the pandemic pedestrian level of service in the study area. Finally, in Section 5, we discuss the results obtained and the advantages of having pedestrian counters to allow for decisions to be made in real time based on pandemic pedestrian levels-of-service (P-PLOS).Levels-of-service (LOS) are useful tools for assessing the capacity of transport infrastructure [13] or a station to accommodate pedestrians or bicycles. In the first analyses of LOS, some authors used pedestrian density to evaluate capacity and space requirements in roadways [14]. Later, other researchers then applied different indicators to assess the level of service, such as the combination of the pavement section and pedestrian density [15]. It is clear that there has been a methodological enrichment in the calculation of LOS. The great advantage of LOS as a tool is that the representation of the results takes place on a scale (from A to F) that is easy to understand. As a result, it has become common in planning and decision making, as reflected in its frequent appearance in reference manuals [16].However, in a situation of normality without restrictions on mobility, pavements are infrastructures on which people are in movement. They may be characterised by the needs of the pedestrians with respect to the structure, safety, comfort, and attractiveness of the street [17]. These characteristics of the street create incentives or disincentives for pedestrian movement on them and encourage or limit interactions with existing shops or other pedestrians. This implies that the choice of itinerary by pedestrians from a point of origin to a specific destination may be different according to the pedestrian quality of the environment [18]. Taking into account these circumstances, it is essential to know how pedestrians are distributed in space and time for the purpose of making decisions, particularly when there is a high pedestrian flow rate, which may have a negative impact on social distancing. Data from pedestrian counters, among other sources (see Section 2.2.), can play a key role. In addition, in recent years, numerous attempts have been made to simulate pedestrian mobility [19,20], and specific simulation tools have been developed on the basis of the PLOS in the context of social distancing [21]. These simulation tools could be enriched or validated by real data from pedestrian counters.In recent years, there has been a revolution in the use of geolocation data [22] for different forms of transport, providing us with a better understanding of movements in cities [23]. In addition, these data from geolocated devices and sensors provide a valuable source of information for pandemic monitoring and control [24], allowing for the identification of spatial transmission [25].The information generated is very heterogeneous, depending on the type of geolocated device. In the case of fixed devices, the information generated does not allow for the reconstruction of trajectories or to have a user profile. However, when the device is individual (such as a mobile phone, shared bicycle, transport card, or credit card), the information generated allows for the reconstruction of spatio-temporal trajectories of the person. This type of device also allows for better user characterisation, as it is usually accompanied by socio-demographic data (age, gender, etc.). This means that mobile phone data, for example, can assist in the modelling of the geographical spread of epidemics [26,27].Beyond mobile phone data, the construction of a big data information system for epidemics has been limited by multiple challenges, such as data acquisition, data integration, or multi-scale dynamics, among others [25]. These challenges are particularly notable with regard to pedestrian mobility. The data acquisition for pedestrian mobility continues to be limited and restricted, both by technology and location. In response to this situation, collaborative initiatives, such as WeCount (https://we-count.net/ (accessed on 5 April 2021)), are beginning to emerge and are of great interest. This collaborative project can carry out mobility and air quality measures in a host of streets throughout Europe.Despite the limited data, there are some examples that assist in the acquisition of big data, such as the data collected by Apple [28], although its level of data disaggregation does not allow for the analysis of changes at a detailed level (neighbourhood, census district, or street). Using mobile phones as a source, Hunter et al. [29] analysed the change in the behaviour of the population of the United States with respect to walking, assessing the variations in time and distance before and after the pandemic. However, the use of this source of data has its limitations. First, the source used provided a great volume of data, which have to be filtered to obtain data that correspond to pedestrian mobility according to the criteria. Second, the data from mobile phones obviously do not include information on people who walk without their mobile phone [29] or who carry it without being connected. One method of overcoming these problems is to use data generated by fixed sensors, such as pedestrian counters. These counters provide data on the number of people who pass by a particular point. The disadvantage, in this case, is twofold. Firstly, a large number of such counters have to be deployed to obtain data from the whole city, something that rarely happens. Secondly, the counters do not measure the socio-demographic attributes of the people passing. Nevertheless, the lack of socio-demographic information is not a determinant in our application, since P-PLOS focuses on the flow of people passing through a given area to decide whether, at a given moment, this flow may entail risk.Social distancing is a response to the significant capacity of the virus to be dispersed through the air. There are many articles and pre-prints providing information on the dispersal of COVID-19, which is closely related to particle size and the force of exhalation [30,31] when talking, coughing, or sneezing. Based on all the scientific information generated, including not only that related to COVID-19 but for other infectious diseases transmitted by air as well, public health organisations at both global and national levels formulated the recommendation to maintain a social distance of between 1 and 2 m [32]. The World Health Organisation (WHO) [33] recommends a distance of at least 1 m. The Centre for Disease Control and Prevention (CDC) [34] proposes 6 feet (about 1.8 m), and the Spanish Ministry for Health recommends a distance of 2 m and the use of masks when a separation of 1.5 m cannot be maintained [35].Some studies were also carried out, such as the one by Córdoba-Hernández et al. [36], who calculated the relationship between the area of the pavement and the population, demonstrating the zones that are deficient with respect to the area of pavements. However, although this information could be of interest, it merely demonstrates the deficit in pedestrian infrastructures with respect to the population and social distancing, in which people are considered non-mobile elements who occupy space on pavements in a static situation. A static scenario like this may be of some use to highlight the need for more pedestrian space to prevent virus transmission, but only in situations where movements are restricted by space and time [36], as implemented in Spain from 1 May 2020 [37].Therefore, it is essential to consider a dynamic scenario of pedestrian mobility, as it is only in these kinds of scenarios that variables that influence the dispersion of the virus and that derive from a subject in motion with more intense exhalations can be included. For the scenario of pedestrian mobility, Blocken et al. [38] proposed a distance of 5 m when walking, 10 m when running, and 20 m in the case of bicycles.To assess the effect of the pandemic on pedestrian mobility, we used the city of Madrid as a case study, more specifically the Centro district (Figure 1). In particular, the Centro district is characterised by its strong cultural component, which makes it a district with a large presence of tourists from all countries, whose digital footprints have previously been measured by García-Palomares et al. [39] and Salas-Olmedo et al. [40]. This distinctive feature results in a predominance of commercial premises (36.8%) and hotels and restaurants (25.5%) according to Madrid’s retail census obtained from the open data website [41].From the point of view of mobility, in 2018, the Centro district joined the low-emission zone, Madrid Central, which promotes sustainable transport, such as walking and cycling, and where public transport is favoured over private transport.In short, it is a district with a high flow of pedestrians compared to other areas, which, in turn, makes it an area where crowding is more likely to cause risky situations. As Madrid is a big city, the amount of data generated at the private and public levels is greater than in other cities, which makes it easier to test the proposed tool. The City Council of Madrid generates a multitude of data relating to mobility (traffic, bicycles, pedestrians, etc.), which it makes available to the public through its open data website (https://datos.madrid.es/portal/site/egob (accessed on 5 April 2021)).The main source of data used to assess the effect of the pandemic on pedestrian mobility has been the pedestrian counters installed in the city of Madrid. To compare trends, we used other mobility data during COVID-19 provided by Apple and Google.The characteristics of these sources of data are specified below:(a)The data provided by Apple [28] refer to the number of requests for directions by country, region, or city, and were compared to the reference data of 13 January 2020 used in this study. These data from Apple users were included anonymously, which implies that the associated identifiers are random and rotating. Moreover, the data do not include any demographic information about the users, so no relations can be established with specific population groups.(b)Google provided Local Mobility Reports on COVID-19 [42], which showed the trend in movement over time at different scales, classified by categories of places (shops and leisure, supermarkets and pharmacies, public transport stations, workplaces, and residential areas).(c)The database of pedestrian counters is available to the public on the Madrid Council’s open data (Datos Abiertos) website [41]. This data source is maintained by the local administration, which periodically publishes data with a high level of disaggregation and in an exhaustive manner, which results in high-quality data. The counter data have been available since 2019, but they are not homogenous, and these limitations must be taken into account if the information is to be correctly processed. A total of 19 counters are available and distributed around the Centro district, some in main streets and others in smaller, more outlying ones. The durations for which the numbers of pedestrians were recorded was 15 min in 2019 and 60 min in 2020 and 2021. This change in frequency must be taken into account and we return to it in Section 3.3.(d)Finally, we used the street plan of the pavements of the city of Madrid as the base map for analysis. It is freely available through the Geoportal of the Madrid Council [43] at a scale of 1:1000, with the latest update in 2016. This information has been updated manually on the pavements that have been extended after the date of publication of the plan.The data provided by Apple [28] refer to the number of requests for directions by country, region, or city, and were compared to the reference data of 13 January 2020 used in this study. These data from Apple users were included anonymously, which implies that the associated identifiers are random and rotating. Moreover, the data do not include any demographic information about the users, so no relations can be established with specific population groups.Google provided Local Mobility Reports on COVID-19 [42], which showed the trend in movement over time at different scales, classified by categories of places (shops and leisure, supermarkets and pharmacies, public transport stations, workplaces, and residential areas).The database of pedestrian counters is available to the public on the Madrid Council’s open data (Datos Abiertos) website [41]. This data source is maintained by the local administration, which periodically publishes data with a high level of disaggregation and in an exhaustive manner, which results in high-quality data. The counter data have been available since 2019, but they are not homogenous, and these limitations must be taken into account if the information is to be correctly processed. A total of 19 counters are available and distributed around the Centro district, some in main streets and others in smaller, more outlying ones. The durations for which the numbers of pedestrians were recorded was 15 min in 2019 and 60 min in 2020 and 2021. This change in frequency must be taken into account and we return to it in Section 3.3.Finally, we used the street plan of the pavements of the city of Madrid as the base map for analysis. It is freely available through the Geoportal of the Madrid Council [43] at a scale of 1:1000, with the latest update in 2016. This information has been updated manually on the pavements that have been extended after the date of publication of the plan.To apply a new pedestrian level of service for a pandemic that can be used together with tactical urbanism to improve pedestrian flows and reduce health risks, we propose a specific methodology (Figure 2) that has been structured into 4 main steps.The first step is fairly technical and includes data preprocessing and street characterisation.The second step entails an overall vision of the effects of the restrictive measures related to COVID-19 on pedestrian mobility.The third step proposes a pedestrian level of service, which allows for an assessment to be made of the spatio-temporal patterns in which pedestrian flows are high and, therefore, involve a risk of contagion.In the final step, the pedestrian level of service is analysed as a whole, together with the layout of the street, so that recommendations for action can be prepared.We have written a script in the Python programming language that allows for the downloading of information in text format (csv) from the websites of Google, Apple, and Open Data of Madrid, checking for errors and missing data.From January to June 2019, the pedestrian counters in Madrid were at a testing phase, so the data included may be overestimated or underestimated. In addition, the location of the counters in this first six-month period was not final, making it difficult, in some cases, to compare the later data. Finally, the frequency of counting also varied. Although the Council’s website suggests that the counting was carried out every 15 min, this frequency was only used in the testing phase; after that, the frequency was every 60 min. Moreover, data were unavailable in July and August 2019 due to a malfunction in the sensors. For all the above reasons, we eliminated all the data before September 2019. This article only covers the period from September 2019 to April 2021.The text files in csv format from Apple and Google were filtered by the Python Pandas library to obtain the information for Madrid, and the format was homogenised to allow for comparison.The area of the pavements was calculated according to the information provided by the Madrid Council, with the data being updated for the pavements whose width was extended after the publication date. After updating the geometry of the streets, their width at the point where the counter was located was obtained.Table 1 shows a brief description of the counter number, the street, the street number at which the sensor is located, and the pavement width. Further details on the characteristics of the section with respect to the modal distribution are included in Appendix A. The counters capture the information on the pavement where they are located, which are distinguished with respect to the number of doorways to homes and the side of the street (odd or even street numbers). In the case of pedestrianised streets, the sensor refers to the whole section.Two periods were used to analyse the changes in the temporal patterns of pedestrian mobility.First, the temporal patterns were analysed at a general level, comparing the data from Google, Apple, and the data on pedestrian counters. For this initial analysis, the data offered by Apple and Google were used, covering the period between January 2020 and April 2021. This analysis only allows for daily change details due to the level of detail of the Google and Apple data.The level of detail offered by the data from the counters was examined and a comparative analysis was carried out on pedestrian mobility in the third four-month period (September to December) of the years 2019 and 2020. These periods of time provide a broader vision with which to compare a normal period and a period of “new normality”, in which there were special restrictions on mobility in specific zones and times. In addition, we analysed the patterns of pedestrian mobility in terms of space by a disaggregated use of the counters. This disaggregated use of the data by the counters allowed us to determine the impact of mobility related to the characteristics of the streets in which the counters are locatedOnce the spatio-temporal patterns of pedestrian mobility during the pandemic were analysed, the pedestrian level of service tool was redesigned to adapt it to a pandemic context, taking into account the recommendations of the health authorities and scientific studies published to date.To calculate the pandemic pedestrian level of service, the variables of the walking scenario had to be established based on the level of social distancing when the population is stationary (Figure 3), which were set at a lateral distance of 1.5 m, according to the recommended interpersonal distance. In movement, the dispersion is mainly produced along the displacement axis. The work by Blocken et al. [38] on the safety distance for aerosol transmission as related to the speed of walking was used as a basis for this work. We established the same walking speed of 4 km/h, and a safety distance of a minimum of 5 m, as a reference. The distance used was the distance at which the minimum presence of particles is detected as expelled when breathing by a person walking at a speed of 4 km/h [38].Using the above data, we calculated the walking-dispersion area, which was used as a reference for the level of service:(1)Ad=Ddfv∗LDd
+          where:
+		  Ad is the walking-dispersion area;Ddf(v) is the walking-dispersion distance based on speed;LDd is the walking-dispersion lateral distance.
+		  Ad is the walking-dispersion area;Ddf(v) is the walking-dispersion distance based on speed;LDd is the walking-dispersion lateral distance.Following this formula, the walking dispersion area at 4 km/h is 7.5 m2 (5 m distance * 1.5 m of lateral distance).Then, the values corresponding to the pedestrian service levels were recalculated (Table 2) based on the original pedestrian service level tool in the Highway Capacity Manual 2000 [16] and the following formula:(2)Vp=SprAp
+          where:
+		  Vp is the flow rate per unit of width (pedestrian/min/m);Spr is the reference pedestrian walking speed (m/min);Ap is the pedestrian space (m2/p).
+		  Vp is the flow rate per unit of width (pedestrian/min/m);Spr is the reference pedestrian walking speed (m/min);Ap is the pedestrian space (m2/p).The width of the pavement for the purpose of measuring pedestrian flow takes into account only the effective width of the actual pedestrian pavement. There are many features along the pavement that may be an obstacle to people and that reduce the pavement width. These obstacles can be fixed, such as signals, streetlamps, some types of tree surrounds, benches, and other types of urban furniture. Other features include temporary obstacles, such as restaurant or shop banners near facades, people exiting from building entries, window shoppers, parked scooters, and so forth. Whereas fixed obstacles are easy to measure, measuring the pavement width considering temporary obstacles may be more difficult and may involve a high cost in terms of personnel and time.Given all of the above, and following the research undertaken by Córdoba-Hernández et al. [36], we consider that the effective width represents 55% of the total pavement.The effect of the COVID-19 pandemic on pedestrian mobility in Madrid (Figure 4) shows how the confinement measures of the first wave of the pandemic (March 2020) led to a swift fall (94% from the reference day) in the number of pedestrians recorded by the counters. This limited number of pedestrians was maintained until May, when the measures were eased, allowing people to walk in the open air. As the conditions were progressively eased, the number of pedestrians recorded increased until the summer months (July and August), when people left Madrid on holiday and there were no arrivals of foreign visitors. With the return of people from their holidays and the start of the school year, there was a slight increase in the number of pedestrians, which was maintained, except for a few weeks, until the end of 2020. The data recorded for the first quarter of 2021 were stable with respect to the number of pedestrians recordedWhen comparing the data recorded by the counters to the data provided by Apple for the analysis of mobility during the pandemic, it can be seen how the temporal pattern is similar, in general (Figure 4). However, there are certain differences in some periods of time due to the scales of the samplings. In the case of Apple, the data cover the whole city of Madrid, while the counters are located solely in the Centro district, so they may record a lower level of pedestrian mobility.The comparison between the third quarters of 2019 and 2020 shows very similar patterns with respect to the distribution of pedestrians on the days of the week (Figure 5a), where there were peaks in the numbers of pedestrians on Saturdays. With respect to the distribution of pedestrians at different times of the day (Figure 5b), and taking into account the significant reduction in the number of pedestrians, the temporal distribution remained the same, with a peak between 13.00 and 14.00 h, and a second peak with a larger number of pedestrians at 20.00 h.Once the results showing the overall number of pedestrians were determined, it was interesting to investigate the temporal patterns by counter to detect differences in the spatio-temporal patterns. Figure 6 shows how the measures and recommendations on mobility have had a different impact depending on the characteristics of the street. The reduction in the number of pedestrians recorded by each of the counters fell considerably, with similar values being recorded in the periods under analysis (except for the counter PEA15). It is worth noting that the counter PEA02, despite showing the biggest fall in the average number of pedestrians, registered values in 2020 at above 1000 pedestrians/hour.Looking in more detail at the temporal distribution by pedestrian counter (Figure 7), clear differences can be seen in the numbers of pedestrians passing by the different counters. These differences are clear for 2019 and even more so for 2020. In both years, three particular counters registered the greatest number of pedestrians: PEA02-PM01, followed by PEA08-PM01 and PEA08-PM02.The results of applying service levels with the data collected (Figure 8) show how for a scenario of normality, the average values of the pedestrian level of service (PLOS) for the period under analysis were at optimal levels (Level A) only for counters PEA02-PM01 and PEA08-PM01 at certain times of the day. For its part, PEA08-PM01 registered a fall in the level of service between 17:00 h and 21:00 h.Regarding the pedestrian levels-of-service during the pandemic (P-PLOS) (Figure 8b), it is clear that the recommended social distancing measures have had an obvious effect on the levels of service. In general, there is a great homogeneity in the levels of service in 2020, for which the peaks of mobility were at the usual times (from 13:00 to 15:00 h and from 18:00 to 21:00 h). This homogeneity was not reflected in the counter PEA02-PM01, which registered the larger number of pedestrians and gave rise to low levels-of-service (D, E) where interpersonal and social distances were recommended, or the speeds were reduced.Finally, it is worth analysing the results of the normal scenario from the pandemic level of service perspective (Figure 8c). This scenario shows how the counters PEA02-PM01 and PEA08-PM01, located in the streets Fuencarral and Gran Vía, respectively, were at level of service F, which is associated with pedestrian flows in which it is difficult to maintain a safe distance.COVID-19 has had a major impact on mobility, both through a change in mobility patterns [6] and changes in public attitudes [44]. During this time, several adaptation actions have emerged, such as the implementation of pop-up bike lanes, among other measures [7]. In addition, this pandemic has put transport systems to the test, showing their weaknesses and strengths, and allowing new challenges to be faced for more sustainable mobility [45].Pedestrian mobility has probably been impacted the most, and it is clear that the pedestrian levels of service that have been so widely used in pedestrian mobility analysis cannot be directly applied in situations with social distancing. At this point, there are two main challenges that need to be met: the need to consider a dynamic scenario (pedestrians on the move) that takes into account the spread of the virus; and the need for pedestrian count data, preferably real and in real or quasi-real time. This would allow decisions to be made once decreases in the pedestrian levels of service are detected.In this context, we have proposed the pandemic level of service (P-PLOS), whereby an adjustment of levels of service is made considering situations with social distancing, and which is fed by pedestrian flow data taken from pedestrian counters. P-PLOS has been tested for the case of the central district of Madrid. The city of Madrid has recently started publishing data from these pedestrian counters (September 2019), but no previous works have used this data yet. The results obtained show that the measures restricting mobility have been useful for avoiding agglomerations (and would have, at least theoretically, produced a lower propagation of the virus outdoors). Once the restrictive measures were eased, pedestrian mobility increased. However, this recovery in pedestrian mobility was not the same in all the streets with counters. In streets with many commercial uses, the recovery was greater due to their attractiveness for pedestrians [46], while in streets without commercial uses, the presence of pedestrians has remained low.At the same time, the measures taken in Madrid to restrict mobility at night had an effect on the presence of pedestrians in the early hours, but it did not represent a modification in the temporal pattern (Figure 7).However, beyond the minor presence of pedestrians in the streets under analysis, it is necessary to analyse this presence in terms of the pedestrian level of service, taking into account the recommended safe distance.From the perspective of the pedestrian level of service during the pandemic, it can be seen that the streets with significant retail activity, such as Fuencarral and Gran Vía, have low (E) or very low (F) levels of service, mostly in the afternoon. These low levels of service also occurred on the dates closest to Christmas, which is traditionally known as the Christmas shopping period. According to the results obtained, at these peak hours, with high numbers of pedestrians, distances and speed were affected, increasing the risk of not maintaining the safe distance.It is in these situations in which the level of service declines that tactical urbanism can play a decisive role. Tactical urbanism allows for action to be taken according to the characteristics of the street to allow for a greater proportion of the pavement to be used by pedestrians [47,48]. Thus, in cases in which the street has a parking lane, this lane may be adapted for pedestrian use to ensure the maintenance of a good pedestrian level of service. In streets where there is no parking lane, but there are a number of lanes of traffic in the same direction (as in the case of Gran Vía), one of them could be used for pedestrians to move in the hours with the greatest expected presence. Finally, in cases in which the street is pedestrianised (such as Fuencarral), actions should be taken to limit the presence of pedestrians to reduce the risk of contagion. These deterrent measures may take various forms, ranging from restricting access to the street or redirecting these pedestrians to other streets. Of course, some of these actions have been put in place by local governments at certain times during the pandemic, but P-PLOS can help them to make better more informed decisions considering local situations and needs.This research also makes clear the usefulness of introducing sensors in cities, as they allow for monitoring, in real or quasi-real time, of a number of aspects of the city, from air quality to mobility, as in the case we investigated. These sensors, which generate a large quantity of data, may be very useful in decision making and the management of mobility in a variety of circumstances. This tool allows for active management of pedestrian mobility, maintaining the distances recommended by the health authorities, and providing appropriate information for tactical measures in certain streets and at specific times during the week or in the day.Among the main limitations of this study are the levels of disaggregation of the data from the different sources. In the case of Apple and Google, there was a limitation of scale that prevented us from knowing the changes in mobility by district or census zone, as well as the fact that the interval between data collections was a day. With respect to the pedestrian counters, there is a limitation in the sample interval, which was increased from 15 min (recommended level) to intervals of 1 h. It should also be noted that the sensors are located in the Centro district. Information on other districts in the city is, therefore, not available for appropriate action to be taken. However, it would be possible to introduce data from other sensors, such as temporary or specific pedestrian time sensors, including the WeCount initiative. In any case, it is precisely the central district where the highest pedestrian flows in the city of Madrid occur [49], making it a strategic area for monitoring possible risk situations. The absence of socio-demographic information on pedestrians could also be seen as a limitation. Of course, this information would be useful for a better characterisation of the flow of pedestrians, e.g., to adjust walking speeds and even to assess the intrinsic vulnerability of pedestrians. However, we believe that P-PLOS can enable the effective monitoring of the flow of pedestrians and be the basis for further analysis and applications.Finally, it is worth clarifying that this study used the distance recommended by the Spanish authorities as the safe distance. However, the measurement of the pedestrian level of service in times of the pandemic can be easily modified to adapt it to the requirements of the health authorities in other countries and regions. Moreover, the measures recommended with respect to the use of masks should be taken into account, as the risk of contagion could vary significantly. Clearly, P-PLOS has proved to be useful considering the restrictions arising from the COVID-19 pandemic, but it can be applied in any situation where a certain safe distance has to be maintained.The social distancing measures, in their different levels of severity, have had a notable impact on mobility, although their effects on pedestrian mobility have, so far, not been extensively analysed.This article proposes an adaptation of the measurement of pedestrian level of service (PLOS), which we named the pandemic pedestrian level of service (P-PLOS), to incorporate the recommendations on interpersonal distancing levels. The data for the calculation are obtained from pedestrian counters and allow us to assess the service level of a segment of a street and, thus, assess in real or quasi-real time whether there are situations of potential risk when the interpersonal distance is reduced. The results of P-PLOS can assist with the development of health and active mobility policies by providing information on the need for tactical urbanism interventions in the locations and at the times when the levels of service are most deficient. This alliance between sensors and tactical urbanism would allow planners to modify the modal distribution of the streets to achieve more sustainable and safer mobility.Conceptualization, R.T.-G. and R.P.-C.; methodology, R.T.-G. and R.P.-C.; formal analysis, R.T.-G.; writing—original draft preparation, R.T.-G. and R.P.-C.; writing—review and editing, R.T.-G. and R.P.-C.; visualization, R.T.-G. All authors have read and agreed to the published version of the manuscript.This research received no external funding.Not applicable.Not applicable.Further data is available at: https://datos.madrid.es/egob/catalogo/300321-1-aforos-peatones-bicicletas.csv (accessed on 5 April 2021); https://datos.madrid.es/egob/catalogo/300321-4-aforos-peatones-bicicletas.csv (accessed on 5 April 2021); https://datos.madrid.es/egob/catalogo/300321-8-aforos-peatones-bicicletas.csv (accessed on 5 April 2021).RTG thanks the Spanish Government for the ‘Juan de la Cierva’ scholarship (Ref FJCI-2017-31662).The authors declare no conflict of interest.Street characterisation of pedestrian counter locations.1 Peak of pedestrian flow (ped/hour) before COVID-19 (Sep-Dec 2019) and peak hour interval.Study area location. Source: Data from [27] and OpenStreetMap contributors.Methodology diagram.Proposal of social distancing for the walking scenario.Variation (%) in the number of pedestrians from the reference day.Temporal patterns. (a) Day of week, (b) time of day.Change in the number of pedestrians by counter station.Temporal patterns by counter (a) day of week 2019, (b) day of week 2020, (c) time of day 2019, (d) time of day 2020.Comparison of the pedestrian levels of service in normality scenario (a), pandemic scenario (b) and P-PLOS in normality scenario (c).Description of pedestrian-counter locations.1 Puente de Segovia with Paseo Ermita del Santo.Comparative values of PLOS and P-PLOS.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11038.txt

@@ -0,0 +1 @@
+Equal contribution.LPAR6 is the most recently determined G-protein-coupled receptor of the lysophosphatidic acid receptor, and very few of studies have demonstrated the performance of LPAR6 in cancers. Moreover, the relationship of LPAR6 to the potential of prognosis and tumor infiltration immune cells in different types of cancer are still unclarified. In this study, the mRNA expression of LPAR6 and its clinical characteristics were evaluated on various databases. The association between LPAR6 and immune infiltrates of various types of cancer were investigated via TIMER. Immunohistochemistry (IHC) for LPAR6 in lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) tissue microarray with patients’ information was detected. We constructed a systematic prognostic landscape in a variety of types of cancer base on the expression level of mRNA. We enclosed that higher LPAR6 mRNA expression level was associated with better overall survival in some types of malignancy. Moreover, LPAR6 significantly affects the prognostic potential of various cancers in The Cancer Genome Atlas Program (TCGA), especially in lung cancer. Tissue microarrays of lung cancer patient cohorts demonstrated that a higher protein level of LPAR6 was correlated to better overall survival of LUAD rather than LUSC cohorts. Further research indicated that the underlying mechanism of this phenome might be the mRNA expression level of LPAR6 was positively associated to infiltrating statuses of devious immunocytes in LUAD rather than in LUSC, that is, LPAR6 expression potentially contributes to the activation and recruiting of T cells (CD8+ T, naive T, effector T cell) and NK cells and inactivates Tregs, decreases T cell exhaustion and regulates T-helper (Th) cells in LUAD. Our discovery implies that LPAR6 is associated with prognostic potential and immune-infiltrating levels in LUAD. These discoveries imply that LPAR6 could be a promising novel biomarker for indicating the prognosis potential of LUAD patients.Lung cancer is one of the most common malignancies of all incident cases in both men and women around the world, and metastasis is the crucial biological procedure leading to a poor prognosis [1]. It is the top diagnosed malignancy in China and the second most common malignancy in the U.S., and is also the main reason of cancer-related deaths both in China and the U.S. [2]. Scientists have made great efforts to treat various types of lung cancer, but there is still a large amount of time and effort left to do. According to histopathological classification, lung cancer could be classified as two broad subtypes, small cell lung cancer (SCLC) and non-small-cell lung cancer (NSCLC), and the NSCLC is more prevalent [3]. Lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD) are the top two most common subtype NSCLC [4]. Surgery is the primary treatment option during the early stages of the NSCLC, while in late stages, surgery is combined with chemotherapies, and/or radiotherapy [4]. However, though these treatment procedures were applied, the prognosis of the patients remains poor, and also the post-treatment recurrence is the main cause of the disease, where the total 5-year survival rate for all stages is only 16.6% [4].NSCLC has been regarded as a kind of non-immunogenic disease in the past 20 years. However, more and more knowledge of tumor immune interactivities has opposed this model in lung cancer and some other types of malignancy. Immune-related interaction mechanisms act as a crucial role in oncogenesis and development, and immune therapy is considered a promising approach for cancer treatment [4,5], based on this, scientists are trying to employ the body’s own immune system to resist and defeat malignancies [6]. Recently, immunotherapies, including adoptive cell transfers, monoclonal antibodies and vaccines, have become more and more applied to the clinic therapeutics of many types of cancer, such as melanoma, and most recently for lung cancer [7]. In recent years, the finding of antibodies that target the immune checkpoints has revolutionized the treatment of NSCLC, including programmed cell death protein-1 (PD-1) and programmed cell death protein ligand-1 (PD-L1) [8], and these two therapeutic approaches above have showed promising anti-tumor performance in melanoma and NSCLC [9,10,11]. In addition, more and more research has demonstrated that tumor-infiltrating lymphocytes (TILs) play a crucial role in modulating the response to chemotherapy and heighten the clinical prognosis potential of various types of cancer [12,13], for example, tumor-infiltrating neutrophils (TINs) [14,15,16] and tumor-associated macrophages (TAMs), they also associate with the prognosis [17,18,19,20]. So, it is an essential and urgent requirement to elucidate the immunophenotypes of tumor immune interactivities and the identification of new immune therapy targets for lung cancer patients.LPA is a kind of lipid that is involved in the proliferation of tumor cells via its G-protein coupled (GPC) receptors [21,22] and one of their receptors—LPAR6—is the latest identified receptor of LPA [23,24]. Moreover, it has been demonstrated to be related to many types of tumors, including prostate [25], liver [26,27], colorectal [28,29] and pancreatic cancer [30]. However, the function of LPAR6 remains highly controversial since in colorectal cancer, scientists found that LPAR6 acts as a tumor suppressor, whereas it acts as a facilitator in the other types of tumors [25,26,27,30,31]. All these indicate that LPAR6 plays an important role in cancer, whereas the relationship between LPAR6 and tumor biology and the underlying mechanism involved is still unclear.Here, we investigated the mRNA expression level of LPAR6 and the correlation with prognosis patterns of cancer patients in databases using bioinformatics tools. In addition, we analyzed the correlation of LPAR6 with tumor-infiltrating immune cells (TIICs) in various tumor microenvironments via TIMER. Moreover, IHC staining for LPAR6 in two separate lung cancer cohorts with patients’ information was detected to analyze the correlation of the expression of LPAR6 and the clinicopathological parameters of lung cancer.All these discoveries indicated the crucial role of LPAR6 in lung cancer and also provided a potential correlation and identified the mechanism involved between LPAR6 and tumor–immune interactions.This project was permitted and under the supervision of Ethics Committee of Shanghai Jiao Tong University School of Medicine.The mRNA expression level of the LPAR6 in various types of cancer was investigated via different database, including Oncomine, TIMER and GEPIA2 [32]. The threshold in Oncomine database was as follows: p-value of 10−4, fold change of 1.5, and gene ranking top 5%.The association between LPAR6 mRNA expression level and survival rate in different types of cancer was investigated using the PrognoScan and GEPIA2 database, which searching for relationships between gene expression and the prognoses of patients, such as overall survival (OS) and disease-free survival (DFS), across a large collection of microarray datasets [33,34]. The threshold was adjusted to a Cox p-value < 0.05.The correlation between the mRNA expression level of LPAR6 and survival rate as well as different cancer staging in various cancers was determined by Kaplan-Meier plotter database [35]. The HRs with 95% confidence intervals (Cls) and log-rank p-value were analyzed.UALCAN [36] could be employed to investigate methylation and mRNA expression levels, also the survival of a specific target gene across several clinicopathological features, such as stages and age. The t-test was performed to compare the statistical significance.GeneMANIA is identified single genes related to a set of input genes [37] to construct the LPAR6 biological network based on a set of function-association data, including co-expression, genetic and protein interaction pathways, colocalization and protein domain homology.Thirty-two types of cancer and over 10,000 patients from TCGA were included in the LinkedOmics database [38]. LinkFinder was employed to determine the differentially expressed genes (DEGs) in TCGA. LUAD and LUSC cohorts whose expression levels correlated with those of LPAR6. These results were investigated by using Pearson’s correlation coefficient. LinkInterpreter was employed to identify the pathways and networks [39].We investigated the mRNA expression level of LPAR6 in various types of malignancy and the association of LPAR6 level the abundance of immune-infiltrating, including various types of T cells (CD4+ T cells, CD8+ T cells), B cells, macrophages, neutrophils, and DCs, via gene modules in TIMER database [40,41,42,43]. In addition, associations between LPAR6 expression level and marker genes of TIICs were explored via correlation modules. The marker genes of TIICs included markers of T cells (CD8+ T cells, general T cells), B cells, TAMs, monocytes, macrophages (type 1 macrophages, M1 and Type 2 macrophages, M2), neutrophils, natural killer (NK) cells, dendritic cells (DCs), T-helper cells (Th1, Th2 and Th17), follicular helper T cells (Tfh), T regulatory cells (Tregs) and exhausted T cells. The gene marker sets are referenced in our previous publication [44,45]. The expression level of the genes was demonstrated by using log2 RSEM.Here, two tissue microarrays were constructed using formalin-fixed paraffin-embedded (FFPE) tissue samples from LUAD (LUC1601) and LUSC (LUC1602) patients, each tissue microarrays (TMA) chip containing 74 and 78 paired tumor and adjacent normal tissues were purchased from the Superbiotek Co., Ltd., (Shanghai, China), respectively. Clinicopathological data including subtype, histological grading, and tumor/nodal stage and information about patient follow-up could be retrieved from the database of the Shanghai Jiao Tong University School of Medicine.The tissue sections underwent immunohistochemical staining using a primary antibody to LPAR6 (Thermo Fisher/Invitrogen, Waltham, MA, USA) (Cat No. PA5-33901) at a dilution of 1: 100. Sections of the TMAs were used for investigating the protein level of LPAR6 following the general standard IHC staining protocols.The results produced via Oncomine as mentioned in Section 2.1. The consequence of Kaplan–Meier plots, GEPIA, and PrognoScan are exhibited with HR and p or Cox p-values from a log-rank test. Further, the correlation coefficient of gene expression was evaluated by Spearman’s correlation and p-values < 0.05 were considered statistically significant. Protein level was determined by the staining intensity and the distribution of the positive cells, which were performed by two independent pathologists blinded to the clinical information of the patients as described [45].To investigate the varied mRNA expression level of LPAR6 in tumor and normal tissues, the LPAR6 expression levels were determined using the dominant online database, Oncomine and GEPIA2. Here, we enclosed the LPAR6 expression was higher in brain and central nervous system (CNS) cancer, gastric, liver cancer, kidney, lymphoma and pancreatic cancer compared to the normal tissues and lower mRNA expression level of LPAR6 was observed in breast, bladder, colorectal, cervical, lung, esophageal, prostate cancer and some other types of cancer compared to the adjacent normal tissues (cancer vs. normal) (Figure 1A). This detail of the expression level of LPAR6 in various types of cancer is summarized in Supplementary Table S1. To evaluate the mRNA expression level of LPAR6 in cancers, we studied the levels of LPAR6 expression employing the RNA-Seq datasets in the TCGA. The varied expression levels between tumor and adjacent normal tissues for LPAR6 across each type of TCGA tumors is demonstrated in Figure 1B. The expression level of LPAR6 was significantly lower in the tumor tissue of bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), head and neck cancer (HNSC), kidney chromophobe (KICH), LUAD, prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ) and uterine corpus endometrial carcinoma (UCEC) compared with adjacent normal tissues and was significantly higher in esophageal carcinoma (ESCA), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), thyroid carcinoma (THCA) compared with the normal tissues (Figure 1B). GEPIA2 generates dot plots to profile gene/isoform expression across various types of cancer and paired normal tissue samples, and each dot representing a distinct sample. The differential mRNA expression level of LPAR6 between tumor and matched normal and GTEx data across all TCGA tumors by GEPIA2 are demonstrated in Figure 1C. LPAR6 expression was significantly higher in glioblastoma multiforme (GBM), KIRC, LAML, lower grade glioma (LGG), PAAD, thymoma (THYM) and lower in adrenocortical carcinoma (ACC), ESCA, KICH, LUAD, PRAD, testicular germ cell tumors (TGCT), UCEC and UCS compared with normal GTEx tissues. From these above, we found that the expression patterns are different in two types of lung cancers, LUAD and LUSC.We analyzed whether the mRNA expression level of LPAR6 was associated with the prognosis specific across cancer patient cohorts. The effects of LPAR6 mRNA expression on the various survival rates were assessed using PrognoScan. The detailed relationship between the mRNA expression level of LPAR6 and Prognostic potential of various cancers are listed in Supplementary Table S2. Notably, the expression level of LPAR6 impacts OS in lung cancer significantly (Figure 2A,C). Two cohorts (GSE3141 and GSE4573) of lung cancer demonstrated that high expression level of LPAR6 was associated with better prognosis (OS HR = 0.53, 95% CI = 0.36 to 0.80, Cox p = 0.00206181; OS HR = 0.53, 95% CI = 0.31 to 0.91, Cox p = 0.0219869). (Figure 2A,B), and better post-progression survival (PPS) (Figure 2D). So, it is conceivable that high LPAR6 expression is an independent risk factor and leads to a better prognosis in lung cancer patients, and a hazard ratio (HR) below 0 indicates LPAR6 expression is a protective factor. Additionally, high LPAR6 expression significantly impacts disease-specific survival (DSS) in bladder cancer (Supplementary Figure S1A) and OS (Supplementary Figure S1B,M), relapse-free survival (RFS) (Supplementary Figure S1C), DSS (Supplementary Figure S1D,H), distant metastasis-free survival (DMFS) (Supplementary Figure S1N) in breast cancer. Moreover, three cohorts (GSE19615, GSE9195 and GSE11121) of breast cancer demonstrated that higher expression level of LPAR6 was correlated with a better prognostic potential of DMFS (Supplementary Figure S1E–G). Further, a higher LPAR6 expression level was associated with better prognosis potential in some other types of cancer (Supplementary Figure S1I–K).In addition to microarray analysis data of LPAR6, the RNA-Seq was also used to analyze the prognosis of LPAR6 in various types of cancers via the same database. A better prognosis in liver cancer is shown to be associated with a higher LPAR6 expression level (Supplementary Figure S1O–Q). The different correlation patterns between adenocarcinoma and squamous cell carcinoma of lung cancer attracted our attention (Figure 2E,F). These data confirmed the prognostic value of LPAR6 in some specific types of cancers, that is, the increased or decreased LPAR6 expression has different prognostic values depending on the type of cancers.Here we studied the association with the expression level of LPAR6 and different clinical characteristics (stages and treatments) in order to better disclosure the mechanisms and relevance of the mRNA expression level of LPAR6 in different types of cancer, especially in different clinical stages of lung cancer patients.We found that high mRNA expression level of LPAR6 was associated with better OS only in Stage 1 and 2 of LUAD (OS HR = 0.27, p = 4.6 × 10−10; OS HR = 0.51, p = 0.0073) not in LUSC (Table 1). This arrestive phenomenon combines with the different survival patterns of LUAD and LUSC in Figure 2E,F may indicate the correlation of LPAR6 expression and the prognosis of different types of cancer depends on the different mechanisms in the carcinogenesis and development.The lower promoter methylation levels of LPAR6 were detected in the earlier stage, implying that lower promoter methylation levels of LPAR6 were correlated with the earlier stages of the progress of lung cancer (Figure 3). We also found that late-stage (stage 4) with the lowest promoter methylation showed levels of LPAR6 in LUSC whereas it is the highest methylation level in entire cancer progress in LUAD cohorts (stage 1–4) (Figure 3A,E), and the lowest promoter methylation levels of LPAR6 appears in an earlier stage (stage 2) of LUAD while in the late stage of LUSC. In addition, the results of promoter methylation here could indicate how LPAR6 expression levels fluctuate in various stages of lung cancer. What interested us is that the same pattern was detected in nodal metastasis analysis, which implies that in the later stage, the promoter methylation levels of LPAR6 are correlated with nodal metastasis in some way (Figure 3D,H). The promoter methylation levels of LPAR6 share a similar pattern in LUAD and LUSC among different races and different ages, respectively; that is, both in the African-American group and younger group of these two cohorts had the lowest promoter methylation levels of LPAR6 (Figure 3B,F).An interaction network of LPAR6 was constructed to determine potential interactions between LPAR6 and other cancer-associated proteins. These data demonstrated that LPAR6 has co-expression with 19 proteins, shared protein domains with ADRB2 and physical interactions with DMD (dystrophin) (Figure 4A). Moreover, the analysis of interactions network of LRP6 with the 19 protein and its impact on lung cancer progression in Human Protein Atlas demonstrated that the “19 protein set” associate to a better prognosis in LUAD rather than that of LUSC (Supplementary Figure S2). Firstly, we found some proteins in the “19 protein set” are significantly positively correlated with the prognosis in LUAD patients, while these genes are significantly negatively correlated with the prognosis in LUSC patients, including MAL (T cell differentiation protein), CXCL12 (C-X-C motif chemokine ligand 12), ADRB2 (Adrenoceptor beta 2). Secondly, the protein level of DMD (Dystrophin) and IL7R (Interleukin 7 receptor) showed a significantly positive pattern in LUAD cohorts rather than that of LUSC cohorts. Thirdly, the protein level of MYCT1 (MYC target 1), EDNRA (Endothelin receptor type A) and F2R (Coagulation factor II thrombin receptor) demonstrated a significantly negative pattern in LUSC patients rather than that of LUAD patients (SI-Figure S2). All these above indicate that the protein level of LPAR6 and the “19 protein set” associate to a better prognosis in LUAD rather than that of LUSC (Supplementary Figure S2). LinkedOmics were then used to analyze the genes that co-expressed with LPAR6 in lung cancers. The volcano plot elucidated that the expression of genes was negatively correlated with that of LPAR6 (green spot; false discovery rate (FDR) < 0.05), while genes expression is positively correlated with LPAR6 (red spot; FDR < 0.05; Figure 4B). The top 50 positively and negatively correlated genes are showed in Figure 4B. These results imply that LPAR6 serves an important role in cancer development. Biological process and molecular function studies were conducted using gene set enrichment analysis, which showed that LPAR6-associated DEGs were involved in several kinds of immune biology process such as ‘interleukin production’, ‘respiratory burst’, ‘leukocyte proliferation’, ‘T cell activation’, ‘adaptive immune response’ were involved in LUAD and LUSC, respectively (Figure 4C). All these data imply that LPAR6 may serves a key role in immune system activation, metabolism, cellular responses to stimulation and many other processes.TILs have been proved as an independent predictor of survival in different types of cancer [46,47]. So, in this study, we determined whether the expression level of LPAR6 correlates with the immune-infiltration levels in various types of cancer. We analyzed the correlations of LPAR6 mRNA expression with the immune-infiltration levels in nearly 40 types of cancer. The results show that the mRNA expression level of LPAR6 has significant negative correlations with tumor purity in 26 types of cancer which indicating LPAR6 somehow related to recruiting lymphocytes to tumor and significant correlated with B cell infiltration levels in 13 types of cancers. In addition, the expression level of LPAR6 has significant correlations with infiltrating levels of CD8+ T cells in 24 types of cancer, CD4+ T cells in 26 types of cancer, neutrophils in 33 types of cancer, macrophages in 20 types of cancer and dendritic cells in 21 types of cancer (Supplementary Table S3 and Figure S3).Given the correlation of the mRNA expression level of LPAR6 with immune-infiltration level in diverse types of cancer, we investigated the distinct types of cancer in which LPAR6 was correlated with prognosis and immune-infiltration. Tumor purity is a crucial factor that influences the analysis of immune-infiltration in clinical tumor samples by genomic approaches [34,41]. So here we selected the cancer types in which LPAR6 expression levels have a significant negative correlation with tumor purity in TIMER and a significant correlation with prognosis. Interestingly, we found that the expression level of LPAR6 expression correlates with better OS and high immune-infiltration levels in breast cancer, liver cancer and LUAD rather than in LUSC.The LPAR6 expression level of LUAD and LUSC are all significantly negatively related to tumor purity (Figure 5). LPAR6 mRNA expression level has significant positive correlations with the infiltrating levels of B cell, CD4+ T cells, CD8+ T cell, macrophages, neutrophils and DCs in LUAD (Figure 5). What interested us is that the correlation with immune cells demonstrated a different pattern in LUAD and LUSC of lung cancers. These findings strongly suggest that LPAR6 plays a specific role in immune-infiltration in different types of lung cancer, and leads to a better prognosis in LUAD instead of in LUSC.To study the association between LPAR6 and different types of TIICs, we focused on the correlations between the expression level of LPAR6 and immune marker sets of various immune cells of LUAD and LUSC. The correlations between LPAR6 expression level and immune marker gene sets of different immune cells, including CD8+ T cells, T cells (general), B cells, monocytes, TAMs, M1 and M2 macrophages, neutrophils, NK cells and DCs were determined in LUAD and LUSC (Table 2 and Figure 6). We also investigated the different types of T cells (Th1, Th2, Tfh, Th17, Tregs and exhausted T cells). After adjustment by purity, the correlation results revealed the LPAR6 expression level was significantly correlated with most immune marker sets of various immune cells and different subtypes of T cells, especially effect T cells in LUAD. However, none of these gene markers was significantly correlated with the LPAR6 expression level in LUSC and other cancer with poor prognosis (Table 2 and Figure 6).These results demonstrated that the mRNA expression levels of the marker genes in T cells (general, CD8+, Naive T, Effector T), natural killer (NK) cell, M1 macrophages and DCs have strong correlations with LPAR6 expression in LUAD (Table 2). More specifically, we demonstrated NOS2, IRF5, PTGS2 of M1 phenotype are significantly correlate with LPAR6 expression in LUAD (p < 0.0001; Figure 4A–H). It is reported that M1 could prevent tumor development. In-depth studies need to be done on whether LPAR6 is a crucial factor that mediating the de-polarization of macrophages and remodel tumor microenvironment. In addition, for Treg cells, LPAR6 does not demonstrate a correlation with the Tregs markers such as STAT5B in liver hepatocellular carcinoma (LIHC) (Table 2). Furthermore, we determined the association between the expression level of LPAR6 and the above marker sets of monocytes and various types of T cells in normal and tumor tissue in LUAD and LUSC (Supplementary Materials—Table S4, Figure 7).The more interesting thing is that the expression levels of most of the marker sets of these immunocytes have strong correlations with LPAR6 expression in the tumor tissue of LUAD patients. In the LUSC, there was no significant correlation between LPAR6 and markers of immune cells (Figure 8, Supplementary Materials—Table S4). This finding suggests that there are different correlation patterns between tumor and normal tissue in LUAD patients. This exciting finding indicates that LPAR6 may regulate macrophage de-polarization in the tumor microenvironment of the LUAD and LPAR6 might be a novel target for LUAD therapy. High LPAR6 expression relates to a high infiltration level of DCs in the tumor tissue of LUAD patients, DC markers such as HLA-DQB1, CD1C and NRP1 show significant correlations with LPAR6 expression both in the tumor tissue in LUAD (Supplementary Materials—Table S4). These results further reveal that there is a strong relationship between LPAR6 and DCs infiltration.We analyzed the protein level of LPAR6 in two independent lung cancer patient cohorts with 74 and 77 paired lung cancer and normal tissues, respectively. LPAR6 is mainly expressed in the cytoplasm of the cells (Figure 9A), and the protein level was lower in the lung cancer tissues compared with the normal tissues (Figure 9B,D). In the next step, we investigated the relationship between the LPAR6 protein level and the clinical characteristics of the LUAD and LUSC patient cohorts. Lung cancer patients with higher LPAR6 levels demonstrated better OS than those patients with relatively lower levels in LUAD patient cohorts, but not in LUSC patient cohorts (Figure 9C,E). Moreover, we found that lower LPAR6 was negatively correlated with the clinical stage of lung cancer and the lymph node metastasis of patients (Figure 9F,G).In summary, we demonstrated that the LPAR6 was downregulated in the tumor tissue of LUAD patients and its mRNA expression was positive associated with the OS of LUAD patients base on the databases and TMA cohorts. The results further confirm that LPAR6 is specifically correlated with immune infiltrating cells in LUAD which suggests that LPAR6 plays a vital role in immune cells recruiting the tumor tissue in LUAD patients. LPAR6 and its modulation on tumor microenvironment may serve as a novel therapeutic target for LUAD.LPA receptors are G-protein-coupled receptor (GPCR) that bind to the LPA and trigger a series of down-streaming cellular responses, including cell proliferation, apoptosis and motility [48,49,50]. Previously, five LPA receptors (LPAR1-5) are well characterized and extensively studied [51]. LPAR6 is a recently determined GPCR, alias as ARWH1, HYPT8, LAH3, P2RY5, at first was considered as purinergic receptor P2Y5 that involved in inherited hair loss [23,52]. Although LPAR6 has not been extensively studied, it was reported that the LPAR6 suppresses tumor cell migration in colorectal cancer [28], and the expression of LPAR6 was decreased in P53-mutated cases [29]. It was also reported that the LPA axis plays a crucial role in hepatocellular carcinoma (HCC) by recruiting and trans-differentiating of peritumoral fibroblasts into TAMs [53]. This offers scientists a promising hint that LPAR6 is involved in the TME. Immunotherapy is a new genre of treatment for patients and has a tightly association with tumor microenvironment (TME) [29].In this study, we announced that different mRNA expression levels of LPAR6 are associated with the prognostic potential in various types of cancer. Higher level of LPAR6 is associated with a better prognosis in three types of cancers, including liver cancer, lung cancer and breast cancer. Moreover, our data demonstrated that the immune-infiltration levels and diverse immune marker panels of the different subtypes of lung cancers (LUAD and LUSC) are associated with the expression level of LPAR6. To this end, our study provides a novel insight into elucidating the potential role of LPAR6 in tumor immunology and its usage as a prognostic biomarker and novel therapy target for LUAD.In this work, we determined the LPAR6 expression levels and constructed a systematic prognostic landscape in various types of cancer using independent datasets and 33 type cancers of TCGA data in online database. The variation expression level of LPAR6 between cancer and normal tissues was determined in many cancer types. Relaying on the Oncomine database, we found that LPAR6, compared to normal tissues, was highly expressed in the tumor tissue of brain and CNS, kidney, gastric cancer, leukemia, lymphoma, liver and pancreatic cancer while some data sets showed that LPAR6 has a lower mRNA expression level in bladder, breast, cervical, colorectal, esophageal, lung and prostate cancer (Figure 1A). However, the redetermination of the TCGA data demonstrated that LPAR6 expression was more expressed in ESCA, KIRC, KIRP and THCA, but significantly lower expressed in BLCA, COAD, BRCA, HNSC, KICH, PRAD, LUAD, UCEC, READ and slightly lower in LIHC compared with paired adjacent normal tissues (Figure 1B). These vary in the mRNA expression levels of LPAR6 in different types of cancer among various databases might be a reflection in data collection approaches and underlying mechanisms involved in different biological properties. Nevertheless, we found similar prognostic associations between LPAR6 expression in bladder, breast, cervical, colorectal, esophageal, lung and prostate cancers in these databases. Investigation of the TCGA database enclosed that the higher LPAR6 expression level is correlated with better prognostic potential in ACC, LGG, skin cutaneous melanoma (SKCM). Furthermore, the determination of patient cohorts from PrognoScan database and Kaplan–Meier Plotter demonstrated a high mRNA expression level of LPAR6 is correlated with better prognosis in breast, lung, bladder, colorectal, eye and ovarian cancer (Figure 2). In two datasets of PrognoScan, higher LPAR6 expression levels could be considered as an independent risk factor for better prognosis in LUAD. Moreover, a high level of LPAR6 expression was shown to be correlated with a better prognosis of LUAD in the early stage with the lowest HR [0.27 (0.17–0.42)] for a better OS when LPAR6 was highly expressed in LUAD, rather than in LUSC. These together strongly suggest that LPAR6 could be a potential prognostic biomarker in LUAD.Another crucial aspect is that the mRNA expression level of LPAR6 is correlated with diverse immune-infiltration levels in cancer, particularly in LUAD. Here, we demonstrate that there’s a strong positive correlation between the infiltration level of T cells (CD4+ T cells and CD8+ T cells), neutrophils, macrophages and DCs and LPAR6 expression in LUAD (Figure 3A,C). Moreover, the correlation patterns of the infiltration level are different in two types of lung cancers (LUSC and LUAD). The correlation between LPAR6 mRNA expression and the marker gene panel of immune cells implicates the role of LPAR6 in regulating tumor immunology in these types of cancer. A possible reason for this striking phenomenon might be that LPAR6 orchestrates the function of multiple immune marker gene sets. This supports the argument that the LPAR6 expression levels are important contributors to human malignancies and indicating the prognostic potential of specific types of cancer.Firstly, the markers of M1 macrophages such as PTGS2 and IRF5 show a weak value correlation with LPAR6 expression in LUAD, respectively (Table 2). Since macrophages are functionally cells. M1 producing type 1 cytokines prevent tumors from developing, whereas type 2 and M2 macrophages M2 inducing type 2 cytokines facilitate tumor growth. Especially in the tumor tissue of LUAD, both IRF5 and NOS2 show significant correlations with LPAR6 expression and PTGS2 shows a significant correlation with LPAR6 expression in the tumor tissue (Supplementary Materials—Table S4). These results indicate the potential regulating role of LPAR6 in de-polarization macrophages against tumor that activated macrophages can be re-polarized into opposite functional phenotypes by microenvironmental modifications and then inhibit tumor growth.Secondly, these results implied that LPAR6 has the potential to activate various types of T cells, including CD8+ T cell, naive T-cell, effector T-cell and natural killer cell and inactivate Tregs and decrease T cell exhaustion. CD8A, a crucial protein on T cells surface, is highly correlated with LPAR6 expression in LUAD which are types of cancers with better prognosis. Further, CD8A did not demonstrate a significant correlation pattern in LUSC (Table 2). This pattern also occurs with the general T cell markers such as CD3D, CD3E, CD2 and most markers of naive T-cell, effector memory T-cell, effector T-cell and natural killer cells, such as LEF1 which has been proven to be a predictor of better treatment response in AML. This is because a higher mRNA expression level of LEF1 was associated with favorable RFS in patients and predicted a significantly better overall survival for AML patients [54].Thirdly, different correlation patterns have been found between the mRNA expression level of LPAR6 and the regulation of markers of T helper cells (Th1, Th2, Th17 and Tfh) in these different cancers. IFN-g is a Th1 cytokine with both pro- and anti-cancer properties [55] which are highly correlated with LPAR6 expression in LUAD, whereas it did not demonstrate significant correlations in LUSC (Table 2). IL-13 is an important immunoregulatory cytokine mainly produced by activated type II helper T cells and is widely involved in tumorigenesis and development, fibrosis and inflammation [56,57]. We found that IL-13 is highly correlated with LPAR6 expression in LUAD, but did not demonstrate significant correlations in LUSC (adjusted by purity), and a similar situation is in the IL-21. So, these would be explanations that show why LPAR6 is an indication of poor prognosis in LUSC and a better prognosis in LUAD. We also studied the prognostic potential of LPAR6 in KIRC and the correlation of LPAR6 and immune-infiltration. We found that it is the same pattern with LUAD (data unshown). So, we believe that LPAR6 might also be a potential biomarker associated with KIRC also via immune infiltration.All these correlations above are implying a potential mechanism that LPAR6 regulates T cell activities in LUAD. These findings indicate that the LPAR6 plays a crucial role in the recruitment and regulation of effective T cells infiltrating in LUAD patients would lead to a better prognosis.In this study, we offered a potential explanation for the mechanism that why the mRNA expression level and protein of LPAR6 correlates with immune infiltration level and associates to a better prognostic potential in some specific types of cancer, especially in LUAD. Hence, the interactions between LPAR6 and the immunocytes in the tumor microenvironment could be an underlying mechanism for the correlation of LPAR6 expression level with the immune infiltration level and a better prognosis in LUAD patients.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111038/s1, Figure S1: Kaplan-Meier survival curves comparing the high and low expression of LPAR6 in different types of cancer in the PrognoScan (A–K) and Kaplan–Meier plotter databases (L–Q). (A) Survival curves of DSS in bladder cancer cohort [GSE13507 (n = 165, p = 0.0067285)]. (B–D) Survival curves of OS, RFS and DFS in the breast cancer cohort [GSE1456-GPL96 (n = 159, p = 0.00575883; p = 0.0000252; p = 0.000210173)]. (E-G) Survival curves of DMFS in the breast cancer cohort [GSE19615 (n = 159, p = 0.00575883), GSE9195 (n = 159, p = 0.0466683), GSE11121 (n = 200, p = 0.0389008)]. (H-J) Survival curves of DSS in the breast, DFS in the colorectal and DMFS in the Eye Cancer cohort [GSE3494 (n = 236, p = 0.00294205), GSE17537 (n = 55, p = 0.0257972), GSE22138 (n = 63, p = 0.00092478)]. (K) Survival curves of PFS in the ovarian cancer cohort [GSE17260 (n = 110, p = 0.0392865)]. (L-N) Survival curves of OS (n = 1402), RFS (n = 3951) and DMSF (n = 1746) in the breast cancer cohorts. (O-Q) Survival curves of OS (n = 364), FPS (n = 370) and RFS (n = 316) in the liver cancer cohort. Supplementary Figure S2: Promoter methylation levels of LPAR6 impacts the clinicopathological parameters in LUAD and LUSC cohorts. Supplementary Figure S3: The correlation of the expression of LPAR6 with immune infiltration level in cancers. Supplementary Material Table S1. LPAR6 expression in Oncomine database; Supplementary Material Table S2. Relation between LPAR6 expression and patient prognosis of different cancer in Prognoscan database; Supplementary Material Table S3: The expression level of LPAR6 is correlated with immune infiltration level in various types of cancers; Supplementary Material Table S4: Correlation analysis between LPAR6 and relate genes and markers of immune cells in tumor and normal tissue in LUAD and LUSC.J.H. contributed to idea, conception and study design. J.H. and Y.S. collected and analyzed the datasets. M.M., R.G. and M.Y. performed TMA analysis. J.H., R.G. and M.M. wrote the manuscript and generated the figures. C.L., J.L. and H.W. revised and proofread the article. All authors have read and agreed to the published version of the manuscript.This study was supported by grants from the National Natural Science Foundation (81630086, 82030099), the National Key R&D Program of China (2018YFC2000700), Shanghai Public Health System Construction Three-Year Action Plan (GWV-10.1-XK15), Innovative research team of high-level local universities in Shanghai of H.W., the Instrument Developing Project of Chinese Academy of Sciences (YJKYYQ20200038 and ZDKYYQ20200004), and SIBET funding (E0290104) of Y.S. and the Start-up Plan for New Young Teacher of SHSMU (KJ30214190026) of J.H.This project was permitted by Ethics Committee of Shanghai Jiao Tong University School of Medicine.Not applicable.The data analyzed or generated during this study are included in this article and its supplementary information.We would like to thank to the funder.All of the authors declare there are no conflict of interest.LPAR6 mRNA expression levels in different types of human cancers in different databases. (A) Increased or decreased LPAR6 in data sets of different cancers compared with normal tissues. Cell color is determined by the best gene rank percentile for the analyses within the cell. (B) Human LPAR6 expression levels in different tumor types from TCGA database. One category of cancer is in one box, and paired tissue (tumor and adjacent) are in grey boxes. p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. (C) LPAR6 expression profile across all tumor samples and paired normal tissues (Dot plot) via GEPIA. Each dots represent the expression of samples.Kaplan–Meier survival curves comparing the high and low expression of LPAR6 in different types of cancer in the PrognoScan and Kaplan–Meier plotter databases. (A,B) Survival curves of OS in two lung cancer cohorts [GSE3141 (n = 111, p = 0.00206181) and GSE4573 (n = 129, p = 0.0219869)] and DSS in bladder cancer cohort [GSE13507 (n = 165, p = 0.0067285)]. (C,D) Survival curves of OS (n = 1926) and PPS (n = 344) of the lung cancer. (E,F) Survival curves of OS of the lung adenocarcinoma (n = 720) and squamous cell carcinoma (n = 524).Promoter methylation levels of LPAR6 impacts the clinicopathological parameters. (A,E) stage, (B,F) race, (C,G) age, (D,H) nodal metastasis status in LUAD and LUSC cohorts respectively. * p < 0.05 compare with the normal tissue, △ p < 0.05 compare within different group in the tumor tissue.Biological interaction network of LPAR6. LPAR6 interaction network in TCGA, different colors represent diverse bioinformatics methods (A) and differentially expressed genes in correlation with LPAR6 and heat maps of positively and negatively correlated genes with LPAR6 in LUAD and LUSC were analyzed by Pearson test (B). Enriched gene ontology annotations of biological process and molecular function analysis of LPAR6 correlated genes in LUAD and LUSC (C). Red indicates positive and blue indicates negative. Dark blue and orange indicate FDR ≤ 0.05, light blue and orange indicate FDR > 0.05. FDR, false discovery rate.Correlation of LPAR6 expression with immune infiltration level in (A) LUAD, (B) LUSC. (A) LPAR6 expression is significantly negatively related to tumor purity and has significant strong positive correlations with the level of B cells, CD8+ T cells, macrophages, neutrophils, and DCs in LUAD (n = 515). (B) LPAR6 expression is significantly negatively related to tumor purity and has weak positive correlations with infiltrating levels of neutrophils in LUSC but no significant correlation with infiltrating levels of B cells, CD8+ T cells, macrophages and DCs (n = 501).Correlation analysis between LPAR6 expression and immune marker sets in LUAD and LUSC. Markers include CD8A and CD8B of CD8+ T cell; CD3D, CD3E and CD2 of general T cell; FOXP3, CCR8, STAT5B and TGFB1 of Treg; PDCD1, CTLA4, LAG3, HAVCR2 and GZMB of exhausted T cells; CD163, VSIG4 and MS4A4A of M2 macrophages; CD86 and CSF1R of monocytes; HLA-DPB1, HLA-DQB1, HLA-DRA, HLA-DPA1, CD1C, NRP1 and ITGAX of Dendritic cell. (A–Z) Scatterplots of correlations between LPAR6 expression and gene markers of CD8+ T cell (A,B), general T cell (C–E), Treg (F–I), T cell exhaustion (J–N), M2 macrophage (O–Q), monocyte (R–S) and dendritic cell (T–Z) in LUAD. (AA–AZ) Scatterplots of correlations between LPAR6 expression and gene markers of CD8+ T cell (AA,AB), general T cell (AC–AE), Treg (AF–AI), T cell exhaustion (AJ–AN), M2 macrophage (AO–AQ), monocyte (AR–AS) and dendritic cell (AT–AZ) in LUSC.Correlation analysis between LPAR6 expression and various T cell marker sets in LUAD and LUSC. (A–AE) Scatterplots of correlations between LPAR6 expression and gene markers of Naive T-cell (CCR7, LEF1, TCF7, SELL) (A–D), Effector T-cell (CX3CR1, FGFBP2, FCGR3A) (E–G), Effector memory T-cell (PDCD1, DUSP4, GZMK, GZMA, IFNG) (H–L), Central memory T-cell (CCR7, SELL, IL7R) (A,D,M), Resident memory T-cell (CD69, ITGAE, CXCR6, MYADM) (N–Q), T cell exhaustion (HAVCR2, TIGIT, LAG3, PDCD1, CXCL13, LAYN) (H,R–V), Resting Treg (FOXP3, IL2RA) (W,X), Effector Treg (FOXP3, CTLA4, CCR8, TNFRSF9) (W,Y–AA), Th1-like (HAVCR2, IFNG, CXCR3, BHLHE40, CD4) (L,AB–AE) in LUAD; (AF–BJ) Scatterplots of correlations between LPAR6 expression and gene markers of Naive T-cell (CCR7, LEF1, TCF7, SELL) (AF–AI), Effector T-cell (CX3CR1, FGFBP2, FCGR3A) (AJ–AL), Effector memory T-cell (PDCD1, DUSP4, GZMK, GZMA, IFNG) (AM–AQ), Central memory T-cell (CCR7, SELL, IL7R) (AF,AI,AR), Resident memory T-cell (CD69, ITGAE, CXCR6, MYADM) (AS–AV), T cell exhaustion (HAVCR2, TIGIT, LAG3, PDCD1, CXCL13, LAYN) (AM,AW–BA), Resting Treg (FOXP3, IL2RA) (BB,BC), Effector Treg (FOXP3, CTLA4, CCR8, TNFRSF9) (BB,BD–BF), Th1-like (HAVCR2, IFNG, CXCR3, BHLHE40, CD4) (AQ,BG,BH, BI,BJ) in LUSC.Correlation analysis between LPAR6 expression and various immune cells in normal and tumor tissue of LUAD and LUSC. (A–R) Scatterplots of correlations between LPAR6 expression and naive T-cell (A,B), effector T-cell (C,D), effector memory T-cell (E,F), central memory T-cell (G,H), resident memory T-cell (I,J), T cell exhaustion (K,L), resting Treg (M,N), effector Treg (O,P), Th1-like (Q,R) in the normal and tissue of LUAD; (S–AJ) Scatterplots of correlations between LPAR6 expression and gene markers of naive T-cell (S,T), effector T-cell (U,V), effector memory T-cell (W,X), central memory T-cell (Y,Z), resident memory T-cell (AA,AB), T cell exhaustion (AC,AD), resting Treg (AE,AF), effector Treg (AG,AH), Th1-like (AI,AJ) in LUSC.Higher expression of LPAR6 was correlated with clinicopathological parameters in LUAD cohort and was associated with increased overall survival (OS) of LUAD and LUSC patients. (A) Immunohistochemistry staining of the LPAR6 in the tumor and adjacent normal tissues. Red arrows indicated the cytoplasm-stained LPAR6. Bar, 50 μm; (B) The immunoreactive score (IRS) of the cytoplasm LPAR6 staining in 74 paired lung cancer tissues in LUAD cohort; (C) The Kaplan–Meier plot of the OS for lung cancer patients with relatively higher or lower LPAR6 expression levels in LUAD cohort (N = 74; Log-rank test, p = 0.02); (D) The IRS of the cytoplasm LPAR6 staining in 78 paired lung cancer tissues in LUSC cohort; (E) The Kaplan–Meier plot of the overall survival for lung cancer patients with relatively higher or lower LPAR6 expression levels in LUSC cohort (N = 78; log-rank test, p = 0.04). (F,G) The proportion of LPAR6 expression level (higher or lower) in different clinical stages, lymph node metastasis, organ metastasis and tumor size of patients in LUAD (F) and LUSC (G) cohorts.Correlation of the mRNA expression level of LPAR6 in different stage and clinical prognostic potential in lung Cancer with different clinicopathological factors.Bold values indicate p < 0.05.Correlation analysis between LPAR6 and relate markers of immune cells.TAM, tumor-associated macrophage; Th, T helper cell; Tfh, Follicular helper T cell; Treg, regulatory T cell; Cor, R value of Spearman’s correlation; None, correlation without adjustment. Purity, correlation adjusted by purity. * p < 0.01; ** p < 0.001; *** p < 0.0001.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11039.txt

@@ -0,0 +1 @@
+Maryland’s growing chicken industry, including concentrated animal feeding operations (CAFOs) and meat processing plants, raises a number of concerns regarding public health and environmental justice. Using hot spot analysis, we analyzed the totality of Maryland’s CAFOs and meat processing plants and those restricted to the Eastern Shore to assess whether communities of color and/or low socioeconomic status communities disproportionately hosted these types of facilities at the census tract level. We used zero-inflated regression modeling to determine the strength of the associations between environmental justice variables and the location of CAFOs and meatpacking facilities at the State level and on the Eastern Shore. Hot spot analyses demonstrated that CAFO hot spots on the Eastern Shore were located in counties with some of the lowest wealth in the State, including the lowest ranking county—Somerset. Zero-inflated regression models demonstrated that increases in median household income across the state were associated with a 0.04-unit reduction in CAFOs. For every unit increase in the percentage of people of color (POC), there was a 0.02-unit increase in meat processing facilities across the state. The distribution of CAFOs and meat processing plants across Maryland may contribute to poor health outcomes in areas affected by such production, and contribute to health disparities and health inequity.Rural farming has changed drastically in the last fifty years, with the influx of corporate entities into the agriculture and livestock business and the decline of small-scale, family-owned farms becoming the norm [1]. The increased industrialization of agricultural operations has resulted in an overall decrease in the number of farms, but an increase in the size. For example, the number of farms in the United States fell from approximately 6.8 million in 1935 to 2.1 million in 2002, and the average farm size grew from 154.8 acres to 434 in the same time frame [2]. The number of animals raised at industrial production facilities increased by nearly 246% between 1982 and 2002 while the total number of livestock raised in the year 2000 was equal to that of the previous 80 years [2].The practice of concentrating farm animals into limited land spaces has led to the characterization of these facilities as animal feeding operations (AFOs) [3]. A facility is characterized as a concentrated animal feeding operation (CAFO) when it confines animals for at least 45 days within a year, no grass or other vegetation grows in the confinement area, and contains more than 100 animal units [4]. At CAFOs, animals are held throughout their lives in indoor stalls until they are transported to processing plants for slaughter.Due to the concentration of animals and the density of industrial farms, these operations pose threats to environmental and human health. For example, thousands of pounds of harmful emissions including volatile organic compounds (VOCs), ammonia, nitrogen, carbon dioxide, methane, particulate matter (PM), and heavy metals are emitted from CAFOs on an annual basis [5,6,7]. For these reasons, the U.S. Environmental Protection Agency (US EPA) has classified CAFOs as point source polluters under the Clean Water Act [8]. Consequently, CAFOs must receive permits through the Clean Water Act’s National Pollution Discharge Elimination System (NPDES) to control for manure, nutrient, and waste runoff into waters [9].These operations pose health problems for individuals who live nearby. For example, local residents face exposure to heavy metals, fertilizers, pesticides, and 355 million tons of animal waste emitted annually that humans may be exposed to via the air and runoff waste in groundwater and surface water [7,10,11]. Researchers found that distance from a CAFO or multiple CAFOs is a key to understanding weekly atmospheric ammonia levels [12] and the closer the populace was to the hog CAFO, the more intense the exposure [13].Additionally, chemicals found in heavily concentrated manure produced from CAFOs are responsible for 37% of national methane emissions and 65% of national nitrous oxide gases released annually [14]. PM10 accounts for one-third of the total emissions released from CAFOs, and excess amounts of arsenic emitted into water sources are linked to health issues like bladder, lung, skin, nasal, liver, kidney, and prostate cancer [15]. CAFOs generate excess nitrogen, above the federal standard, with chicken farms making up 60% of operations exceeding the acceptable nitrogen levels [2]. Overexposure to nitrates is associated with methemoglobinemia, hypertension, infant mortality, goiter, stomach cancer, thyroid disorder, cytogenetic defects, and birth defects [16].The poor air quality surrounding CAFOs differentially impacts nearby residents and on-site workers who have higher exposures than the general population in occupational settings [17]. High rates of respiratory and gastrointestinal problems, as well as irritation in the eyes and throat from airborne emissions, have been recorded by communities in close proximity to swine CAFOs, with the highest rate being experienced by employees of the operation [18]. More recent studies show that 25% of employees of CAFOs report respiratory problems thought to be caused by exposure to endotoxin, a family of Gram-negative bacteria membrane lipopolysaccharide fragments that are mainly found in environments that have high exposure to organic dusts, that are known to be present in high concentrations specifically in chicken CAFOs [19,20,21]. A study conducted in North Carolina found that children of CAFO workers are more than twice as likely to contract methicillin-resistant Staphylococcus aureus (MRSA) than children in households with no contact with these operations [22]. Similarly, the prevalence of Staphylococcus aureus and MRSA in CAFO workers has been found to be higher than in the general U.S. population [23].Animal agriculture has been shown to differentially burden low-income communities and communities of color [17,24,25,26]. A study in Mississippi found hog operations were located in areas with high percentages of low-income, African American residents [27]. North Carolina studies have found that low-income communities and communities of color were more likely to be located near a CAFO than their wealthier, White counterparts [20,28]. A 2015 study determined that communities with high percentages of Hispanic individuals are disproportionately burdened by CAFOs in Ohio [29]. Furthermore, various North Carolina studies have determined that vulnerable populations are disproportionately burdened by poor air quality and high ammonia levels as a result of close proximity to CAFOs [13,30].From 2011 to 2018, the number of CAFOs in the state of Maryland rose from 150 to 573 (a 282% increase), suggesting that it is a continuously growing industry. In the same time frame, there was a 46.5% increase in egg production in Maryland [31]. To this point, no study has examined whether or not there are environmental justice issues associated with chicken farming in the state of Maryland. This paper explores potential disparities in the distribution of chicken farms in Maryland along sociodemographic lines, and discusses the implications on human health and quality of life in the region.We obtained geocoded data for concentrated animal feeding operations (CAFOs) from the Maryland Department of Environment’s (MDE) AFO permit database, which provides both CAFO and AFO data for different animal feeding industries. The Topologically Integrated Geographic Encoding and Referencing (TIGER) file, a census tract boundary file, was downloaded from the MDE’s 2010 Maryland Census Tract boundary dataset [32]. This was then spatially joined with sociodemographic features from the 2018 Census Bureau American Community Survey (ACS) 5-year estimates data obtained from the Census [33]. The sociodemographics used for analyses were: (1) percentage people of color; (2) percentage under 18; (3) percentage below poverty; (4) percentage individuals without a high school (HS) diploma over age 25; (5) percentage below poverty who are under 18; (6) median household income; (7) per capita income; (8) percentage of home ownership; and (9) percentage of homes built before 1949. Housing pre-1949 was chosen to represent suburbanization based on historical trends in development [34]. Similar variables have been used in previous environmental justice research to represent race/ethnicity and socioeconomic status [18,20,27]. Due to the inconsistent reliability of ACS census data, we excluded any census tracts where the coefficient of variation was greater than 40, which the census bureau considers unreliable [35]. This reduced the total number of populated census tracts in the analysis from 1394 down to 1309 total census tracts. Additionally, data related to poverty, education, and housing pre-1949 were removed from the analysis due to its particular unreliability.We obtained the address and primary activity for all meat processing facilities in the United States from the US Department of Agriculture’s (USDA) Meat, Poultry, Egg Inspection Directory [36]. The address of each facility was geocoded and then mapped as individual points in ArcGIS.The first stage of analysis required identifying CAFO host and non-CAFO host census tracts. A “CAFO host tract” was defined as a census tract that contained 1 or more CAFOs, whereas a “non-CAFO host tract” was defined as a census tract that did not contain a CAFO within its boundaries. The same definitions of host and non-host census tracts were used for meat processing facilities. These census tract categorizations were then stratified by majority “rural” or “urban” designations based upon urban and rural housing unit data from the 2010 Census. Census tracts with more than 50% urban housing units were considered “urban” and those with more than 50% rural housing units were designated as “rural”. ArcGIS 10.7 (ESRI, Redlands, CA, USA) was used to find these census tract classifications, and these were then analyzed alongside sociodemographic statistics to draw correlations between presence/distance from CAFO and meat processing sites versus socioeconomic status and race/ethnicity.A hotspot analysis using the “Optimized Hot Spot Analysis” test was performed on CAFOs and meat processing facilities in Maryland in order to find areas with large concentrations of CAFOs and meat processing plants. For CAFOs a distance band of 25,284.14 m was calculated, while one of 12,256 m was calculated for meat processing facilities. Because of the concentration of CAFOs on the Eastern Shore, a secondary “Optimized Hot Spot Analysis” test was used to further investigate CAFO concentrations in this region. The distance band used for this analysis was 19,887.27 m. The average sociodemographics of areas determined to be CAFO/meat processing hotspots or cold spots were found in order to understand the primary population characteristics impacted by the animal agriculture industry. A comparison between the mean differences between both CAFO and meat processing hotspots was completed using the study variables.To build upon the results from the hotspot analysis and identify statistically significant relationships, statistical tests were run to further understand the relationships between the chicken industry and socioeconomic indicators: Exploratory and Ordinary Least Squares (OLS) regression were performed to determine if there was a statistically significant relationship between sociodemographic variables in census tracts versus the presence of CAFOs and related sites using ArcGIS. Additionally, multivariate regression tests were performed in PAST 4.03 (University of Oslo, Oslo, Norway) to determine if there is a relationship between the presence of the chosen demographic, income, and education variables versus the number of CAFOs/meat processing facilities. However, due to the number of census tracts that contained zero CAFOs or meat processing facilities, a zero-inflation regression model was used instead. This was performed because the large frequency of zero CAFO/meat processing facility census tracts skewed the model, causing it to not be normally distributed due to the excessive number of zeroes. A separate zero-inflation regression was also used to assess the relationship between the sociodemographics and CAFOs on the Eastern Shore, as this area was identified as a chicken CAFO hot spot. The zero-inflation model was performed in RStudio (R Consortium, Boston, MA, USA). In order for the test to be completed, the median household income variable was scaled down by 1000. Finally, the regressions were offset using total population per census tract in order to adjust for the differences in population between urban and rural areas.Table 1 shows that census tracts that host CAFOs in more urban areas feature a 27.5% POC population vs. those in rural areas (15.3% POC). Additionally, these census tracts tended to have lower median household incomes and levels of homeownership in comparison to rural areas with CAFOs. However, there were 35 rural census tracts that hosted CAFOs, while there were only 12 urban census tracts that were hosts (See Figure 1).In Figure 2, the primary hotspots of CAFOs are centered around the Eastern Shore and the Delaware border, whereas the CAFO coldspots are located in the Baltimore–Washington Metropolitan area. The meat processing facility hot spots are generally more dispersed throughout the state, with a larger concentration of statistically significant hotspots occurring in and around Baltimore. However, due to the distribution of meat processing facilities throughout the state, there are no cold spots of meat processing facilities. When looking at the Eastern Shore of Maryland exclusively, the southern part of the Eastern Shore, around Dorchester and Wicomico Counties, is identified as a concentration of CAFO hotspots as seen in Figure 3. The hotspot classifications were used in Table 1 and Table 2 as categories for highlighting the differences in means between hot spots and cold spots of CAFOs and meat processing facilities.Figure 1 shows the location of chicken CAFOs and federal meat processing plants in the state of Maryland. Chicken CAFOs are shown to be concentrated on the Eastern Shore of the state, with very few outliers in other regions. meat processing plants, conversely, are primarily located in more urban areas, specifically in and around the city of Baltimore. However, there are other meat processing plants located throughout the western part of Maryland. There is very little overlap between hotspots for chicken CAFOs and the location of meat processing plants.Figure 4 demonstrates the locations of CAFOs and their colocation to populations of people as stratified by median household income. CAFOs are abundant on the Eastern Shore of Maryland, which also has many areas with a large percentage of people with lower median household incomes.The largest concentrations of people of color tend to be in the Baltimore area which showed a spatial relationship to the presence of CAFO coldspots (Figure 5). On the other hand, there is a comparatively smaller population of people of color on the Eastern Shore where large CAFO hotspots exist (Table 2). However, the percentage of people with low median household incomes tended to be high in both the Baltimore and the Eastern Shore areas. When comparing this to the CAFO and meat processing facility locations in Figure 1, there appears to be a tendency for CAFOs and meat processing facilities to be located in areas with low median household incomes (see Figure 5).With regards to race/ethnicity, meat processing facilities appear to be located in areas of higher POC, in contrast to the distribution of CAFOs which appears to be in smaller POC percentage census tracts. The percentage of homeownership, as opposed to rented homes, is slightly lower in CAFO hotspots than in CAFO coldspots (Table 2).However, the median household income of census tracts in these hotspots was inversely related to the presence of a CAFO hotspot, with the highest median household income areas being statistically significant coldspots or statistically insignificant census tracts, which tended to be urban or suburban areas. The coldspots, however, included the areas around Baltimore which had much higher than average median household incomes than inside Baltimore.Table 3 displays the results of the zero-inflated Poisson regression for chicken CAFOs. This analysis produced a number of statistically significant results. Increases in median household income across the state were associated with a 0.04-unit reduction in CAFOs. As the percentage of homeownership increased, there was a 0.05-unit increase in CAFOs.Table 4 displays the results of the zero-inflated Poisson regression for meat processing facilities. This analysis shows that for every unit increase in the percentage of POC, there is a 0.02-unit increase in meat processing facilities across the state.Based upon the results of the hotspot analysis that demonstrated statistically significant clustering of chicken CAFOs in Maryland’s Eastern Shore, a zero-inflated Poisson regression was used to assess relationships between sociodemographics and the presence of CAFOs in the region. Prevalence ratios were also used to determine if the prevalence of CAFOs was modified by sociodemographics. The zero-inflated Poisson regression model shows that for every unit increase in median household income, there is a 0.04-unit reduction in CAFOs on the Eastern Shore (see Table 5). Increases in % homeownership were associated with a 0.05-unit increase in CAFOs on the Eastern Shore.The goal of these analyses was to examine the relationship between sociodemographic characteristics and presence to chicken CAFOs and meat processing plants in the state of Maryland, specifically assessing communities of color and low-income communities to determine if they are disproportionately burdened by industrial chicken farming and processing plants. Results from the zero-inflated Poisson regression model (Table 3) show a positive relationship between POC and chicken CAFOs statewide, yet this association is not statistically significant at a p-value of 0.05. This finding is also consistent with work performed in similar studies, which demonstrate a higher percentage of African American residents in communities located near CAFOs [20,26,27,28]. We found that the largest clusters of CAFOs are located on the Eastern Shore. Table 2 demonstrates that 25.2% of the population in CAFO hotspots are POC. Additionally, this region features some of the lowest median household incomes in the state (Figure 4). This is a significant finding because socioeconomic status appears to have a strong correlation with the location of chicken CAFOs, so much so that economically depressed areas seem to be targets for CAFO operations [37]. The trend held true beyond the state level and extended to Maryland’s CAFO hot spot on the Eastern Shore. Zero-inflated Poisson regression restricted only to the Eastern Shore also demonstrated that low median household incomes were associated with chicken CAFOs. This is detrimental to the wellbeing of residents living within close proximity to CAFOs due to the health implications such as respiratory diseases and irritation of the eyes, nose, and throat [20,27]. Manure, antibiotics, and heavy metals are all produced by these animal farms and have the potential to seep into nearby water sources or contribute to pollution in the area [7].Statistical analyses demonstrated that chicken CAFOs are unequally distributed throughout the state, with a disproportionate amount located near low-income communities. This finding is consistent with work performed in previous studies, which determined that CAFOs in North Carolina and Mississippi have the tendency to be located near low-income communities [27]. On the Eastern Shore, Bernhardt et al. (2015) found that there were 438 chicken operations in Caroline, Dorchester, Somerset, Wicomico, and Worcester counties in 2013, totaling 243,891,955 animals [38]. The high concentration of CAFOs in the region is an environmental justice concern, given these five counties’ SES. All five fall below the state’s 2014 median household of USD 74,149 (Caroline—USD 55,605, Dorchester—USD 45,628, Somerset—USD 36,716, Wicomico—USD 52,301, and Worcester—USD 58,820). All five are also above the state average of people below the poverty level (10%)—15.3% in Caroline, 16.9% in Dorchester, 23.7% in Somerset, 17.4% in Wicomico, and 11.1% in Worcester. Only Somerset, however, has a higher African American population (42.3%) and a lower White population (53.5%) than the State (29.4% Black and 58.2% White, respectively) [39]. This study fills an important gap in the literature with respect to assessing environmental justice concerns related to Maryland’s chicken industry.Studies have shown that CAFOs are usually developed in existing communities of color and low-income communities, instead of the CAFO attracting these populations [26]. Individuals that have the financial means to leave the area following the development of a CAFO usually do, leaving residents of low SES to suffer the environmental and public health consequences of living near a CAFO [40].This analysis showed a positive relationship between the percentage of homeownership and the presence of CAFOs, indicating that homeownership may not be an accurate indicator of socioeconomic status, but rather an indicator of rurality or urbanity. Homeownership is more common in rural areas [41], which would help explain why homeownership is associated with an increased presence of CAFOs, which primarily affect rural populations [42].In contrast to CAFOs, meat-packing plants were shown to be distributed more evenly across the state. Numerous plants were located in Baltimore City (Figure 3). Air quality in Baltimore City has already caused concern regarding residents’ health and well-being [43], and the high concentration of meat-packing plants and increasing industrialization have the potential to contribute to the declining public health of the city [44]. Statistical analysis from Table 4 demonstrates comparable results to Table 2, showing that meat processing facilities, like chicken CAFOs, are disproportionately located near low-income communities of color. This is consistent with studies performed in other areas of the United States, specifically the Midwest, which determined areas containing meat processing facilities to have high percentages of immigrants and low-income families [44].Table 4 shows higher percentages of homeownership associated with meat processing facilities, similar to CAFOs. As stated previously, this distinction is consistent with studies showing higher rates of homeownership in rural areas [41]. Although there are more meat processing facilities in urban areas than there are CAFOs, there are still enough in areas of high homeownership to create an overall positive correlation between homeownership and meat processing facilities.Existing research regarding demographics surrounding meat processing plants shows an overwhelming correlation between the location of meat processing plants and areas with high immigrant populations [44,45,46,47]. This correlation is consistent with the analysis shown in Table 3, indicating increases in POC associated with a unit increase in meat processing facilities, as 76.5% of immigrants in the state of Maryland are POC [39]. Additionally, in the state of Maryland meat processing facilities are concentrated in Washington, DC and Baltimore City which both have immigrant populations that make up approximately 13–15% of the total population [48]. Future research regarding meat-packing facilities in the state should investigate the specific meat-packing plants that exist in Baltimore City and how their presence in and around the city impacts the health of residents.This study has several limitations. Many of the demographic variables are proxies to try to assess urban/suburban/rural relationships, and therefore cannot offer more precise relationships. For example, the large percentage of houses built before 1949 in Baltimore and in the Eastern Shore is due to these areas’ histories as urban and agricultural centers, respectively. The mass suburbanization of America occurred after World War II, meaning the majority of suburban homes were built following 1949 [34]. Suburbanization has also led to larger concentrations of POC living in the inner cities, and despite a large growth of POC moving to the suburbs, large areas of suburbia remain majority White [49]. This proxy measure cannot fully capture the dynamics related to suburbanization, and therefore is a limitation in the results.There were several census tracts that had to be disqualified from this study due to unreliable ACS data. Because of this, it is difficult to assess how these tracts may affect the overall results for the state of Maryland. Additionally, there were several variables that similarly were excluded from the analyses due to unreliable ACS data due to high coefficients of variation. Because of this, it was impossible to measure associations between the presence of CAFOs, meatpacking facilities, and poverty—which could illuminate the discussion regarding the association between socioeconomic status and the presence of industrial chicken farming.Performing these analyses at a lower level of analyses could have produced more precise effects. Other investigations of environmental justice variables in relation to CAFOs have been conducted at the block group level instead of the census tract level [20,27]. Aggregation of data at a higher level of analysis produces more heterogeneous data, which can mask effects occurring on a smaller scale [50].The results of this investigation have implications for both policymaking and future research studies. Because this work has demonstrated that CAFOs and meat processing facilities disproportionately burden areas with higher concentrations of POC and low socioeconomic status communities in Maryland, more policy provisions should be implemented to protect these populations. Public comment periods where citizens can give input on a federal level are a way to temporarily deal with problems due to CAFOs [19,51]. In addition, the Maryland Department of the Environment requires the inclusion of the distance to and water quality status of the nearest body of water and potential outdoor air quality risks in the Comprehensive Nutrient Management Plan (CNMP), which is necessary to get a permit to build a CAFO [51]. While MDE does have several recommended stopgaps in place regarding the establishment of new CAFOs, this study suggests that Maryland’s chicken industry policy needs to employ an environmental justice lens. Environmental justice-oriented solutions include policy that promotes community empowerment; building infrastructure that supports sustainable communities; enhancing community-based pollution prevention strategies; and creating community-based sustainable development [52]. Essentially, policy solutions should be holistic in their approach to providing protection for environmental justice communities through initiatives that empower communities rather than merely being prescriptive. Policy that adopts the environmental justice framework also focuses on prevention as opposed to treatment [52]. Therefore, a public health approach to policy that prioritizes the prevention of exposure to the deleterious byproducts of CAFOs and meat processing plants is essential to the regulation of these industries.This study produced notable results regarding the unequal distribution of CAFOs and meat processing plants amongst communities of color and low socioeconomic status in Maryland. Other research should build upon these findings using cohort studies in order to further analyze the relationships between environmental justice communities and hazards associated with the chicken industry in Maryland. The present study relied on an ecological design. Future investigations should examine the presence of contaminants associated with the chicken industry, such as ammonia or nitrate, in environmental justice communities located near such sites in Maryland in order to measure actual exposure levels [12,53]. Previous research that could provide guidance to future work in this area includes Mirabelli’s (2007) cohort designs that used questionnaires to ask participants about their exposure history to agents that may exacerbate asthma [54]. Other research on agricultural contaminants has included biological sampling from a cohort of livestock workers to demonstrate exposure to antibiotic-resistant Staphylococcus aureus [55]. A biological sampling of nearby water supplies or residents’ homes and properties could provide a better understanding of exposures related to industrial chicken farming.A dearth of pertinent data proved to be a challenge to the present investigation. Passing the Maryland Community Healthy Air Act would facilitate better data collection and further analysis. The proposed legislation creates a protocol for sampling, quantifying, and reporting various data related to large animal-feeding operations across Maryland [56]. These data—which include ammonia, particulate matter, and VOCs—would provide a better understanding of the types and amount of exposure that CAFO host communities experience, providing a stronger basis for policy and advocacy around environmental health and environmental justice.As two of the leading food production businesses in the state [57], poultry CAFOs and meat processing facilities in Maryland must acknowledge the threats they pose to environmental and public health. Low socioeconomic status communities and communities of color are disproportionately burdened by chicken CAFOs and meat processing facilities across Maryland, making the state’s chicken industry an environmental justice concern. This study has produced results that suggest a need for further research into the lasting effects of industrial agriculture on community health. Efforts should be made by the state to reduce these disparate effects of chicken CAFOs and meat processing facilities through regulation and by providing protection for the communities that live and work in affected areas. Additionally, consistent data collection and reporting on environmental factors such as water and air quality is essential to monitoring and ensuring the environmental health of these areas. Future studies should continue to investigate Maryland’s chicken industry with respect to environmental justice concerns in order to provide further basis for policy to improve the environmental public health of affected communities throughout Maryland.Conceptualization, S.W., K.B.-N. and R.M.; methodology, J.H., L.K. and C.J., K.B.-N. and R.M.; software, J.H.; resources, N.K., I.B. and C.E.; data curation, J.H., L.K. and C.J.; writing—original draft preparation, J.G., N.K., I.B. and C.E.; writing—review and editing, J.G.; visualization, J.H.; supervision, J.G.; project administration, S.W.; All authors have read and agreed to the published version of the manuscript.This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.Not applicable.Not applicable.Data for these analyses are available from the Maryland Department of Environment’s (MDE) AFO permit database, the Terrestrial Initiative in Global Environmental Research (TIGER), and the 2018 American Community Survey from the Census.We would like to thank Daniel Polsky and other student researchers from the Center for Community Engagement, Environmental Justice, and Health for their contributions to this manuscript.The authors declare no conflict of interest.Locations of CAFOs and Federal Meat Processing Facilities in Maryland.Optimized Hot Spot Analysis for Chicken CAFOs in Maryland.Optimized Hotspot Analysis for Chicken CAFOs on the Eastern Shore of Maryland.Chicken CAFOs by Quintiles of % Population of Median Household Income.Chicken CAFOs by Quintiles of % Population POC.Distribution of Features/Sociodemographic Features, CAFO Host vs. Non-Host.# of = “Number of”.CAFO and Meat Processing Hot/Coldspots vs. MD Sociodemographics.Zero-inflated Poisson Regression Modeling for MD Chicken CAFOs.Signif. codes: “***” p < 0.001.Zero-inflated Poisson Regression Modeling for MD Meat Processing Facilities.Signif. codes: “***” p < 0.001, “**” p < 0.01.Zero-inflated Poisson Regression Modeling for MD Chicken CAFOs in the Eastern Shore.Signif. codes: “***” p < 0.001.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11040.txt

@@ -0,0 +1,47 @@
+The purpose of our study was to investigate which organizational levels and factors determine the number of resident handlings in eldercare. We conducted a multi-level study, stratified on day and evening shifts, including information on four levels: nursing homes (n = 20), wards within nursing homes (day, n = 120; evening, n = 107), eldercare workers within wards (day, n = 619; evening, n = 382), and within eldercare workers (i.e., days within eldercare workers; day, n = 5572; evening, n = 2373). We evaluated the influence of each level on the number of resident handlings using variance components analysis and multivariate generalized linear mixed models. All four levels contributed to the total variance in resident handlings during day and evening shifts, with 13%/20% at “nursing homes”, 21%/33% at “wards within nursing homes”, 25%/31% at “elder-care workers within wards”, and 41%/16% “within eldercare workers”, respectively. The percentage of residents with a higher need for physical assistance, number of residents per shift, occupational position (only within day shifts), and working hours per week (only within day shifts) were significantly associated with the number of resident handlings performed per shift. Interventions aiming to modify number of resident handlings in eldercare ought to target all levels of the eldercare organization.Musculoskeletal disorders (MSDs) are highly prevalent among the working-age population [1,2,3] and impose significant costs for the individual, workplaces, and society [4,5]. Eldercare workers represent a highly vulnerable working group for MSDs, where annual prevalence of low back pain and neck/shoulder pain is between 51%–71% and 31%–52%, respectively [6,7]. Additionally, eldercare workers have high rates of sickness absence and early retirement, predominantly attributed to MSD [8,9].Resident handling activities comprise one of the core tasks of eldercare work [10]. These tasks include lifting, repositioning, turning, help with support stockings, pushing and pulling residents in portable chairs, and kneeling. Previous studies conducted in healthcare settings have found that the number of resident handling activities per work shift is between 7.5 to 9.5 times for repositioning and between 2.8 and 20 times for transferring [11,12,13]. These resident handling tasks are documented to increase the workers’ risk of MSD [14,15] and sickness absence [8].Individual physical and psychosocial factors are associated with the pathogenesis of MSD among eldercare workers [16,17,18,19,20,21]. Other factors are related to the policies and available ergonomic aids for resident handling activities at the workplace (i.e., the organization level), or the staff ratio and the number of residents with a high need for resident handling activities who are being handled within a ward (i.e., a ward level organizational factor) [22,23].Previous studies examining resident handling activities have usually been conducted with a one-level exposure approach (i.e., they do not consider the influence of higher organizational levels such as the ward or nursing home) [15,24,25,26,27,28,29]. Thus, eldercare organizations have limited evidence about which nursing home levels and factors need to be targeted in order to reduce the number of resident handlings performed per shift. Taking into account the complexity of the simultaneous influence of the organizational and individual factors in a multi-level approach is likely to provide a more holistic understanding to reduce the number of resident handlings.Furthermore, the vast majority of studies investigating resident handling activities are primarily based on self-reported information [11,12,13,15,25,26], which can lead to imprecise and biased results. Only a few studies have used direct observations as their core methodology [30,31] and with assessments either conducted for three 2-hour periods [32] or among a limited number of participants [30]. A more comprehensive quantification of resident handlings utilizing observational exposure assessment tools may increase the evidence of factors associated with number of resident handlings.Additionally, the majority of the studies assessing resident handling tasks among healthcare personnel were performed during day shifts only [30,31], hence not accounting for the possible differences between shifts. Research in the eldercare setting has demonstrated that there are notable differences between day and evening shifts [33]. Furthermore, day-to-day variation of the number of resident handlings performed during day and evening shifts has not yet been explored. Thus, it is highly important to investigate which specific nursing home levels and factors influence the number of resident handlings during day and evening shifts. Furthermore, possible differences may exist between days in eldercare, making it important to study the day-to-day variation of factors within eldercare work.To the best of our knowledge, no studies have explored the influence and extent of organizational levels and factors on the number of resident handlings in eldercare using a multi-level methodology and direct observations. The purpose of this study was to investigate which organizational levels and factors determine the number of resident handlings in eldercare and to evaluate the influence of each level of the nursing home on the number of resident handlings performed during day and evening shifts. This is a cross-sectional multi-level study utilizing baseline data from the Danish Observational Study of Eldercare work and musculoskeletal disorders (DOSES). A detailed description of the study design, data collection, and methodology has previously been published [34]. DOSES received ethical approval from the Danish Data Protection Agency and the Ethics Committee for the Capital Region of Denmark (H-4-2013-028). In total, 83 nursing homes located in the administrative Region Zealand, in the Greater Copenhagen area in Denmark, were invited to participate in DOSES. Twenty nursing homes (18 municipal and 2 private), including 126 wards and 941 eligible eldercare workers, agreed to participate. Eligibility criteria for participation in DOSES were: aged between 18 and 65 years old, employed more than 15 hours per week, working on day, evening, or rotational shifts, and allocation of at least 25% of their working time on activities related to direct resident care. Individuals that were on long-term sickness absence, pregnant, not permanently employed, or not spending a minimum of 25% of their working time on tasks related to the direct care of the elderly were excluded from the study. In order to maintain as much data as possible and minimize selection bias, the only criterion for the eldercare workers to be included in the present study analysis was filling in the work schedules at baseline providing information on which residents they had been taking care of over a three-week period. Thus, the final study population for the day shifts consisted of 20 nursing homes, 120 wards, 619 eldercare workers, and 5572 registered days. For the evening shifts, the final study population consisted of 20 nursing homes, 107 wards, 382 eldercare workers, and 2373 registered days (Figure 1).Baseline data collection at the 20 nursing homes were conducted from September 2013 to December 2014. Data from each participating nursing home were collected over a period of 1 to 2 weeks. Organizational factors were collected at the ward level with a web-based questionnaire to the team managers. Eldercare workers’ factors were collected at the eldercare worker level with a structured self-administered questionnaire. Finally, information about the number of resident handlings performed per shift were collected through real-time workplace observations by trained observers. We linked data between the different levels by identification numbers for nursing home, ward, eldercare worker and day (date), which was registered for each outcome “Number of resident handlings performed per shift”.All organizational factors were collected at ward level. At baseline, team managers of each ward (41 team managers administered the 120 wards in the day shifts and 40 team managers administered the 107 wards in the evening shifts) answered a web-based questionnaire. Information on the number of residents located at the ward and the number of staff normally present at the ward in a shift, were used to calculate staffing ratio (total number residents divided by eldercare workers working in a shift) for day and evening shifts separately. Furthermore, information on the residents’ need for physical assistance (light, moderate, comprehensive, or complete) were provided by the team managers on standardized lists. We calculated the percentage of residents with high needs for physical assistance (i.e., comprehensive or complete) for each ward. During meetings with upper-and team-managers, we collected information regarding the type (somatic, dementia, rehabilitation, or independent living) of each ward. The type of ward was dichotomized (somatic versus dementia/other).We collected information on age, sex, occupational position (i.e., social service helper, social service assistant or other), and working hours mainly through standardized lists provided by the management. If the information were missing at the lists, we collected it, as with information on seniority, through a structured questionnaire provided to the eldercare workers at their workplace. At the “within eldercare workers” level (i.e., days within eldercare workers), the number of residents each eldercare worker were responsible for providing care per shift were drawn from work schedules (i.e., allocation of the residents between eldercare workers) collected over a three-week period at baseline.In DOSES, more than 4700 real-time workplace observations were conducted at baseline during a 4-hour period for the day shifts and during a 5-hour period for the evening shifts. The observations contained information on all caring activities needed for each resident at the 20 nursing homes, including information on the number and types of resident handlings. Types of resident handling activities included lifting, repositioning, turning, help with support stockings, push and pull resident in portable chair, and kneeling. The observations were conducted by trained observers following a strict protocol [35]. The observations were performed using tablets with the software Noldus Observer XT pocket observer (Noldus, Wageningen, The Netherlands). The inter-rater reliability of the observation instrument was shown to be good with corresponding agreement coefficients ranging from −0.27 to 1.0 [35].Observations of the resident handlings required for each respective resident were merged with the information from work schedules, stating which residents each eldercare worker were responsible of providing care to each shift. We calculated the total number of resident handlings per shift performed by each eldercare worker as the sum of needed handlings of the residents they were responsible for.We used variance components analysis (VCA) to calculate the proportion of variance in the number of resident handlings performed per shift explained by the levels: nursing homes, wards within the nursing homes, eldercare workers within the wards and within eldercare workers (i.e., day-to-day variation). Nursing home, ward, and eldercare worker, were entered into the model as hierarchical random effects. Table 1 and Figure 1 illustrate and describe the four levels in the VCA. The analysis was stratified for day and evening shifts.The proportion of variance is presented as “percentage of variability” and is calculated from the Between Group Variance (BGV) for each level and the Total variance. The intra-class correlation coefficient (ICC) is a descriptive statistic that quantitatively describes the proportion of variance that is explained by a grouping factor in multi-level / hierarchical data [36]:
+          (1)
+          
+            
+              
+                Percentage
+                 
+                of
+                 
+                variability
+                =
+                ICC
+                ∗
+                100
+                =
+                
+                  
+                    
+                      Between
+                       
+                      Group
+                       
+                      Variance
+                       
+                    
+                    
+                      (
+                      
+                        BGV
+                      
+                      )
+                    
+                  
+                  
+                    
+                      Total
+                       
+                      variance
+                    
+                  
+                
+                ∗
+                100
+              
+            
+          
+        To identify factors associated with the number of resident handlings performed per shift for day and evening shifts within each level of the nursing homes, we used the following two-step procedure. First, univariate analyses were conducted with each potential determinant added individually as a fixed-effect to the developed VCA model (i.e., retaining nursing home, wards, eldercare worker, and within eldercare workers as hierarchical random-effects). All variables with a p-value less than 0.10 were selected for further investigation. Second, we constructed a multivariate model by forward selection starting with the variables with the lowest p-value. Variables with a p-value less than 0.10 and resulting in the lowest -2loglikelihood were retained in the final model. We set the significance level for variable selection at 0.10 so that critical variables relevant to the outcome and of practical and / or clinical relevance are not omitted.The extent of missing data differed across each variable and to keep most data in the analysis we analyzed the organizational, eldercare worker, and within eldercare worker factors with a different sample size for every variable (i.e., age, day = 5381, evening = 2309; sex, day = 5417, evening = 2310; seniority, day = 4889, evening = 1860; working hours per week, day = 5326, evening = 2274). However, the extent of missing data was not substantial.To identify the percentage reduction of different sources of variance to the total variability in the resident handlings per shift for day and evening shifts in eldercare work across all levels, the variables that were found statistically significant in the multivariate model were individually introduced to the raw VCA model. The percentage reduction across all levels was calculated as the difference of the estimate of the between group variance of the raw VCA model subtracted by the estimate of the between group variance of the raw VCA model including the individual statistically significant factors.For descriptive statistics, SPSS (IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp.) was used. All other analyses were conducted using the procedure PROC MIXED in SAS version 9.4 software (SAS Institute, Cary, NC, USA).In total, 619 eldercare workers within 120 wards from day shifts and 382 eldercare workers within 107 wards from evening shifts filled in work schedules. The eldercare workers working in day shifts registered on average 9 days of information, giving a total of 5572 days of registration. Accordingly, the eldercare workers working in evening shifts registered on average 6 days of information, giving a total of 2373 days. As the analysis of this study is stratified on day and evening shifts, it is only possible for the eldercare workers to have one registration each day. This means that the number of days registered by the eldercare workers and the corresponding measured day-to-day variation is equivalent to the number of shifts registered by eldercare workers and shift-to-shift variation. We provide an overview of the study participants and number of measured days, their hierarchical structure, and clustering in Table 1.Within the 20 participating nursing homes, the majority of the wards were somatic (day shifts 75%; evening shifts 75%) and the rest were dementia/other (day shifts 25%; evening shifts 25%). The average (SD) staff ratio (residents per worker) was 3.4 (0.9) for the day shifts and 7.5 (2.6) for the evening shifts. The eldercare workers were generally middle aged (day shifts’ mean [SD] = 44.4 [10.8]; evening shifts’ mean [SD] = 47.3 [11.0]), predominately female (day shift 95%; evening shift 94%) and most were employed as social service helpers (day shifts 42%; evening shifts 47%). The average (SD) job seniority (in years) was 15.1 (11.1) for the day shifts and 16.1 (11.2) for the evening shifts. The average (SD) working hours per week was 33.1 (3.2) for the day shifts and 29.5 (3.4) for the evening shifts. Table 2 illustrates the characteristics of the organizational factors on ward level and eldercare workers stratified by day and evening shifts.All four levels within the nursing homes contributed to the total variance of the number of resident handlings performed per shift during day and evening shifts (see Table 3). We found clear differences in the percentage of variance between day and evening shifts. The greatest source of variance occurred within eldercare workers (day-to-day variation; 41.4%) for the day shifts and between wards within nursing homes (33.2%) for the evening shifts. The rest of the variance for the day shifts occurred, in descending order, between eldercare workers within wards (25.0%), between wards within nursing homes (20.9%), and between nursing homes (12.7%). Accordingly, the rest of the variance for the evening shifts occurred, in descending order, between eldercare workers within wards (31.3%), between nursing homes (19.9%), and within eldercare workers (day-to-day variation; 15.7%). For the organizational factors, an increased percentage of residents with a higher need for physical assistance was associated with a higher number of residents handlings for both day (β = 0.07; 95% CI 0.04 ; 0.10; p = <0.0001) and evening (β = 0.09; 95% CI 0.05 ; 0.13; p = <0.0001) shifts. Regarding the eldercare workers factors, being a social service assistant was associated with lower resident handlings only for day shifts (β = −0.65; 95% CI −1.24 ; −0.06; p = <0.05), when compared with being a social service helper. Furthermore, a higher number of working hours per week was positively associated with an increased number of resident handlings for day shifts only (β = 0.10; 95% CI 0.01 ; 0.19; p = <0.05). Finally, for the within eldercare workers factors, the number of residents per shift was a significant predictor for the number of resident handlings for both day (β = 1.47; 95% CI 1.37 ; 1.57; p = <0.0001) and evening shifts (β = 0.82; 95% CI 0.76 ; 0.87; p = <0.0001). Table 4.For the organizational factors, when the percentage of residents with higher need for physical assistance determinant was added in the crude VCA model, a greater percentage reduction in the total variance in resident handling per shift was observed between nursing homes for the day shifts (44%) and evening shifts (62%). For the eldercare worker factors, when the occupational position determinant was introduced to the crude VCA model, a percentage reduction in the total variance in resident handlings per shift was observed between nursing homes (6%), between wards (1%), and between eldercare workers (3%) for day shifts only. In contrast, increased total variance of resident handlings was observed between nursing homes (−3%), between wards (−7%), and between eldercare workers (−1%) for day shifts only, when the working hour per week determinant was added to the crude VCA model. Finally, for the within eldercare workers factors, the greater percentage of the total variance in resident handlings per shift occurred between the eldercare workers (15%) for the day shifts, and between nursing homes (43%) for the evening shifts. Table 5.This cross-sectional multi-level study identified that all levels of the nursing homes (i.e., nursing homes, wards within nursing homes, eldercare workers within wards, and within eldercare workers) contributed to the total variance in the number of resident handlings performed per shift for day and evening shifts. Furthermore, we found that an increased percentage of residents with a higher need for physical assistance (both day and evening shifts), being a social service helper (day shifts only), a higher number of working hours per week (day shifts only), and an increased number of residents per shift (both day and evening shifts) were significantly associated with a higher number of resident handlings performed per shift.To the best of our knowledge, this is the first comprehensive multi-level study to investigate to what extent the number of resident handlings performed per shift is attributed to four hierarchical levels of nursing homes (i.e., nursing home, wards within nursing homes, eldercare workers within wards, and within eldercare workers). No previous studies have demonstrated that all four investigated levels of nursing homes have contributed to the total variance in the number of resident handlings performed during day and evening shifts.The main result is that during day shifts, the highest variance in the number of resident handlings was observed within and between eldercare workers. There are only few studies using multi-level methodology to investigate how different hierarchical levels such as nursing home, ward, and eldercare workers, and registered days may influence the number of resident handlings. However, in accordance with our results, a previous study by Koppelaar et al. (2012) found that across all types of resident handlings (i.e., transfers, personal care of patients, repositioning, putting on or taking off anti-embolism socks) the greatest source of mechanical load was observed within nurses’ factors during day shifts only [37]. The organizations and the wards within them had limited contribution to the total variance in the mechanical load across all resident handling activities. This finding suggests that within and between eldercare worker factors are the main drivers to the number of resident handlings performed in nursing homes.Although nursing homes in Denmark are under the governance of the assigned municipality, they present a high degree of independence, which could explain some of the variance in the number of resident handlings found between nursing homes during both day and evening shifts. However, Danish nursing homes have to follow standardized protocols and eldercare workers have to comply with certain principles and guidelines, regulated by the municipalities. This could explain the low level of variance in the number of resident handlings contributed by the nursing homes.On the other hand, for the evening shifts, we found that the greatest source of variance in the number of resident handlings was observed between wards within nursing homes and between eldercare workers within wards. The variance could be due to the decreased number of eldercare workers during evening shifts. The observed differences in the variance in the number of resident handlings performed per shift during day and evening shifts are important, as it suggests that the number of resident handlings performed during day and evening shifts are highly driven by the different hierarchical levels of nursing homes.Several determinants across all hierarchical levels of nursing homes were significantly associated with the number of residents handlings performed per shift. An increased percentage of residents with high demands for physical assistance (day and evening shifts), a higher number of working hours per week (day shifts only), and a higher number of residents per shift (day and evening shifts) were positively associated with an increased number of resident handlings performed per shift. Having an occupational position as a social service assistant compared with a social service helper was associated with less resident handlings per shift (day shifts only). For some of the investigated factors, our results are in accordance with previous studies. Koppelaar et al. (2012) found that a higher ratio of nurses per patient was associated with a decreased frequency of resident handling activities, such as manual lifting of patients, personal care of patients, patient transfers, and putting on and taking off anti-embolism stockings [37]. However, in that study, the determinants were investigated only for day shifts, compared with our study, which included both day and evening shifts.The finding that residents with lower physical functional level are associated with more resident handling activities reflects that the demands of the eldercare occupation are closely related to the need for care among the residents. Engaging the resident during resident handling activities will potentially promote their mobility and decrease the need for physical assistance, thus reducing the physical demands on the eldercare workers during a shift [38].Our finding that a higher number of residents per shift is associated with an increased number of resident handlings is highly intuitive. Increasing the staff ratio during a shift will potentially decrease the number of resident handlings performed per shift. The major strengths of this study are the large sample size and the multi-level design, which provides important insights into understanding the levels that contribute to the number of resident handlings performed per shift in eldercare work. Data on organizational and eldercare worker factors were collected from independent sources on each level and not with the use of questionnaires solely to eldercare workers. Furthermore, data about the number of resident handlings per shift were collected through real-time observations minimizing recall and measurement bias. However, a major limitation is the cross-sectional design, which prohibits attribution of causality to any detected association.This study has identified multiple factors across different levels of nursing homes associated with the number of resident handlings in eldercare work. Some of these factors (i.e., percentage of residents with a higher need for physical assistance) are not easy to change due to their nature. However, targeting modifiable factors such as reducing the number of residents assigned to each eldercare worker per shift or allocating the residents better between the eldercare workers can potentially reduce the fraction of eldercare workers with very high number of resident handlings per shift. Our results may be used to prevent excessive workload, pain, and sickness absence related to a high number of resident handlings performed per shift among eldercare workers. In the present study, all levels of nursing homes (i.e., nursing homes, wards within nursing homes, eldercare workers within wards, and within eldercare workers) contributed to the total variance in resident handlings during day and evening shifts. Furthermore, a higher number of resident handlings was significantly associated with the percentage of residents with higher need for physical assistance, the number of residents per shift, the occupational position (only within day shifts), and working hours per week (only within day shifts). These findings can be leveraged in order to reduce the number of resident handlings per shift by increasing the staff ratio, optimal workforce planning, and resource allocation within wards and increasing the use of assistive devices.Conceptualization, K.K., M.L.S., K.S. and A.H.; methodology, K.K., M.L.S., K.S. and A.H.; software, K.K.; validation, K.K.; formal analysis, K.K.; investigation, K.K.; resources, A.H.; data curation, K.K.; writing—original draft preparation, S.K. and K.K.; writing—review and editing, S.K., K.K., M.L.S., K.S. and A.H.; visualization, S.K.; supervision, A.H.; project administration, K.K.; funding acquisition, A.H. All authors have read and agreed to the published version of the manuscript.This research was funded by The Danish Work Environment Research Fund, grant number 03-2017-09. The DOSES cohort was financed by a grant from the Danish government.The study was conducted according to the guidelines of the Declaration of Helsinki, and ap-proved from the Danish Data Protection Agency and the Ethics Committee for the Capital Region of Denmark (protocol code H-4-2013-028 and date of approval 3 April 2013).Informed consent was obtained from all subjects involved in the study.The study datasets are available at the Danish National Archives, https://www.sa.dk/en/k/about-us.We gratefully acknowledge the entire DOSES research group and personnel who contributed to the data collection and the 20 nursing homes that participated in DOSES. We also wish to thank the group of data managers at the National Research Centre for the Working Environment for their great work in data processing.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.Scheme of the four-level hierarchical data structure of the present study and the level in which the specific investigated factors are measured at. Registered days (RD) (level 4) are clustered within eldercare workers (level 3), eldercare workers are clustered within wards (level 2), and wards are clustered within nursing homes (NH) (level 1), all stratified on day and evening shifts.Descriptive of the hierarchical structure and the clusters within the four levels of the dataset. N = Total population.Characteristics of organizations on ward level and eldercare workers stratified on day and evening shifts work. (Res. = Respondents/cases; SD = Standard Deviation; % = percent; n = number.)Estimated contribution of different sources of variance to the total variability in number of resident handlings performed per shift in eldercare work stratified on day (N = 5572) and evening shifts (N = 2373) work. (BGV. = Between Group Variance; 95% CI = 95% confidence intervals.)Univariate and multivariate analysis of the association between organizational, eldercare worker factors, within eldercare worker factors and the number of resident handlings performed per shift stratified on day and evening shift work. (Est. = Estimate; S.E. = Standard Error; 95% CI = 95% confidence intervals.)* p = <0.1; ** p = <0.05; *** p = <0.01; ‡ p = <0.001; † p = <0.0001; a = Continuous variable; b = Categorical variable; c Not included in the multivariate model for evening shift; d Number of cases without missing data, day = 5381 and evening = 2309 (only for the univariate model); e Number of cases without missing data, day = 5417 and evening = 2310 (only for the univariate model); f Number of cases without missing data, day = 4889 and evening = 1860 (only for the univariate model); g Number of cases without missing data, day = 5326 and evening = 2274 (only for the univariate model).Percentage change to the total variability in resident handlings per shift (at each level), due to the inclusion of specific factors.a = Continuous variable; b = Categorical variable; c Not statistically significant in the multivariate model in evening shifts; Note: This table shows how the percentage variance derived from each level changes when adding individual specific variables into the crude VCA model. For example, adding the percentage of residents with higher needs for physical assistance variable into the crude VCA model reduced the total variance in resident handlings explained at the nursing home level by 44% for the day shifts.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11041.txt

@@ -0,0 +1 @@
+Schools are an important arena to curb the decline in physical activity (PA) in youth. School-based interventions with accelerometer-measured PA are warranted. This study aimed to increase accelerometer-measured PA in adolescents following a 12-month school-based intervention. Two school-classes of 16–18-year-old Swedish students were allocated to intervention group and control group. Accelerometer-measured PA was gathered at baseline, 6- and 12-month follow-up. Mixed-effects linear regression was used to investigate between-group and within-group differences in mean minutes per day (min/day) of moderate to vigorous PA (MVPA), light PA (LPA) and sedentary time (ST). Fifty-seven students participated (intervention group = 31, control group = 26). At 12-month follow-up, the intervention group performed 5.9 (95% CI: −4.3, 16.2) min/day more in MVPA, 1.8 (95% CI: −17.9, 14.2) min/day less in LPA, and 4.1 (95% CI: −27.3, 19.2) min/day less in ST compared to the control group. Within the intervention group, there was no significant change in PA. Within the control group, LPA decreased (95% CI: −19.6, −0.2; p = 0.044) and ST increased (95% CI: 1.8, 30.8; p = 0.028). Although no between-group differences in PA were statistically significant, the within-group changes may suggest a preventive impact on the decline in PA during adolescence.Physical activity (PA) has a beneficial effect on mental and physical health [1], whereas excessive sedentary time (ST) is a risk factor for morbidity in adults [2]. The transition from childhood to adolescence is characterized by increasing ST and a decrease in PA [3]. Notably, both globally and in Sweden, a majority of adolescents fail to achieve the World Health Organization (WHO) recommendation of at least 60 min of moderate to vigorous PA (MVPA) per day [4,5]. Since PA in childhood and adolescence predicts PA in adulthood [6,7], and since PA of any intensity is associated with a substantial reduction of all-cause mortality in a dose-response pattern in adults [8], finding ways to increase PA in youth could have a positive long-term impact on public health [9]. Adolescents spend a large proportion of their waking hours at school [10], where ST is predominant [11]. Since schools can reach a majority of adolescents, the school environment is an important arena to promote PA in youth [12]. However, previous school-based interventions to increase PA in adolescents have shown modest positive results. For example, a meta-analysis by Borde et al, including 12 randomized controlled school-based PA interventions in adolescents, reported a standardized mean difference (SMD) of 0.24 min in MVPA between intervention and control groups (95% CI: −0.08, 0.56) [13]. Similarly, Love et al. included 17 cluster-randomized controlled school-based PA interventions and found a SMD of 0.02 min in MVPA in between-group comparisons (95% CI: −0.07, 0.11) [12].Another problem is the frequent use of self-report methods to measure PA in previous studies, which are known to overestimate PA while underestimating ST compared to objective measures [14]. For example, in a systematic review by Hynynen et al., four out of seven included randomized school-based interventions that reported significantly positive effects on PA were solely based on self-report data [15]. An additional challenge is that relatively few interventions have targeted older adolescents [13,15].Considering the importance of preventing a decline in PA in late adolescence, and the scarce amount of previous school-based interventions with objectively measured PA in this age group, we report the findings of a pragmatic trial that aimed to increase accelerometer-measured PA in a group of students aged 16–18 years. Specifically, the trial was set in a Swedish upper secondary school during 2019 and aimed to increase accelerometer-measured PA at 12 months following an experimental school-based intervention, with the long-term ambition to also promote academic performance and mental health. The “Active and Participating by Innovative Classrooms” study was a pragmatic school-based intervention study set in an upper secondary school in a small city in the middle of Sweden during 2019. In January 2019, a group of 16–17-year-old (age at recruitment) students attending the first year of the same upper secondary program and grade but in two separate school classes, were invited to participate. Parental consent was obtained in a separate information meeting in January 2019. The school-classes were pre-determined, by the local project coordination group, to be an intervention group and a control group respectively, resulting in two non-randomized groups. Specifically, in the selection procedure of intervention vs control group condition, it was noted that one of the two school classes consisted of students attending additional physical education (PE) (2–3 additional PE-classes per week). These additional classes of PE were part of an extended PE-program at the school, and thus, were similar to the ordinary PE-classes in terms of content and duration. The school class with a higher frequency of PE-classes was assigned the control condition, emulating an active control group with expected greater baseline PA, and maximizing potential health benefits for the least physically active class. The trial was pre-registered [16], and approved by the Swedish Ethical Review Authority (Reference Number: 2019-01128).The intervention group was exposed to a PA-promoting intervention consisting of (1) activity breaks during lectures in English and Mathematics, (2) adjustments to the classroom environment, and (3) time for reflection during PE classes (Figure 1). The multi-component intervention approach was based on theories of social-ecological influence on behavioral patterns, linking individual change to the social and physical environment [17], which have been shown to be effective to increase PA in youth [18]. The total length of exposure was 14 weeks during the spring semester and 14 weeks during the fall semester of 2019. During lectures in English and math, the exposure consisted of (1) low-intensity PA breaks of 5 min twice per lecture, and (2) classroom adjustments consisting of standing tables, desk bikes, and additional equipment such as balance pads, exercise balls, and elastic resistance bands. Activity breaks served to interrupt longer sessions of ST, during which the additional equipment served to facilitate unstructured PA, and the use of desk bikes can increase levels of LPA [19]. Lectures in math were scheduled three times per week, and in English two times per week, all extended by 10 min to enable activity breaks (i.e., a total extension of 50 min per week). To ensure an equal time distribution between standing tables, desk bikes, and regular sitting tables respectively, the intervention group was divided into three groups and assigned an equipment type for two weeks followed by rotations.Finally, during PE classes once per week, the intervention consisted of (3) time for reflection on the intervention and the impact of PA on physical and mental health in general, to increase commitment and participation [20]. To facilitate these reflections, the intervention group listened to 15-min podcasts with inspirational content while walking outdoors during PE classes once every two weeks.The trial involved three measurement points with accelerometer-measured PA; baseline assessment in January, mid-intervention (6 months) in June, and 12-month follow-up in December. PA was measured using Actigraph GT3X accelerometers, providing greater reliability and validity compared to subjective assessments [21]. Recommended settings for wear protocol were considered; measures were made with accelerometers at the right hip during waking hours for seven consecutive days at all measurement points [22]. Minimum wear time was set to ≥10 h per day and non-wear time was defined as ≥60 min with zero activity counts (i.e., no accelerations due to body movement) [22]. A valid measurement period was defined as ≥1 day with ≥ 10 wear time hours. Activity counts were analyzed as vector magnitude (Vm) in 15-s epochs, in line with current recommendations for children and adolescents [22]. PA-intensities were calculated based on recommended cut-off values [23]: ST was defined as <720 counts per minute (CPM), LPA as 721–3027 CPM, MPA as 3028–4447 CPM, and VPA as ≥4448 CPM. Raw accelerometer data files were processed using ActiLife software version 6.13.4.Descriptive characteristics were collected from questionnaires at all measurement points, where participants were asked to detail their date of birth, sex (female/male), organized sports participation (yes/no), distance to school (kilometers), season-dependent active commuting (bicycling/walking all year, during spring or fall semester, or not at all), and describe if they had a foreign background (born outside Sweden or both parents born outside Sweden). Descriptive characteristics and baseline accelerometer data on PA are presented using relevant measures of central tendency and dispersion. All analyses were performed using an intention-to-treat analysis. The primary outcome was between-group differences in MVPA (mean minutes per day) at 6- and 12-month follow-up. Secondary outcome measures included between-group differences in LPA and ST (mean minutes per day), and within-group changes of each activity intensity. To model the mean activity in each group at each time point and its change over time we employed linear mixed-effects regression. To account for the repeated nature of data (i.e., the same participants over time), all models were fit with a random intercept for individuals and with robust standard errors. Each intensity of PA was analyzed in separate models and all models were adjusted for sex and accelerometer wear time hours. All statistical tests were two-sided and a p-value < 0.05 was considered statistically significant. All data analyses were performed in Stata version 15.1 (StataCorp, College Station, TX, USA).Of the two enrolled classes, a total of 61 students chose to participate of which 32 were in the intervention group and 29 in the control group. At baseline measurement, valid accelerometer data were provided by 31 participants in the intervention group and 26 participants in the control group (Table 1). At 12-month follow-up, valid accelerometer data were provided by 26 and 25 participants in the intervention and control group, respectively.The intervention group had a high proportion of females (80.6%) compared to the control group (57.7%), and the control group spent a slightly higher proportion of accelerometer wear time in MVPA and LPA at baseline. Baseline accelerometer wear time was similar in both groups at 13.3 (SD 1.5) and 14.0 (SD 1.1) hours/day in the intervention and control group, respectively. Mean accelerometer wear time remained similar in the groups at mid-intervention (6 months) (13.3 (SD 1.2) and 13.0 (SD 1.1) hours/day in the intervention and control group, respectively) and 12-month follow-up (12.8 (SD 1.6) and 13.3 (SD 1.2) hours/day in the intervention and control group, respectively).At the baseline measure, the intervention group performed approximately 3.9 (95% CI: −13.7, 5.9; p = 0.437) minutes/day less MVPA compared to the control group (Figure 2). At 6- and 12-month follow-up, the intervention group performed 2.0 (95% CI: −9.5, 13.5; p = 0.732) minutes/day and 5.9 (95% CI: −4.3, 16.2; p = 0.255) minutes/day more MVPA compared to the control group, respectively. There was a statistically significant difference in both secondary outcomes at baseline (Figure 2); the intervention group performed less LPA (95% CI: −34.3, −2.8; p = 0.021) and more ST (95% CI: 0.7, 44.0; p = 0.043). However, there were no statistically significant between-group differences in mean minutes per day of ST and LPA at either 6-month or 12-month follow-up. Specifically, at 6 months the intervention group performed 12.6 min/day less LPA (95% CI: −29.8, 4.6; p = 0.152) and 10.4 min/day more ST (95% CI: −12.9, 33.7; p = 0.380) compared to the control group. At 12-month follow-up the intervention group performed 1.8 min/day less LPA (95% CI: −17.9, 14.2; p = 0.823) but, contrary to 6 months, 4.1 min/day less ST (95% CI: −27.3, 19.2; p = 0.732) compared to the control group.In the intervention group, the mean minutes of MVPA increased across all measurement points, resulting in an estimated increase of 3.3 (95% CI: −3.9, 10.6; p = 0.367) minutes/day from baseline to 12-month follow-up (Table 2). Contrary, in the control group the mean minutes of MVPA decreased across all measurement points with an estimated decrease of 6.5 (95% CI: −14.3, 1.3; p = 0.102) minutes/day at 12 months.From baseline to 12-month follow-up, the intervention group increased LPA by 6.8 (95% CI: −4.2, 17.8; p = 0.228) min/day and decreased ST by 10.1 (95% CI: −25.6, 5.4; p = 0.203) min/day, whereas the control group decreased LPA by 9.9 min/day (95% CI: −19.6, −0.2; p = 0.044) and increased ST by 16.3 min/day (95% CI: 1.8, 30.8; p = 0.028) (Table 2).We have presented the findings of a pragmatic trial that aimed to increase accelerometer-measured PA in a group of students aged 16–18 years following a 12-month experimental school-based non-randomized intervention. There were not any statistically significant between-group differences in MVPA (main outcome), LPA or ST (secondary outcomes), at any of the measurement points. However, from baseline to 12-month follow-up, the intervention group increased time spent in accelerometer-measured MVPA and LPA and decreased in ST, whereas the opposite was observed in the control group. The observed within-group changes may reflect the proposed transition towards increased ST, which was observed in the control group but not in the intervention group. In specific, the control group significantly reduced time spent in LPA by 9.9 min/day (95% CI: −19.6, −0.2; p = 0.044) and increased time in ST by 16.3 min/day (95% CI: 1.8, 30.8; p = 0.028). As such, our secondary findings suggest that the intervention may have had a preventive impact on the transition from PA to ST that typically occurs during adolescence [3]. Nevertheless, the positive but not statistically significant between-group effects on PA in the present study are in line with several previous school-based PA interventions [12,13]. Notably, a large proportion of previous school interventions reporting positive effects on PA are based on subjective measures (e.g., questionnaires) [12,24]. However, in contrast to subjective measures of PA, studies utilizing objective measures of PA have failed to demonstrate any effect of school-based interventions [12]. A potential reason could be the risk of overestimating PA with subjective measures compared to objective methods [14]. Therefore, comparisons with results from studies with subjective measures are challenging as the findings may differ due to measurement reliability and validity. An additional challenge is that relatively few previous interventions have targeted older adolescents specifically, since the effectiveness of interventions may vary between age groups [15].A major strength of this pragmatic trial was the use of accelerometers to objectively measure PA, with greater reliability and validity as compared to self-report methods [22]. However, accelerometers have a limited ability to capture movements such as swimming, weight training, or bicycling, which may result in an underestimation of PA [25]. The use of accelerometers also implies attaching a foreign element to the body, which could cause compliance issues in the event of discomfort or a lack of knowledge on how to use the equipment. Hence, careful how-to-use instructions and appropriate preparation of the equipment is important to limit the risk of reduced accelerometer wear-time. An additional strength was a comprehensive multi-component intervention, including adjustments to the physical environment as well as curricular activities with reflections, walking podcasts, and increased recess time, since multicomponent intervention approaches are most effective in promoting PA in adolescents [26]. Furthermore, the 12-month duration of the trial included measures across seasonal variations, which increases the reliability of the findings as objectively measured MVPA in adolescents is higher during the spring [27].However, the non-randomized small-scale nature of the pragmatic trial has several limitations. The study population consisting of 57 individuals was relatively small compared to previous school-based interventions, making analyses particularly vulnerable to invalid accelerometer data and/or insufficient wear time. To maximize the sample size, we decided to adjust for wear time rather than exclude participants with insufficient wear time according to traditional inclusion criteria, typically defined as ≥3 days with ≥10 wear time hours/day [21]. This is an acknowledged limitation, as the relaxed inclusion criteria increase the power at expense of the reliability of estimated time in different PA-intensities [22]. Furthermore, there was a substantial difference in sex distribution between the groups, particularly skewed in the intervention group consisting of 80.6% females. Since females engage in less PA than males during adolescence [28], this could contribute to the higher levels of PA and lower ST observed at baseline in the control group. Although all models were adjusted for sex, our findings are likely affected by residual confounding, and as such our findings should be interpreted with caution. In addition, some circumstances regarding the implementation and execution of the intervention are important to consider. The project leader reported that students tended to choose certain table types depending on individual preferences, implying that the 6-week intervention cycles with different equipment were not perfectly performed according to the protocol during the whole intervention. Although the use of a desk bike or standing position has limited representation in hip-worn accelerometer-measured PA, since the registration of such activities is weak [25], it might influence individual change in PA over time. This is due to the theoretical relation between behavioral change and the physical environment, which is a central idea of the present intervention approach [17].Schools are important to prevent the decline in PA in adolescents and provides an opportunity to promote positive long-term health behaviors. In general, efforts to increase and measure PA objectively are expensive and resource-demanding, and few previous interventions have focused solely on late adolescents. Hence, this pragmatic study that aimed to increase accelerometer-measured PA in a group of students aged 16–18 years is a valuable contribution, adding to the pool of objective data on PA in this age group and describes certain challenges in changing the physical school environment to increase PA. Although the present intervention study did not result in a significant increase in PA, it highlights the importance of actions to promote PA as the control group may reflect the typical transition to more ST during adolescence. In addition, the results suggest that an experimental intervention approach in a small-scale setting may have a preventive impact on the decline in PA. This potential benefit should be further explored in future research using larger sample size and randomizationFollowing a 12-month experimental school-based intervention there were not any statistically significant between-group differences in MVPA, or LPA and ST. Yet, the within-group changes may suggest that the intervention had a preventive impact on the transition from PA to ST that typically occurs during adolescence.Conceptualization, F.C., V.H.A., D.B.; methodology, F.C., V.H.A., M.N., H.L., D.B.; formal analysis, F.C., V.H.A.; investigation, F.C., V.H.A.; writing—original draft preparation, F.C., V.H.A.; writing—review and editing, V.H.A., M.N., H.L., D.B.; visualization, F.C., V.H.A.; supervision, D.B.; funding acquisition, M.N., H.L., D.B. All authors have read and agreed to the published version of the manuscript.The present study received funding by the Swedish National Agency for Education. The study was approved by the Swedish Ethical Review Authority (Reference Number: 2019-01128).Informed consent was obtained from all subjects involved in the study. Data is available upon reasonable request provided approval from appropriate authorities.All authors of the present study declare no conflicts of interest.Overview of the Intervention Content. Abbreviations; PA = physical activity, LPA = light physical activity, ST = sedentary time, PE = physical education.Between-group differences in physical activity and sedentary time from baseline to 12-month follow-up. Adjusted for sex and accelerometer wear time. Abbreviations; MVPA = moderate to vigorous physical activity; LPA = light physical activity; ST = sedentary time.Characteristics of the study population and crude accelerometer data at baseline (T1).Abbreviations; LPA = light physical activity; MVPA = moderate to vigorous physical activity; ST = sedentary time; km = kilometers.Within-group changes in physical activity and sedentary time from baseline to 12-month follow-up.Abbreviations; MVPA = moderate to vigorous physical activity; LPA = light physical activity; ST = sedentary time.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11042.txt

@@ -0,0 +1 @@
+The impact of the lockdown, during the period from March to June in 2020, upon the air quality of the Basque Country in northern Spain is analyzed. The evaluation accounts for the meteorology of the period. Daily and sub-daily analysis of aerosol and ozone records show that the territory was repeatedly affected by episodes of pollutants from outer regions. Three episodes of PM10 and ten of PM2.5 were caused by transported anthropogenic European sulfates, African dust, and wildland fires. The region, with a varied orographic climatology, shows high and diverse industrial activity. Urban and interurban road traffic of the region decreased by 49% and 53%, respectively, whereas industrial activity showed a lower reduction of 20%. Consequently, the average concentrations of NO2 in the cities during the period fell to 12.4 µg·m−3 (−45%). Ozone showed up to five exceedances of the WHOAQG for the daily maximum 8-h average in both rural and urban sites, associated with transport through France and the Bay of Biscay, under periods of European blocking anticyclones. However, averages showed a moderate decrease (−11%) in rural environments, in line with the precursor reductions, and disparate changes in the cities, which reproduced the weekend effect of their historical records. The PM10 decreased less than expected (−10% and −21%, in the urban and rural environments, respectively), probably caused by the modest decrease of industrial activity around urban sites and favorable meteorology for secondary aerosol formation, which could also influence the lower changes observed in the PM2.5 (−1% and +3% at the urban and rural sites, respectively). Consequently, in a future low NOx traffic emission scenario, the inter-regional PM and ozone control will require actions across various sectors, including the industry and common pollution control strategies.COVID-19, a disease caused by the SARS-CoV-2 coronavirus, expanded rapidly throughout the world during the first months of 2020, reaching pandemic status on March 11. Many countries were presented with unprecedented challenges with the intensive care units of their national health systems reaching saturation point. At this point (June 2021), the number of infections has exceeded 175 million and almost 4 million deaths have already been confirmed in the world (https://www.worldometers.info/coronavirus/; accessed on 14 June 2021). To reduce the spread of SARS-CoV-2, lockdown measures were implemented worldwide with varied timing and severity according to the onset of the epidemiological crisis and the evolution of infections. The reduction of mobility and changes in industrial/commercial activity affected global emissions, but many details of these changes are yet unclear. Annual national emission inventories previously lagged three years behind the present [1], and thus, the year 2020 is not yet available to assess the impacts of COVID-19. Present estimations of emissions can be based on mobility and industrial activity changes, electric and fossil fuel energy demand, economic output, and indirect evidence of air pollution changes (including ground-based and satellite observations) with respect to a previous reference period [2,3].Using the year 2016 as the reference for the emission changes, Ref. [4] estimated average emission reductions for the period 20 January–26 April 2020 to be −33% for NOx, −8% for Non-Methane Volatile Organic Compounds (NMVOCs), −7% for SOx, and −7% for PM2.5 in the EU-28 plus Norway and Switzerland. These reductions vary according to the studied reference period and the analyzed countries, which were not subject to the same mobility–activity restrictions during the period. Activity and mobility in the EU-27 (EU-28 excluding the UK) were sharply reduced in the first half of March 2020 as a result of the implementation of lockdown measures which were initiated on February 21 in Lombardy (Italy) and then extended to the rest of the Union. Mobility gradually recovered in May in some areas faster than in others depending on how quickly lockdown measures were eased [5]. The reported shifts in human mobility and the industrial activity resulting from the lockdown offer a unique opportunity to identify their effects on urban and rural air quality.As a result of the aforementioned emission reductions, many studies on air quality changes have been published. Some of them are focused on urban environments [6,7,8] or both urban and rural environments [9,10,11]. Such studies range from the local to the continental and global scale. Most of them estimate the average changes over a period for a selection of pollutants such as NO2, NO, O3, PM10, PM2.5, and CO2. In Spain, NO2 levels fell below 50% of the WHO annual air quality guidelines (WHOAQGs), but those of PM2.5 were reduced less than expected attributed to, among other causes, meteorological conditions favoring high secondary aerosol formation [11]. Decreases in PM10 levels were greater than in PM2.5. Daily maximum 8-h average (MDA8) O3 experienced a generalized decrease in all the rural receptor sites when lockdown restrictions were relaxed (June–July) with −20%. For urban areas (including Bilbao), the average MDA8 O3 period responses were heterogeneous, with increases or decreases depending on the period and location. However, the O3 WHOAQG was still exceeded during the lockdown in several cities. Sicard et al. (2020) also detected strong reductions in NO2 mean concentrations in four European cities in Italy, France, and Spain: ~53% at urban stations, comparable to Wuhan (57%), and ~65% at traffic stations. In Europe, NO declined even further, ~63% at urban stations and ~78% at traffic stations. Reductions in PM2.5 and PM10 at urban stations were much smaller. The NOx concentrations during the lockdown were on average 49% lower than those at weekends of the previous years (2017–2019) in all cities. The lockdown effect on O3 production was ~10% higher than the weekend effect in the studied southern European cities. However, very few studies are devoted to pollution episodes and their analysis together with the observed emission/air pollution/meteorology variations inside or around the target domain, which could be important to understand the origin of the observed changes or to identify eventual transport episodes. In this respect, the evaluation of the concentration series at a daily/hourly resolution is needed, together with the observed meteorology/emission changes, in order to show the real length and intensity of episodes, search for their origin, and suggest actions for the eventual control strategies. In this manuscript, we show a comprehensive analysis of the impact of the COVID-19 lockdown on the air pollution of the Basque Country, a region in northern Spain with important industrial activity, and three major cities distributed throughout complex mountainous topography next to the French border. Special attention is paid to the occurrence of episodes during the period as well as their origin.The Basque Country (B.C.) is an autonomous Spanish community located in the north of the Iberian Peninsula, at the eastern end of the Cantabrian coast and near France (Figure 1). The orography of the B.C. is mountainous, with moderate altitude peaks of approximately 1000–1500 m.a.s.l (Figure 1). Three climatic zones can be distinguished: (1) the coastal zone and the North Atlantic slope, very rainy with moderate temperatures softened by the influence of the sea, where the main urban areas of Bilbao and Donostia-San Sebastián are located; (2) a transition zone between the oceanic and the Mediterranean climate, with mild temperatures and less precipitation than in the Atlantic slope, where Vitoria-Gasteiz is located; (3) the southernmost area corresponds to the Ebro valley, which drains into the Mediterranean, with important seasonal temperature fluctuations, dry and hot summers, and little precipitation throughout the year [12]. The climate of the coastal valleys and mountains can be classified as Cfb following the Köpen–Geiger climate classification, while the southernmost region shows a Csb-type climate [13]. Anticyclonic conditions have historically been reported to be behind the most severe pollution episodes in the region [14,15]. They are associated with weaker winds, calm, and/or re-circulations, which are also accompanied by a limited vertical dispersion of pollutants due to subsidence and the presence of temperature inversions at the lower troposphere. The subsidence inversions prevent pollution from efficient vertical ventilation and dispersion, as was already reported in the last century in the region [14,16,17] and can efficiently transport aged air masses with high ozone concentrations from more remote area sources into the studied region [18]. Anticyclones during the spring and summer seasons bring longer and more intense sunlight periods, which enhance ozone production after photochemical reactions from NOx and NMVOCs precursors. For ozone in the Basque Country (B.C.), an efficient transport mechanism from the upwind regions of precursor emissions ideally occurs under high-pressure conditions in western Europe and southern France [18,19,20,21]. Easterlies and northeasterlies from the European landmass, when they persist for several days, are responsible for such transport. We documented that the local ozone concentrations can rise well above 120 µg·m−3 for the daily 8-h averages during consecutive days, giving rise to the most severe/long episodes registered in the regional monitoring network of the B.C. Almost half of the 2.2 million inhabitants of the B.C. live in the core and commuting zone of the city of Bilbao (the Greater Bilbao), the most populated city of the three provincial capitals. With 350,184 inhabitants in the core city, the rest are located in approximately 35 municipalities, most of them following the 15-km long estuary of the lower Nervión valley, from Bilbao city into the sea. The other two main capitals are Vitoria-Gasteiz with 253,996 inhabitants, and Donostia-San Sebastián with 188,240 inhabitants [22]. Regarding the economic activity of the whole region, industry represents 24.2% of the gross value in 2017 which was above the European EU-28 (19.6%) and the Spanish (16.2%) averages [23]. The most relevant sectors are metallurgy, machine tools, electricity generation, and transportation. The activity of the port of Bilbao represents 1.29% of the gross domestic product of the B.C. [24] and it is the most important Atlantic Port of Spain and the gateway to the European Atlantic trade routes. In Spain (312 veh·km−2), the province of Bizkaia ranks third for vehicle density, behind Madrid and Barcelona [11]. The province of Gipuzkoa also has a high vehicle density (238 veh·km−2), while Álava shows the lowest values (70 veh·km−2) of the B.C. [25]. The three provinces (Bizkaia, Gipuzkoa, and Álava), in Figure 1, show a different industrial distribution around their main cities (Figure 2). The location of the EPRTR industries [26] in and around the Greater Bilbao, follows the main estuary area and the two main valleys that drain into the city from the southeast and south. In Gipuzkoa, only a small fraction of its industry is located around the city of Donostia-San Sebastián, while the main fraction is well-distributed inside the three coastal valleys which drain directly into the sea, west of the city. In contrast, the main fraction of the industry in Álava is concentrated around the city of Vitoria-Gasteiz. The majority of the metal industry is located in the B.C., as shown in Figure 2, but there is also high diversity in industries distributed across the entire territory devoted to the production of chemicals and petrochemicals, cement, pulp and paper, waste treatment, waste incineration, and gas-fired power plants, among others.The Air Quality Monitoring and Control Network managed by the Basque Government operates with the aim of evaluating the air quality in the whole territory. Currently, the network has 50 stations: 21 located next to industrial sources, 18 next to roads where traffic may be the main pollution source, and 11 at rural sites. Air pollutants are automatically measured on an hourly basis, as each analyzer is fitted with a suitable measurement technique and meets reference standards. This study is focused on regulated pollutants comprising sulfur dioxide (SO2), nitric oxide (NO), nitrogen dioxide (NO2), tropospheric ozone (O3), and particulate matter (PM10 and PM2.5) [27].To assess the impact of the lockdown restrictions on the air pollution of the three provincial capitals, a selection of urban background sites and traffic stations was made, following the European Environmental Agency (EEA) criteria. This organization uses a viewer that tracks the weekly and monthly average concentrations of NO2 and PM https://www.eea.europa.eu/themes/air/air-quality-and-covid19 (accessed on 18 October 2021) to assess the effect of the COVID-19 lockdown upon the air pollution of European Countries. In this application, Bilbao (BI*) is represented by five stations and Vitoria-Gasteiz (VG*) and Donostia-San Sebastián (DS*) by four (Table 1). The Facultad de Farmacia station in Vitoria-Gasteiz was added to the initial three stations of the EEA list in order to include at least one ozone monitoring site in the city. Their respective locations are shown in Figure 1. “Cluster-station data” are used for each city, as in the EEA application, which represents averaged values for each urban site (BI*, VG*, and DS*), as shown in Table 1. Moreover, to complete the study, and to assess rural air pollution, three monitoring stations, one for each province, were also selected (Figure 1). Table 1 gathers the sites and the measured air pollutants on each station. In order to identify anomalous data, each individual hourly average concentration was tested and validated following an exhaustive procedure [28]. To avoid unreliable trends, daily, weekly, and yearly averages have been calculated only when more than 75% of the hourly data were available. For example, to obtain daily means, data with at least an 18 hourly average were neededIn Spain, a state of alarm was established on Saturday 14 March 2020, and it was extended six times every two weeks until Sunday 21 June [29]. To study the lockdown impact on air quality in the B.C., 18 weeks were selected: 3 weeks before the lockdown (named w−3, w−2, and w−1), week 0 for the period 9–15 March, and the following 14 weeks of the lockdown (named w1–w14). Selected Monday-to-Sunday weeks, confinement, and mobility restrictions are all summarized in Table 2. Schools, colleges, and universities closed from 9 March onwards in Vitoria-Gasteiz, and from 13 March onwards in the remaining Basque territory. General restrictions enforced on 14 March included confinement and mobility limitations. Teleworking was enforced whenever possible, as non-essential services were ordered to close. In addition, from 30 March to 9 April, all non-essential service workers were forced to stay home. During this period, mobility was only allowed through force majeure, such as medical appointments or the acquisition of basic goods or food. Relaxation measures began on 4 May [29]. Mobility was restricted to the municipality and the use of public transport increased gradually. Allowance to leave was by age groups and time slots, and some establishments were open for appointments only, such as hairdressers, restaurants, bars, etc. Given the different situation of the Spanish Autonomous Communities, from 11 May onwards measures were dictated by the autonomous government [30]. In the following stages 1 and 2, social distancing measures were reduced progressively. From June 18 onwards, mobility between provinces was allowed and a situation of “new normality” started on 20 June [30]. In order to compare air quality data recorded during the 2020 lockdown (Table 2), corresponding periods for the five previous years were selected according to the criteria of the European Environment Agency [31], including the weekend effect. Selected Monday−to-Sunday weeks are detailed in Table 3. The largest road traffic contribution in the B.C. corresponds to the province of Bizkaia (48–52% of the B.C. total in Table 4), and most of it corresponds to the Greater Bilbao’s urban area and its commuting zone. Data for the traffic changes during the lockdown (Section 3.1) were obtained from traffic counting stations at the three Basque Provinces [32] and as personal communication from each town council (Bilbao, Vitoria-Gasteiz, and Donostia-San Sebastián). Shipping emissions in ports are related to port traffic activity. The main port of the B.C. is in the estuary of the Greater Bilbao [33]. The contribution of the second port, in the province of Gipuzkoa, to the total traffic in the B.C. is marginal (7.7%), as shown in Table 4. The monthly variations on total traffic [34] have been used to estimate changes during the w0–w14 period (Section 3.1). The three airports in the B.C. are located 13–20 km from their respective core cities. The Vitoria-Gasteiz airport accounts for 99.8% of the B.C.’s merchandise (air goods), while the main airport of Bilbao accounts for 94% of passengers (Table 4). The estimation of aircraft emissions at ground level and their air pollution effect near airports have been analyzed in different studies. Indeed, the near surrounding areas of the airports, between 1 and 5 km, are the most sensitive to NOx pollution, mainly emitted during the take-off and climbing phases. CO and Hydrocarbons (HC) are mostly emitted during the taxiing of aircraft, while SO2, PM2.5, and other trace compounds are primary air pollutants or precursors of secondary pollutants [35].In the B.C. there are three natural gas combined cycle power plants with a total installed capacity of 2000 MW. They are all located in the commuting area of the Greater Bilbao: two of them at 14 and 17 km to the nothwest of the core city, and the third one to the southeast, following the valley of the Ibaizabal River, draining into the city (Figure 2a). Their emissions can impact the whole urban area under adequate meteorology. In the B.C., electricity from non-renewable sources comes from the three mentioned combined cycle thermal power plants (58%), a cogeneration plant (36%), and a waste incinerator (6%) [36], marked in closed red triangles in Figure 2a. Industrial activity and power plants are the main sources of NOx in the region: they emit more than road traffic (5/4 ratio) and more than the residential domestic sources (5/1). They are also practically the only sources that emit SO2 and major emission sources of primary PM2.5 in a 4/1 ratio with respect to road traffic [37]. Annual emissions for the year 2018 (last update), and source distribution for NOx, SOx, PM, and NMVOCs, are shown in Figure 2. Industrial activity changes during the lockdown could be assessed by using the energy (electricity and gas) consumption changes discussed in Section 3.1. The Industrial Production Index (IPI) can also be used as an industrial activity indicator [38]Meteorology is one of the main drivers of the pollutant concentration of a region. Intense winds bring ventilation periods with low concentrations while weak winds and/or local and mesoscale wind re-circulations give rise to the accumulation of pollutants and pollution episodes. A reduction of pollution concentrations in a region, like during the lockdown period, can be attributed to fewer emissions, good ventilation conditions, or both. Contrarily, an increase or the mere absence of a trend in a low emission scenario makes it necessary to search for adequate reasons, which could be related to non-accounted pollution sources and/or the concurrence of adverse meteorology. As it is discussed in Section 3.3, Section 3.4, and Section 3.5, this is the case for some of the pollutants monitored in the Basque network. To search for the role of meteorology in the observed concentrations during the lockdown period, meteorological anomalies with respect to the expected average conditions in the region are estimated in Section 3.2. Meteorological variables such as wind, cloud cover, temperature, and pressure distribution were selected to search for anomalies during the March–June 2020 lockdown period, with respect to the five-year 2015–2019 averages of the same four-month period. Hourly ERA-5 re-analysis data are used for the study [39], and the Grid Analysis and Display System (GrADS) for the data processing and representation of anomalies [40]. The six-hourly NCEP Climate Forecast System Reanalysis (CFSR) historical archive in Wetterzentrale (http://www.wetterzentrale.de/; last accessed: 14 June 2021) and the satellite imagery of NASA’s Global Imagery Browse Services (GIBS) were also used to analyze the main O3 and PM episodes recorded in the region during the lockdown in Section 3.5.The concentrations of air pollutants recorded during the lockdown period in 2020 (Table 2) are compared with data obtained during the corresponding periods of the previous five years (2015–2019) in Table 3. Inter-annual and lockdown average concentrations of the full lockdown period, both at the selected three clustered urban sites and three rural sites, are estimated in Section 3.3 with their corresponding ensemble urban/rural averages. Hourly data availability is between 90% and 100% for all pollutants and sites, except for PM2.5, NO, and NO2 at the Pagoeta site during the 2015–2019 period, with 76%, 85%, and 85% of the data available, respectively, due to anomalous data for approximately three weeks which were removed from the database. The series of weekly concentrations during the lockdown can show changes with respect to climatological values not represented in the full lockdown averages. Comparisons at a higher time resolution (weekly or daily) of the NO2, O3, and PM10 concentrations during the lockdown with the corresponding 2015–2019 inter-annual values are performed in Section 3.4 and Section 3.5. They can help to study the effect of each stage of the lockdown in the air quality or to detect episodes of “anomalous” concentrations at this shorter time-scale above/below the inter-annual averages of the daily values. It is assumed that ozone and particles (natural or anthropogenic) episodes can occur randomly during the March–June period throughout the years. Those all events are included in the climatological baseline of daily averages and variability of the concentrations of the previous five years. In a scenario of low emissions, like during the lockdown, the positive anomalies can be related to transport episodes of ozone and particles (natural or anthropogenic) under adequate meteorology. The episodes last a variable number of days (usually less than a week) and the list of episodes [41] evaluated using the Navy Aerosol Analysis and Prediction System (NAAPS) with satellite imagery and high-resolution trajectories are used to search for positive anomalies associated with African dust, European sulfates, and wildfires over the climatological records (Section 3.4 and Section 3.5).In order to study mobility decrease during the state of alarm in the B.C., the percentage of traffic reduction with respect to the same 2019 March–June period was calculated for urban and interurban roads (Table 4). Average urban and interurban traffic reduction in the B.C. during the lockdown was −49% and −53%, respectively. Its effect on air quality is discussed in Section 3.3. Traffic started to decrease during the first week with general restrictions (w1) and a minimum was reached during 6–12 April (w4), coinciding with restrictions on non-essential activities (Table 2). The decrease was significant for intercity roads (−82%) and down to −85% for highways for light vehicles with respect to w0 and less for heavy vehicle circulation on highways (−65%) as the transport of goods was an essential activity. From week 4 onwards, traffic tended to recover, but values previous to the lockdown were not reached. Regarding urban traffic, the largest reduction with respect to w0 was −81%, −70%, and −85% at Bilbao, Vitoria-Gasteiz, and Donostia-San Sebastián, respectively, during week 4. The time evolution of the traffic reduction is discussed in Section 3.3 in the context of the observed NO2 and O3 concentration changes in the three cities.The activity and logistics operations of the port of Bilbao, with the highest contribution, barely changed during the lockdown period (Table 4), ensuring the delivery of essential services and facilitating supply sources for industry and consumer products for customers [33]. Monthly variations on total traffic ranged between −3% (March) and −14% (June 2020) [34], with an average of −9.1% for the corresponding 2019 March–June period (Table 4). On the other hand, the three airports in the B.C. showed a decrease in their activity during the 2020 lockdown, although differently in each of them. Total passenger air traffic variation ranged between −91.0% and −94.4% at the three airports of the B.C., compared to the corresponding 2019 March–June period. The decrease in total merchandise air traffic affected Bilbao and Donostia-San Sebastián (−87.3% and −100%) more significantly. However, the Vitoria-Gasteiz airport, which accounts for 99.8% of the B.C.’s merchandise (air goods) traffic totals, barely changed during the lockdown period at −12.8% [42].The observed changes in the electricity consumption by main sectors in the B.C. during the period March–June 2020 relative to the same period in 2019 are summarized in Table 5 (left column). The table is based on the monthly data of electricity consumption by sector (buildings, domestic, services, iron and steel industry, and rest of industry) forwarded by the Basque Energy Agency [43]. The electricity consumption decreased during the lockdown (−14.3%), dragged down by a decrease in the demand in industry and services, which was not balanced by the slight increase in demand (+4.8%) in the domestic sector after the “stay at home” order.Between March and June 2020, with respect to the same months of 2019, the total consumption of natural gas decreased moderately (−11.1%), as shown in Table 5 (right column). This variation is supported by a moderate decrease in domestic and commercial consumption (−8.7%), a greater decrease in industrial consumption (−25%), and a significant growth for thermal power plants (+57%). Industrial activity changes during the lockdown could be assessed by using the energy consumption changes by sector (Table 5): an average decrease between 18% and 22% during the period March–June 2020 with respect to the same period in 2019 could be inferred from the observed decrease in the industrial electricity demand. The Industrial Production Index (IPI) is also used as an industrial activity indicator. In the B.C., the IPI decreased by 21% during the period March–June 2020, with respect to the preceding months of January and February [38]. This decrease was similar (−20.4%) to that observed in Spain [44]. The estimated change was uneven among the different sectors of the industry, depending on the evolution of demand and the storage capacity of the industrial production, among other reasons. In this sense, the refinery located close to the port of Bilbao, the largest industrial emission source of the B.C. (Figure 2), maintained its full activity during the first two months of the lockdown and reduced it by 40% from the beginning of May onwards, due to the persistent low fuel demand after the drastic reduction in mobility [45]. On the contrary, the three natural gas combined cycle power plants increased their activity by 57% (Table 5), with the corresponding increase in their relatively important NOx contribution.Emission changes out of the B.C. during the lockdown need to be considered in our analysis because they can explain the transport of pollution from upwind regions. The increasing level of pollutant concentrations in the surrounding regions can contribute to local/regional episodes (see Section 3.5) under the appropriate meteorology. Following the IPI monthly activity indicator [44], the industrial activity in France, similar to that of Spain, decreased by 20.4% during the period March–June 2020 with respect to the preceding months of January and February. However, a lower decrease is found for the average industrial activity changes in the EU-27 member states (−16.4%).Concerning urban and interurban road traffic, the largest mobility decrease in Europe, using mobile positioning data, was observed in Spain [5] with an average decrease of 55% during the period March–June 2020 relative to February before the lockdown, followed by France with an average decrease of 41%. The rest of Europe showed lower decreases. Ground transport CO2 emissions decreases were the largest in Spain (−16.6%), while France (−13.7%) and Italy (−13.0%) showed more modest decreases [46], although in this case the changes were estimated over the first seven months of 2020 with respect to the same period in 2019. Meteorological anomalies during the lockdown period (16 March–21 June 2020) with respect to the five-year 2015–2019 averages of the same corresponding period are represented in Figure 3. ERA-5 re-analysis of temperature (shaded colors in °C) and cloud cover anomalies (contours in %) are presented in Figure 3a,b, while the mean sea level pressure (contours in hPa) and wind anomalies (shaded colors in m·s−1 and vectors) are included in Figure 3c,d. The lockdown period was characterized by positive temperature anomalies in the north and northwest of the Iberian Peninsula and negative temperatures over the south and southeast (Figure 3a,b). The highest temperature anomalies were located over the southwest of France and in “the most southeast region” of the Bay of Biscay. The same relative warm areas showed negative cloud cover anomalies, which also affected the B.C. region (Figure 3b). The pattern of pressure and wind anomalies observed in Figure 3c,d show more intense European continental easterlies blowing over the higher-pressure anomaly located in western France and offshore over the Bay of Biscay, while the wind intensity in the B.C. approaches the average values, with a slightly negative 0.5 m·s−1 anomaly. Consequently, in a low emission scenario, the actual ventilation conditions in the B.C. would have contributed to a comparative reduction of ambient concentrations of primary pollutants near the sources (Section 3.3). Pressure, wind direction, temperature, and cloud cover are compatible with a higher-than-normal frequency of anticyclones over France during the period: the continental E winds would have brought the observed warm and dry anomaly to the northern regions of the Iberian Peninsula. Sea breezes during the daytime, which are frequent from March to October in the Cantabrian Coast, are expected to contribute to the inland convergence of the European continental easterlies over the Bay of Biscay into the northern coast of Iberia [14,18,19]. All these processes are responsible for the transboundary transport of ozone (and secondary aerosols) from southern France, which adds to the local production to give rise to the most severe episodes registered in the region (Section 2.1). The changes in the lockdown period average values of PM and ozone between 2020 and previous years (Section 3.3) could be related to the mentioned meteorological anomalies (temperature, cloud cover, and winds). All the ozone episodes identified during the lockdown in Section 3.5 also show the synoptic forcing responsible for the observed anomalies in Figure 3.Due to the greater total traffic and industrial emissions in the Greater Bilbao (BI*) with respect to the other capitals (Section 2.3), its average inter-annual concentrations of pollutants (NOx, SO2, PM10, and PM2.5) were the highest among the three urban environments (Table 6). Similarly, DS* showed higher concentrations than VG*. This is true for most pollutants except for O3, which showed a reversed behavior, with lower concentrations at BI* due to an increased O3 titration by NOx. Inter-annual NO2 (27.2 μg·m−3) average concentrations in BI* (the highest among all the sites) were well below the annual WHO’s Air Quality Guideline (WHOAQG) for this pollutant (40 μg·m−3). The SO2 concentrations for the inter-annual values were also low (5.9 μg·m−3) and there was no exceedance of the 24-h average guideline (20 μg·m−3). However, PM10 (17.7 μg·m−3) and PM2.5 (9.5 μg·m−3) were close to exceeding their respective annual guidelines of 20 μg·m−3 and 10 μg·m−3. The largest urban O3 concentrations were registered in VG* (66.2 μg·m−3) and it was relatively low when compared to the rural averages. For the three rural sites MU, PA, and VA, the ensemble average of the inter-annual concentrations of NO (1.5 μg·m−3), NO2 (3.7 μg·m−3), and SO2 (1.9 μg·m−3) were much lower than those recorded in the urban sites (7.7, 22.6, and 3.9 μg·m−3 for the respective three pollutants). In this respect, the rural averages of NO and NO2 were only 19% and 16% of the respective urban concentrations while the rural SO2 were also relatively low: 49% of the urban values. This seems to be caused by the relatively large distance from their main sources (road traffic and industry) around the respective urban sites and the relatively short lifetimes of these pollutants. The average rural PM10 concentrations were 68% of the values registered in the cities. PM2.5 was even more uniformly distributed in the rural environment of the B.C. (5.6–6.2 μg·m−3), approximately 69% of the PM2.5 found in the average urban environment. The observed urban to rural concentration differences regarding NO and NO2 seem to be in accordance with the longer lifetime of PM2.5 and PM10, with resultant larger urban-to-rural differences at shorter lifetimes. On the contrary, the O3 field distribution was controlled both by the photochemistry and the availability of NOx and NMVOC species. Thus, its concentration in the rural sites, out of the NOx- saturated urban environments, was higher than in the cities. The highest averages were found in VA (84.1 μg·m−3 in Table 6). This monitoring station showed the largest history of O3 episodes in the B.C. [21,47,48,49,50,51]. The occurrence of O3 exceedances of the WHOAQG (100 μg·m−3) for the MDA8 O3 concentration during the lockdown is discussed below, in Section 3.5.During the lockdown, both NO and NO2 urban concentrations showed the largest decreases among all the registered pollutants in the monitoring stations of the B.C. The ensemble urban average in the three sites during the period March–June 2015–2019 was 7.7 µg·m−3 and 22.6 µg·m−3 for NO and NO2, respectively. During the lockdown, they decreased an average of 4.1 µg·m−3 for NO and 10.2 µg·m−3 for NO2, an important reduction of −53% and −45%, respectively, when compared to the inter-annual values. These average values were not equally distributed among the cities, with VG* showing the largest NO2 reduction (−53%) and BI* the lowest (−42%). These reductions are in accordance with the reported road traffic decrease (−49% and −53% in urban and interurban road traffic, respectively) in the whole B.C. (Table 4), which had a direct impact on the urban NOx. Other contributors to the observed NOx changes could have been the reported variations in the industrial and domestic commercial activity (Section 3.1), which use natural gas as the main source of heat or electric power. However, based only on the reported moderate reduction of the total gas consumption (−11% in Table 5) during the lockdown. It can be concluded that changes in road traffic were the main driver of the NOx urban decrease.The rural inter-annual concentrations of NO and NO2 in the B.C. were low because of their relatively short lifetime and long distance from the sources, as discussed above. Ensemble averages of the three sites were 1.5 µg·m−3 and 3.7 µg·m−3 for NO and NO2, respectively. During the lockdown, NO2 showed an important reduction of approximately 1.5 µg·m−3 in all three rural sites, while NO concentrations showed lower changes around its already very low value, with greater effect in MU than in VA and PA.SO2 concentrations were very low in both urban and rural sites: the ensemble urban average during the period February–June 2015–2019 was 3.9 µg·m−3, and 1.9 µg·m−3 in the VA rural site. The urban SO2 was significantly higher in BI* (5.9 µg·m−3) and lower in VG* (2.7 μg·m−3) and DS* (3.2 µg·m−3). The observed distribution could be attributed to industry, which is the main source of ambient SO2 in the region (Section 2.1), with the relative higher impact of a large refinery located close to the mouth of the estuary of BI*, together with the ship traffic in its port area. During the lockdown, SO2 showed a reduction of −0.8 µg·m−3 in the urban sites and stayed constant in the rural ones. This represents a decrease of 21% in the cities for inter-annuals, which is in accordance with the estimated average 20% reduction in the activity of the refinery and the reduction (−18% to −25%) of the general industrial activity, based on their energy consumption. Traffic reduction in the port of Bilbao during the lockdown (−8.1% with respect to the same 2019 March–June period), could also have contributed to the observed SO2 decrease.The PM10 lockdown average concentrations, as for the case of SO2, showed a moderate reduction in the urban and rural environments of the B.C. The ensemble average in the three urban sites during the period February–June 2015–2019 was 16.5 µg·m−3 and 11.2 µg·m−3 in the rural one. During the lockdown, PM10 decreased an average of −2.4 µg·m−3 and −1.7 µg·m−3 in the rural and urban environments, respectively. This represents a small reduction of 10% in the cities, which is lower than the observed one for SO2, and a larger reduction of 21% in the rural PM10 for inter-annual values. Reduced road traffic emissions, less construction/demolition activities, and the more modest industrial activity decrease (approximately −20%) could be the cause of the observed decrease. Thus, during the lockdown the decrease of the PM10 concentrations affected the rural environment more, resulting in an increased urban-to-rural concentration gradient. The reduction did not affect all cities or all rural sites equally: the lowest reduction of 3% in all the cities and 10% in all the rural sites corresponded to DS* and MU, respectively. Both monitoring sites are located at the seashore, and the sea salt aerosol could have played a role in the observed smaller reductions [11]. Sea salt aerosol contributed to a fraction of the total concentration, which probably did not change during the lockdown for inter-annual values, as can be inferred from the absence of wind anomalies in the coastal area of the B.C. (Figure 3). PM2.5 average concentrations during the lockdown stayed constant without significant changes for inter-annual values in both urban and rural environments of the B.C. For PM2.5, the ensemble inter-annual average in the three urban sites during the period March–June was 8.4 µg·m−3 and 5.8 µg·m−3 in the three rural sites. During the lockdown, similar concentrations were found both in the ensemble urban (8.3 µg·m−3) and rural (6.0 µg·m−3) averages. Moreover, concentrations remained similar at every specific site during the lockdown for their inter-annuals, except for the rural site MU which increased slightly (+13%) over a relatively low inter-annual value (6.2 µg·m−3). As shown in Section 3.2, wind velocity during the lockdown, which controls the ventilation/dispersion of pollutants, was close to the 2015–2019 average conditions. Thus, in a first approach, the lockdown values could be directly comparable with their inter-annuals without a wind velocity correction. After observing the quasi-persistent PM2.5 concentrations, with respect to inter-annuals, it was determined that urban and interurban road traffic could not be the main source for the observed PM2.5 concentrations in the B.C. because road traffic decreased significantly (49–53%). A significant PM2.5 reduction would have had a major impact on the cities, which was not the case. The observed reduction could only be attributed to a main PM2.5 source that did not change so much during the lockdown. In this respect, a spatial and sectorial source allocation study performed in 150 European cities, including the Greater Bilbao, using an adapted chemical transport model [37], showed that road transport contributes only 7% of the PM2.5 in Bilbao, while the industry contribution is 46%. The rest is distributed among agriculture (16%), residential (4%), natural (12%), and other minor sources. These percentages show high variability among the reported cities. For the case of Bilbao, which shows one of the largest industrial contributions of the European cities, more than 50% of the PM2.5 (primary and secondary) originated inside the Greater City, which could be helped by adopting the efficient local reduction policies on PM2.5 concentrations. During the lockdown, the main decrease affected road traffic, and industry showed a lower decrease, whereas, the activity of power plants increased (Section 3.1). Taking into consideration road traffic and industry only, as agriculture did not change significantly during the lockdown because the activity of agri-food companies was guaranteed, including crop and livestock holdings, we could expect a decrease of −1.2 µg·m−3 in Bilbao during the period, but the actual decrease was only −0.2 µg·m−3. The impact of specific meteorological conditions such as less rainfall and more insolation than during average March–June (2015–2019) meteorology (Figure 3), could have played a role in the more efficient formation of secondary aerosol, which could increase PM2.5 concentrations to the observed values [11]. The more uniform urban/rural PM2.5 distribution in the B.C. with respect to the PM10 can be attributed both to the location of their main primary and precursor sources (industry, road traffic, and domestic, among others) inside or close to the greater cities and their lifetime differences with PM10, which allowed a more efficient transport of PM2.5 to the more remote rural sites. The ensemble O3 average concentration of the three urban environments of the B.C. during the lockdown did not show significant changes with respect to the inter-annual value, though important differences were appreciated among the three cities. Inter-annual O3 concentrations in the urban sites showed an average value (60.1 µg·m−3) below one of the rural sites (82.1 µg·m−3) with the lowest concentrations in BI* (51.5 µg·m−3), due to a greater effect of the O3 titration with respect to DS* (62.7 µg·m−3) and VG* (66.2 µg·m−3). During the lockdown, a significant increase was observed in BI* (+11%) and a moderate decrease in VG* (−7%) and DS* (−2%), which accounts for the observed quasi-unchanged average of the three sites (60.2 µg·m−3) concerning the inter-annual average (60.1 µg·m−3). On the contrary, a more uniform decrease (approximately −11%) was observed in the three rural environments making the differences between urban (60.2 µg·m−3) and rural (73.2 µg·m−3) sites smaller. The observed changes in the urban concentrations of O3 during the lockdown mimics the weekend effect in the three cities out of the lockdown period, as discussed in the next section.Summarizing the results shown in Table 6, the average air pollutant concentrations during the March–June 2020 lockdown period with respect to the March–June 2015–2019 concentrations show different trends in urban and rural sites depending on both site and pollutant. At the urban sites, average concentrations of most air pollutants trend to decrease affecting NO and NO2 (−53% and −45%, respectively) more, SO2 and PM10 less (−21% and −10%, respectively), with little change for PM2.5 (−1%) and O3 (0%). The highest concentrations during the lockdown, as for the inter-annual averages, were registered in BI* and the lowest in VG*, except for O3, which showed a reversed behavior with a minimum at BI*. The observed decrease of pollution in the rural sites during the lockdown affected NO (−20%) and NO2 (−41%) resulting in very low concentrations of 1.5 µg·m−3 and 3.7 µg·m−3, respectively. The reduction was also important for SO2 and PM10 (−21%), moderate for O3 (−11%), and non-significant for PM2.5 (−1%).The urban traffic decrease had a significant impact on the observed reduction of NOx concentrations, as discussed in Section 3.3. PM10 and O3, among other pollutants, were also affected by traffic in the cities. NO2 weekly series for BI* (Figure 4a) and VG* (Figure 4b) are shown during both the lockdown period and the corresponding 2015–2019 period together with the percentage change in urban traffic during the lockdown in each city (dashed lines). The selection corresponds to the environments with the highest (BI) and the lowest (VG) pollution levels of the three cities. The NO2 full lockdown period average decrease in the cities (Table 6) is also well represented at the weekly scale for the full period. The lowest NO2 weekly average concentration in BI* corresponded to w7 with 11.1 µg·m−3, which was also a minimum in VG* (4.0 µg·m−3). However, these lowest concentrations are concurrent with the first week of the relaxation of the confinement and the mobility limitations, with the corresponding traffic recovery in the cities. The lower w7 values are more related to a period of favorable ventilation conditions with intense westerly winds, overcast conditions, and rain on the northern coast of Iberia. It is important to notice that, due to their respective rate of emissions, the lowest 11.1 µg·m−3 in BI* are above most of the weekly averages in VG* during the whole lockdown period (w2 to w14). The daily averages during the lockdown and the inter-annual 2015–2019 values with their standard deviations are also depicted for the same two cities BI* (Figure 4c) and VG* (Figure 4d). Inter-annual NO2 concentrations in VG* (dashed line) are well below the WHOAQG for the annual mean (40 µg·m−3) and during the lockdown they were even lower. However, the inter-annual concentrations in BI* are close to the WHOAQG; sometimes they exceeded it and, most of the time, the upper sigma interval is above the WHOAQG limit, at least until May-June, when its seasonal decrease is more evident. During the lockdown and after the road traffic decrease, the NO2 24-h concentrations in BI* sank below the lowest values of the normal variability of the inter-annual values (gray shading in Figure 4c) and remained well below the WHOAQG values.As in Figure 4a,b, for NO2, the weekly PM10 series for BI* and VG* are represented in Figure 5c,d. The generalized decreased concentrations observed in the three cities during the full lockdown period (Table 6) was also registered for all the weekly averages in Figure 4a,b, except for weeks w1–2, when the full lockdown and mobility restrictions were already active, and w11 during the relaxation stage. In these three weeks, simultaneous increases above the inter-annual values were registered in three cities. As discussed in Section 3.5, these periods were concurrent with African dust episodes. The O3 series for BI* and VG* are shown in Figure 5a,b. O3 concentrations in VG* during the lockdown (Figure 5b) showed a general trend to decrease with respect to the weekly inter-annual values in the same panel, except for weeks w2–3 and w10–11, which were concurrent with transport episodes of ozone from peripheric regions and under specific meteorological conditions (Section 3.5). In contrast, weekly ozone in BI* (Figure 5a) showed a trend to increase along the period of highest road traffic decrease in the city (w1-to-w12 in Figure 4a), with the exception of w7, with the above-mentioned exceptional meteorology, which resulted in a highly depleted O3 photochemistry. The same weekly series in DS* (not represented in the figure) showed no significant changes during the lockdown for inter-annual values. Thus, every city showed a different response in O3 concentrations to reduced traffic, despite being subject to the same restrictions and being located a short distance from each other (60–80 km). This may be related to the O3 chemical regime in each city. After interpreting their respective NOx pollution levels, discussed above, and the O3 response to the traffic reduction, we interpreted the O3 increase in BI* as an indicator of O3 formation in NOx-saturated environments. In the same way, VG* is NOx-limited and DS* can be either limited or saturated depending on the specific situation. Unfortunately, the lack of NMVOC monitoring limited our capacity to fully interpret O3 changes at the urban locations. The analysis of the weekend effect in the three cities during the recent historical (2015–2019) series shows that O3 changes during the lockdown followed similar (positive–negative) trends than those observed between weekdays (Monday-to-Friday) and weekends (Saturday and Sunday) of the historical series. Figure 6 shows the hourly averages of O3 during the mean week (Monday to Sunday) of the lockdown period (solid red line) together with the corresponding inter-annual averages (dashed black line) and the standard deviations (gray shading) for the three clustered urban stations BI* (b), DS* (c), and VG* (d): the hourly O3 concentrations in BI* during the lockdown were, most of the time, above the represented range of the five-year inter-annual variability (gray shading); in DS* they were inside that range, while in VG* were below. In BI* these changes led to average increases of +7% and +5% in the MDA8 concentrations for the lockdown weekdays and weekends, respectively, compared to the inter-annual values (Figure 6a). Corresponding decreases of −8% and −10% were observed in VG*, while DS* showed the smallest changes (−3% and −1%). Similarly, the historical (2015–2019) weekend changes (Figure 6b) record an average increase in BI* (from 73.6 to 76.8 µg·m−3), a decrease in VG* (from 86.0 to 84.9 µg·m−3) and almost constant values in DS* (82.4–82.9 µg·m−3). Weekly averages of O3 during the lockdown, discussed in the preceding section (Figure 5), show some weeks of ”anomalous” high O3 concentrations above the inter-annual averages in the more “rural” of the three cities (VG*), which showed a generalized trend to decrease after the road traffic decrease of the lockdown (weeks w2–w3 and w10–w11). “Anomalous” high PM10 concentrations were also recorded for the weeks w1–w2 and w11 in the three cities. On the contrary, the NO2 weekly averages were kept well under their inter-annual values in all locations during the whole lockdown as shown in Figure 4 for the most polluted city (BI*) and the least polluted (VG*) in the B.C. These anomalous periods for O3 and PM hide transport episodes from outer regions and with a different origin, as discussed next. A representation of the series at a (daily) higher resolution uncovers the real length and intensity of the detected episodes and it shows new episodes of short duration out of the signaled weeks.A depiction of the daily time series of the concentrations of O3 and PM helps to better discriminate the episodes, which for the case of O3, lasted only an average of 3–4 days in the B.C. [19,20]. Thus, the recorded concentrations have been compared with the WHOAQGs. Figure 7 shows the MDA8 O3 concentrations during the lockdown and a short period of 4 weeks before (pre-COVID-19) at the urban sites. Five O3 episodes of an average duration of 3–4 days and with the daily 8-h standard above 100 µg·m−3 were recorded simultaneously in all urban sites (episodes numbered 1 to 5 in the figure). The simultaneous exceedance of the standards was selected for the identification of both O3 and PM transport episodes. The same O3 episodes were also recorded at all rural sites (Figure S1), which reached even higher concentrations and exceeded the long-term objective of the daily 8-h standard in the EU (120 µg·m−3) during episode number 5.All episodes occurred with an upper-ridge extending from northwestern Africa to the Iberian Peninsula and a high-pressure surface, which expanded over the Bay of Biscay and the British Islands and France (Figure 7). This forced easterly winds in the marine boundary layer over the northern coast of Spain as well as sea-breeze convergence into the northern coast. As discussed in Section 2.1 and Section 3.2, the described synoptic scenario was behind most of the O3 episodes in the B.C. [19], and the main fraction of the O3 impacting the B.C. during these events could be attributed to transport across southern France. [21] concluded that local emission reduction policies would have a limited effect on the reduction of O3 levels in the B.C., as has been confirmed during the lockdown. Out of these “short” episodes of O3 transport during the lockdown and during a relatively long period (w4 to w8) its concentration in the rural/urban sites remained below the inter-annual averages and even below the lower sigma interval depicted in shaded gray color in Figure 7 and Figure S1. The city of BI, the most NOx-saturated environment of the three cities, and, to a lesser extent, DS, showed concentrations close to the inter-annual averages during the same period w4-w8, which mimics a rise in the O3 concentrations similar to the weekend effect in BI. Consequently, the ensemble urban average (60.2 µg·m−3 in the three cities) of the whole lockdown period did not show changes for the inter-annual values (60.1 µg·m−3), hiding the different behavior of each city. Similarly, the rural ensemble background average (73.2 µg·m m−3) showed a decrease of −8.9 µg·m m−3 (−11%) for inter-annuals (Table 6), also hiding episodes more severe than in the cities.The observed changes during the lockdown reinforce these results: (1) the relatively high impact on the whole B.C. of the O3 importation from the continental Europe during episodes (rural and urban areas), and (2) the high local O3 anomaly in the city of Bilbao, which seems to be related with a NOx-saturated area. The city kept out of the NOx- sensitive conditions even after the reported emission reduction during the lockdown. In this respect, it is important to note that the observed behavior of the O3 averages during the low emission scenario of the lockdown in the three cities (Table 6) reproduced the observed weekend effect of the three cities (Section 3.4): BI* recorded higher O3 concentrations, DS* did not show significant changes, and VG* showed a decrease.Figure 8 shows the 24-h average concentrations of PM10 at the three main urban sites. Horizontal red lines mark the WHOAQG values for the PM10 concentrations: 20 μg· m−3 for the annual mean and 50 μg·m−3 for the 24-h mean. Up to three PM10 events (numbered 1 to 3 in the figure) with the daily averages above 20 µg m−3 were recorded simultaneously in all urban sites during the whole period. Simultaneously high concentrations can be related to transport from distant sources. In this case, they all were concurrent with desert dust outbreaks from northern Africa after the development of large Rossby waves over the region, and/or after the evolution of isolated low-pressures moving to northwestern Africa, as shown in Figure 8. Out of these events occurring in weeks w1–w2, w8, and w12, which pushed the 24-h average concentrations above the inter-annual values, PM10 concentrations were slightly under those averages and near the lower boundary of the sigma intervals. As a result, the full period averages also remained under the inter-annual values (Table 6) in the urban environments of the B.C., with a larger decrease in VG* (−16%), moderate in BI* (−12%), and with almost no variation in DS* (−3%). In contrast to the O3 concentrations, which show greater average values in the rural sites in comparison to the urban ones (Table 6), PM10 concentrations in the rural environments show values well below those registered in the urban sites during the lockdown. There is a complete absence of concentrations above the reference 20 µg· m−3 level, except for the MU location during week w1 (Figure S2). The more important decrease in PM10 with a more local regional origin in the three rural sites in comparison to the urbans has prevented the African transport events 1, 2, and 3 from causing the PM concentrations to exceed the level of 20 µg·m−3. The two extreme PM10 peaks during the pre-COVID-19 period (weeks w−2 and w−3) in Figure 8 and Figure S2, corresponded to a wildfire in Tineo (Asturias), west of B.C., for the w−3 episode, and an African dust episode simultaneously with a wildfire for the w−2 episode.The 24-h average concentrations of PM2.5 of the three urban sites are shown in Figure 9. The recorded concentrations during the lockdown were close to the inter-annual averages in the figure, and most of the time, they remained inside the sigma interval depicted in shaded gray color. The horizontal red lines mark the WHOAQG values for PM2.5: 10 μg·m−3 for the annual mean, and 25 μg·m−3 for the 24-h mean. Similarly, for the identification of the PM10 episode, and using the annual WHOAQG for PM2.5 (10 µg·m−3), the same events registered in weeks w1–w2, w8, and w12 for PM10 emerge for the fine fraction (marked with an arrow in the figure), with the daily average above the WHOAQG simultaneously in all urban sites. In addition, more simultaneous peak events can be identified for weeks w3, w4, w6, w10, and w11 (not arrowed in Figure 9), which are not present in the PM10 record because they did not reach enough concentration in the three cities. Following the list of episodes in MITECO (2021), European sulfates were behind the first three PM2.5 events (w3, w4, and w6), and w10 was an African dust episode. None of them were identified in the PM10 records. The recorded African dust episodes during weeks w1–w2 and w12 were also identified by the system together with concurrent PM from wildfires. All the episodes identified by the surveillance system for the B.C. during the period are clearly represented in the PM2.5 records and vice-versa with the exception of the peak in week w11, which was recorded simultaneously in the three cities but not identified by the system. This unexplained peak is concurrent with an O3 transport episode discussed above, and it could be related with (non-sulfate) secondary aerosols generated by photochemical reactions, and transported together with O3 into the B.C. Similar high 24-h averages, close to inter-annual values, were observed for PM2.5 in the rural sites (Figure S3). The observed relatively high daily PM2.5 concentrations are in line with the calculated period average concentrations in Table 6, which show small changes with respect to the inter-annual values, both in rural and urban environments. Contrary to PM10, which showed a more drastic reduction in concentration in the rural sites during the lockdown, the more uniform distribution of PM2.5 (Table 6) contributed to reaching the reference concentration (10 µg·m−3 ) in the three rural (Figure S3) and urban sites (Figure 9) during weeks w1-w2 and w11. The rest of the recorded PM2.5 peaks laid, approximately, at the reference 10 μg·m−3 level, but without exceeding it simultaneously in the three rural sites. The two extreme PM2.5 peaks during the pre-COVID-19 period (week w−2) in Figure 9 and Figure S3 correspond to the same wildfire and African dust episodes also observed in the PM10 records (Figure 8 and Figure S2).We have evaluated the impact of the lockdown during the COVID-19 pandemic for the three main urban sites of the B.C. region and their adjacent rural sites during the period of confinement and mobility restriction (14 weeks). We selected 16 stations of the Air Quality Monitoring and Control Network which represented the whole territory. As described in many other regions of the world including the rest of Spain [11], the concentrations of most pollutants decreased in the B.C. due to COVID-19 restrictions, with the urban and interurban road traffic of the region falling an average of 49% and 53%, respectively, and a lower reduction of approximately 20% of the general industrial activity. The observed changes of the full-period averages, however, were not uniform throughout the territory, and they varied with each pollutant. In addition, the whole territory was repeatedly affected by aerosol and O3 episodes transported from outer regions and added to the local pollution. They occurred under specific meteorological conditions.At the urban sites, the observed decrease of pollution during the lockdown affected NO and NO2 more than any of the other registered pollutants (−53% and −45%, respectively) resulting in a very low concentration for the ensemble urban NO2 average (12.4 µg·m−3) well below the WHOAQG for the annual average of 40 µg·m−3. In the rural sites, a similar reduction affecting the already low inter-annual values of these sites resulted in very low concentrations of both NO (1.2 µg·m−3) and NO2 (2.2 µg·m−3). The observed decline was more in accordance with the reported road traffic decrease, which has a direct impact on the urban NOx, and less with the more limited reduction of the industrial activity as discussed above. The decrease was even more moderate for SO2 (−21%) in both rural and urban environments, in line with the moderate reduction of the general industry, although this decrease affected their already very low inter-annual concentrations (3.1 and 1.5 µg·m−3 for the urban and rural environments, respectively). During the lockdown, PM10 decreased moderately (−10% and −21% in the urban and rural environments, respectively). Inter-annual period averages in BI* (15.5 µg·m−3) are close to the WHOAQG for the annual averages (20 µg·m−3). Alternatively, during the lockdown, the most uniform concentration distribution was registered for PM2.5 (8.3 µg·m−3 and 6.0 µg·m−3 in urban and rural sites) and it also showed the lowest changes with regards to pre-pandemic values (−1% and +3% at the urban and rural sites, respectively). The greater average concentrations in BI* (9.5–9.3 µg·m−3) places this location very close to the WHOAQG for the annual averages (10 µg·m−3), even during the lockdown. Apart from the relative weight of the local industry emissions, which could have contributed to the observed modest changes in the PM2.5 concentrations, the impact of the documented meteorological anomalies during the period (less cloud cover than during the average March–June 2015–2019 period) could have also contributed to the more efficient formation of secondary aerosol, and the subsequently relatively high PM concentrations. O3 concentrations showed low changes in the inter-annual averages in the cities (ensemble average 0%) but, unlike PM2.5, showed important differences among the three cities: BI* increased (+11%), DS* remained almost unaltered (−2%), and VG* decreased (−7%). These changes occurred in the same direction as the weekend effect already observed in historical data. The changes in the rural sites showed a moderate decrease (−11%) from the relatively high average concentrations (82.1 µg·m−3) of the inter-annual records. This is in line with lower emissions of anthropogenic precursors. The widespread decrease in the rural sites did not avoid the reported five O3 episodes during the period and the exceedance of the long-term objective of the daily 8-h standard (120 µg·m−3) in all these sites. Simultaneous exceedances of the WHOAQG for the MDA8 of the O3 concentrations were also observed in the three cities. These episodes are attributed to transport from France of important background levels of ozone under anticyclonic conditions, which added to the local production. We did not find exceedances of the WHOAQG for either PM10 and PM2.5 24-h average concentrations during the lockdown. However, frequent episodes of consecutive PM2.5 daily values above the annual WHOAQG were identified and less frequent for the PM10. African dust transport, wildland fires, and anthropogenic European sulfates, together with more frequent spells of favorable meteorology for secondary aerosol formation at the local regional scale, seem to have been behind the observed exceedances.The observed air quality response to the activity and road traffic emission reductions during the lockdown period suggests that the results of having a high proportion (above 50%) of the vehicle fleet switching to electric in the near future will affect the air quality of the B.C., based on the NOx emission reductions. However, the reduction of fine and ultrafine fractions of particles, due to the use of electric vehicles, important from the point of view of their number, not their mass, can mean a non-negligible improvement from a health point of view [52,53]. Sustained improvements in the PM10, PM2.5, and O3 of the region will require actions across various sectors, including industry, together with inter-regional European initiatives to establish pollution control strategies based on inter-regional policies. The limited number of monitoring stations used for the current study does not represent the whole region in full, though the selected clustered urban sites characterize the main three core cities and their peripheral commuting zone, where the main fraction of the B.C. population resides. Thus, the caveat of our assessments on the impact of the COVID-19 lockdown upon the air quality is that outside the reach of the major cities there may be smaller villages where the relatively low decline in the local industrial activity could have resulted in non-significant changes in the local air pollution. This could be important when considering diffuse emissions, odor impacts, or pollutants not registered in the monitoring network.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111042/s1, Figure S1: As in Figure 7, MDA8 ozone daily series during a pre-COVID19 four-week period the lockdown period (solid red line) and during in the three background rural sites, together with the inter-annual averages (dashed lines) and their standard deviations (grey shading). The five simultaneous O3 exceedances (marked with arrows) of the WHOAQG of 100 μg·m−3 are also detected in the three rural sites. Figure S2: As in Figure 8, 24-hour average concentration series of PM10 during pre-COVID four-week period and the lock-down period (solid red line) in the three background rural sites, together with the inter-annual averages (dashed lines) and their standard deviations (grey shading). The PM10 episodes observed in the urban sites (marked with arrows) above the WHOAQG reference level of 20 μg·m−3 are not detected in the rural sites. Figure S3: As in Figure 9, 24-hour average concentration series of PM2.5 during a pre-COVID four-week period and the lockdown period (solid red line) in the three background rural sites, together with their inter-annual averages (dashed lines) and standard deviations (grey shading). The same three PM10 episodes marked in Figure 8 and Figure 9 (arrows) can also be detected in these PM2.5 series (above the WHOAQG of 10 μg·m−3), but not in all three rural sites.Conceptualization, G.G., M.d.B., M.C.G., A.R.-G., E.T.-P., E.G.-R., E.S.d.C., I.Z., J.A.G. and V.V.; methodology, G.G., M.d.B., M.C.G., A.R.-G., E.S.d.C., I.Z. and J.A.G.; software, A.R.-G.; validation, A.R.-G. and E.T.-P.; formal analysis, G.G., M.d.B., M.C.G., A.R.-G., E.T.-P. and J.A.G.; data curation, A.R.-G. and M.C.G.; writing—original draft preparation, G.G., M.d.B., M.C.G., A.R.-G. and E.S.d.C.; writing—review and editing, G.G., M.d.B., M.C.G., E.T.-P., E.G.-R., E.S.d.C., I.Z., J.A.G. and V.V.; visualization, G.G., M.d.B., E.S.d.C. and I.Z.; supervision, G.G.; funding acquisition, G.G. All authors have read and agreed to the published version of the manuscript.This research was funded by the Basque Government and the University of the Basque Country (GIC15/152 and GIU13/03) and by the Environment Vice-Department of the Basque Government for the measurement of biogenic volatile organic compounds in Valderejo Natural Park.Not applicable.Not applicable.All the public data and software used are contained in sources cited in the body text and in the references.The authors wish to thank the Basque Government and the University of the Basque Country (UPV/EHU) as the source of our main financial support: GIA consolidated Research Groups GIC15/152 and GIU13/03. Additional financing was provided by the Environment Vice-Department of the Basque Government for the measurement of biogenic volatile organic compounds in Valderejo Natural Park. These financing bodies have played an exclusively economic role in the study. We also want to thank the Department of Economic Development, Sustainability and Environment of the Basque Government and the Bilbao City Council for providing air quality data and road traffic information, respectively. Hersbach, H. et al., (2018) was downloaded from the Copernicus Climate Change Service (C3S) Climate Data Store. The results contain modified Copernicus Climate Change Service information 2020. Neither the European Commission nor ECMWF is responsible for any use that may be made of the Copernicus information or data it contains.The authors declare no conflict of interest.(a) Location of the Basque Country in Spain; (b) Topographic map. Colored scale indicates altitude (m) above the mean sea level (a.m.s.l.), bold lines separate the provinces (Bizkaia, Gipuzkoa, and Alava), main urban areas are shaded in gray and main roads are marked with thin solid lines. Measurement sites correspond to provincial capitals, clustered urban sites: Bilbao (BI*), Donostia-San Sebastian (DS*), and Vitoria-Gasteiz (VG*), and rural sites: Mundaka (MU), Pagoeta (PA), and Valderejo (VA); (c) Monitoring sites for each clustered urban traffic station (BI*, DS*, VG*).(a) Location of the companies registered in the European Pollutant Release and Transfer Register (EPRTR), Basque Country (2018), main urban areas, and roads. (b–e) Estimated annual emissions of NMVOC, NOx, PM, and SOx, respectively, from companies registered in the EPRTR.Temperature, cloud cover (a,b), mean sea level pressure, and wind anomalies (c,d) during the 2020 lockdown period with respect to the 2015–2019 corresponding period.(Left) Weekly NO2 averages during the March–June (2020) lockdown period and the corresponding 2015−2019 inter-annual averages for the urban BI* (a) and VG* (b) monitoring stations. Dashed lines in the same panels represent the percentage change in weekly traffic during the lockdown compared to week 0. (Right) NO2 daily averages during the lockdown (solid line) and the inter-annual averages (dashed) with the standard deviation (gray shading) for the same two cities BI* (c) and VG* (d).(Left) Weekly O3 averages during the March–June (2020) lockdown period and the corresponding 2015–2019 inter-annual averages in the urban BI* (a) and VG* (b) monitoring stations. (Right) Same type of representation for the PM10 weekly averages in BI* (c) and VG* (d).Daily maximum 8−h average (MDA) O3 values for weekdays, weekends, and the full period are shown in the table for the lockdown period and the corresponding years 2015–2019 in the three cities (a). The hourly averages of ozone during the mean week (Monday to Sunday) of the lockdown period are represented (solid red line) together with the corresponding inter-annual averages (dashed black line) and the standard deviations (gray shading) for BI* (b), DS* (c), and VG* (d).MDA8 ozone daily series during a pre−COVID19 four-week period and the lockdown period (solid red line) in the three urban sites. The series is represented in Figure 4c,d for NO2. Inter-annual averages (dashed) with their standard deviation (gray shading) can be compared with the lockdown daily values (solid red). Five simultaneous O3 exceedances of the WHOAQG of 100 μg·m−3 (horizontal red line) are marked (arrows) during weeks 2, 3, 9, 10, and 11 of the lockdown period. The synoptic forcings of the respective episodes are shown (upper panels) in the Climate Forecast System Re-analysis (CFSR), one panel per episode: the 500 hPa geopotential heights (gpdams) and mean sea level pressure (MSLP) contours (hPa) at 12:00 UTC are depicted (Source: Wetterzentrale).24-h average concentration series of PM10 at the three main urban sites. As in Figure 7, inter-annual values (dashed lines) with their standard deviation (gray shading), and the lockdown daily values (solid red) are represented. Three PM10 episodes, above the reference level of the WHOAQG of 20 μg·m−3, for the annual average concentration (horizontal red line), are marked (arrows) during weeks 1–2, 8, and 12 of the lockdown period. The respective synoptic forcings are shown (upper panels) in the Climate Forecast System Reanalysis (CFSR), one panel per episode, as in Figure 7.24−h average concentration series of PM2.5 at the three main urban sites. As in Figure 8, inter-annual values (dashed lines) with their standard deviation (gray shading), and the lockdown daily values (solid red) are represented. The same three PM10 episodes marked in Figure 8 (arrows) can also be detected in the PM2.5 series (above the WHOAQG of 10 μg·m−3). More episodes above the reference line (10 μg·m−3) can also be observed. The origin, date, and the pandemic week reference of each episode are shown in the upper panel table, as informed by the surveillance program of the Spanish Government, based in the Navy Aerosol Analysis and Prediction System (NAAPS) by the U.S. Naval Research Laboratory (NRL).Selected urban background, traffic and rural sites, corresponding stations, and measured pollutants. Clustered urban background and traffic sites are marked with an asterisk *.Weekly periods studied and mobility and confinement restrictions applied in the Basque Country.Corresponding selected pre-lockdown and lockdown weeks for the years 2015–2019 and 2020.Contribution (%) and variation (%) of mobility in 2020 compared to 2019 in the three provinces of the Basque Country during the period studied (w0-w14) in different areas: interurban and urban traffic, air goods, passengers, and port traffic. Large reduction (from −100.0% to −25.0%), reduction (from −24.9% to −10.0%), small reduction (from −9.9% to −0%).Variation of the energy consumption (electricity and natural gas) by main sector in 2020 relative to 2019, for the March–June average. Large reduction (up to −25.0%), reduction (from −24.9% to −10.0%), small reduction or increment (from −9.9% to 10.0%), increment (above 10.1%).Average and ± standard deviation of NO, NO2, SO2, PM10, PM2.5, and O3 (μg·m−3) concentrations during the lockdown period (2020) and previous years (2015–2019) at clustered urban traffic sites (marked with an asterisk) and selected rural sites in Basque Country. Ensemble urban and rural averages are also included. The percentage of reduction or increment during the 2020 lockdown with respect to 2015–2019 is included in shaded colors: Large reduction (up to −25.0%), reduction (from −24.9% to −10.0%), Small reduction or increment (from −9.9% to 10.0%), increment (above 10.1%). Clustered urban background and traffic sites are marked with an asterisk*.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11043.txt

@@ -0,0 +1 @@
+Gender norms prescribe domestic labor as primarily a female’s responsibility in developing countries. Many domestic tasks depend on access to water, so the physical, emotional, and time demands of domestic labor may be exacerbated for women living in water-insecure environments. We developed a set of domestic work experience (DWE) measures tailored to work in rural areas in developing countries, assessed rural Nigerian women’s DWE, and examined relationships among the measures. Interviewer-administered survey data were collected between August and September from 256 women in four rural Nigerian communities. Latent factors of DWE were identified by analyzing survey items using confirmatory factor analysis. Pearson’s correlation was used to examine relationships among latent factor scores, and multivariate linear regression models were used to determine if factor scores significantly differed across socio-demographic characteristics. The DWE measures consisted of latent factors of the physical domain (frequency of common domestic tasks, water sourcing and carriage, experience of water scarcity), the psychosocial domain (stress appraisal and demand–control), and the social domain (social support). Significant correlations were observed among the latent factors within and across domains. Results revealed the importance of measuring rural Nigerian women’s DWE using multiple and contextual approaches rather than relying solely on one exposure measure. Multiple inter-related factors contributed to women’s DWE. Water insecurity exacerbated the physical and emotional demands of domestic labor DWE varied across age categories and pregnancy status among rural Nigerian women.Despite growing female participation rates in the workforce and the growing contribution of women to economic growth and gross domestic products (GDP) globally, the fact remains that women are still primarily responsible for domestic and care work [1]. Millions of women in rural Asia and Sub-Saharan Africa are estimated to spend 10–17 h per day performing domestic work [2,3]. In developing countries, domestic tasks entail engaging in demanding tasks such as cleaning, food preparation and cooking, water fetching, manual washing of dishes and clothes, child/elder care, and engaging in subsistence agriculture [4,5,6]. For example, in Gujarat, India, women spend between three to four hours of their day collecting only water [7]. Ownership of modern household appliances is often limited in rural sub-Saharan Africa. Thus, domestic tasks such as food preparation can be long, often strenuous, involving tasks such as water fetching, heavy lifting, pounding, grinding, and cooking for long hours [8]. Water insecurity further worsens the burden of domestic labor. Water infrastructure is poor, and access to water is unreliable in many rural communities in sub-Saharan Africa, including Nigeria. Nigeria had the second-highest proportion of women spending more than 30 min/day on water collection in a study analyzing the time spent on water collection labor across 24 Sub-Saharan African countries [9]. Since most domestic tasks depend upon water, the physical, emotional, and time demands of cooking and other domestic work may be exacerbated for water-insecure households in low-income countries (LIC). Performance of many domestic tasks for those who depend heavily on rainwater storage and rainfall-recharged wells in rural Nigeria can become problematic during drought periods [9,10]. The responsibility for procuring water in these water-insecure environments rests on women who spend much time and physical effort on water collection [9,11].Beyond economic and geographic influences, socio-cultural norms and gender ideologies impact household division of labor and women’s domestic work experiences. The traditional gender role theory described by Parson and Bales (1955) delineated the paid work domain and instrumentality as important for men, with the home domain and expressiveness as integral for women [12]. This theory on societal perception of domestic responsibilities is still relevant today in 21st-century Nigeria. Akanle and Ejiade (2012) aptly described gender roles in a traditional Nigerian society [13]. In Nigeria, performance of domestic labor is viewed as the ‘woman’s domain’ regardless of women’s engagement in paid employment outside of the household [14]. In Yoruba Nigerian culture (the primary study population in this paper), a housewife with long leisure time is considered lazy. Additionally, a housewife’s sense of pride and accomplishment is often linked to her resourcefulness and ‘busyness’ within the home [15]. Likewise, a man performing normative roles may be perceived as feminine (e.g., cooking) or called a ‘women wrapper’ in traditional, rural Nigeria society [16]. Thus, socio-cultural norms may influence women’s domestic experiences, their bargaining power over the time allocated to domestic labor, and male involvement in domestic responsibilities [17].Domestic labor and water insecurities have mainly been studied as separate issues affecting women’s physical and psychosocial health. However, water carriage, which is a consequence of water insecurity [18], and high domestic work burden are associated with musculoskeletal pain (MSP) among women [19]. As a result, evidence is still lacking on how these risk factors are interrelated, specifically on how the experiences of water insecurity could influence the performance of tasks and women’s domestic work experiences (DWE). Measures commonly used to define DWE include: the number of working hours [20,21,22,23,24], the level of familial support with domestic responsibilities [25,26,27], the biomechanical demands of domestic tasks [5,28,29,30], perceived stress and demands of work [31], and the frequency of task performance per week or self-reported ‘intensity-ratings’ of specific domestic tasks [32,33,34]. Identifying the interdependencies between these DWE measures, including other measures such as water insecurity and water carriage, could improve understanding of rural women’s DWE and related health outcomes.Furthermore, most studies on women’s DWE originated in developed countries and considered the synergistic effects or consequences of paid and unpaid work demands [35,36,37]. DWE was assessed using ‘time’/‘frequency’ measures in these studies. DWE influence on physical and psychosocial health was secondarily evaluated, based on their capacity to confound the association between the risk factors from paid work and health or contribute to psychosocial strain from poor work-life balance among women residing in primarily urban environments [17,32,36,38].The consequence for engaging in domestic work can be far-reaching, beyond time lost. It influences women’s capacity to pursue productive economic activities, increasing physical and psychosocial stress, especially for those balancing domestic care work with paid employment [25,39,40,41]. Unpaid care work has been associated with school absenteeism among girls, thus perpetuating the cycle of poverty and gender inequality. Understanding women’s domestic work realities and the consequences of such work on women’s health will be crucial towards developing interventions focused on improving the redistribution of work responsibilities between men and women [41]. The purpose of this study was to develop DWE measures tailored to the context of work in rural areas of LICs, such as Nigeria, and assess women’s DWE. Domestic tasks are mostly performed manually, and water insecurity (which may increase physical stress) is prevalent in these environments. We hypothesize that physical, psychosocial, and social factors of living and working conditions all contribute to women’s DWE, and that DWE varies across socio-demographic variable categories (Figure 1). These DWE, in turn, could impact women’s physical and mental health (Yellow Box, Figure 1).Physical Factors [Green Box, Figure 1]: Long hours spent on domestic work, including the difficulty and frequency of domestic tasks, contribute to DWE by leading to work overload, increased time pressure, reduced opportunities for rest and recovery, and increased perceived stress [21,35]. Furthermore, the performance of tasks in extreme postures, water insecurity, fetching, and carriage activities could influence the efficiency of task performance, contributing to DWE and physical stress.Psychosocial Factors [Orange Box, Figure 1]: Psychosocial stress results from the interaction between external demands, such as work responsibilities, and the cognitive appraisal of the capacity to cope with those demands [42]. The transactional model of stress theory posits that stress appraisal may influence stress reactivity, including coping strategies used to manage stressful situations. Likewise, women’s perception of domestic work responsibilities as stressful or not (cognitive appraisal) contributes to their DWE [43,44]. The demand–control model of occupational psychosocial stress posits that psychological job demands and decision latitude (level of autonomy or agency over work) influence perceived stress and negative stress reactivity [45,46]. Women who perform time-demanding domestic tasks over which they have low control/agency may experience psychological distress [17,47,48]. Social Factors [Blue box, Figure 1]: Social support and networks influence women’s capacity to adapt to stressful domestic work conditions and mitigate the adverse effect of strenuous domestic work burden on their physical and mental health [22,26,27,49]. Low social support or social networks reduces opportunities for responsibility-sharing and division of labor [50,51], while greater social support, especially instrumental support (division of domestic labor), could improve women’s DWE and health [27].Demographic Factors [Grey Box, Figure 1]: Situational and contextual determinants of health, such as socioeconomic status or life stage and pregnancy status, may moderate women’s DWE, including the exposure and reaction to domestic work stressors [49,52,53]. Although variability in levels of domestic work stress/exposure is not the focus of this study, the model proposes that the magnitude of domestic work exposure varies over time, life stage, season, and place. We hypothesized that there would be interrelatedness within and across the physical, psychosocial, and social factors of DWE and that sociodemographic characteristics would moderate DWE. In this study, we used confirmatory factor analysis (CFA) to test our hypotheses about the relationships among items representing the physical, psychosocial, and social domains of DWE for rural, low-income Nigerian women. This approach allowed us to identify the latent factors of women’s DWE, the extent of factor interrelatedness, and the factors that most influence DWE, including the direction of the inter-relationships.A cross-sectional study was designed to measure factors that contribute to DWE among women in rural Ibadan, Nigeria. Ibadan, the capital of Oyo state, is one of the largest cities in Africa in terms of landmass and one of the most densely populated Nigerian cities. The projected population by the year 2025 in the city is 5.03 million people [54]. There are 11 local governments in Ibadan, five in the metropolitan urban area, and the rest primarily made up of semi-urban/peri-urban and rural settlements. Increased population growth and rural–urban migration has led to the formation of peri-urban, semi-rural slums on the outskirts of Ibadan metropolis. These fringe communities live in the reality of abject infrastructural deprivation. For example, at least 60% of rural and semi-rural households in Oyo State, Nigeria, must fetch water from streams, rivers, and unprotected wells to meet their water supply needs [55]. We recruited 365 women of reproductive age (i.e., 18–49 years) from four neighboring rural communities in Lagelu and Akinyele Local Government Areas, Oyo State, Nigeria. These communities represented typical communities where cultural and gender norms promote domestic responsibilities as ‘women’s work’ and where household water access is low [13,55]. Item Development: A literature review was conducted between February and March 2019 to identify items from validated surveys for consideration in a DWE instrument. Items included those related to two primary areas: physical work and psychosocial work. Physical work included experience of water insecurity [56]; water sourcing and proximity to water source from the Joint Monitoring Program (JMP) core questions [57]; load carrying [58]; and posture and movements [29,59,60]. Psychosocial work included items related to decision authority and psychosocial job demands from the Job Content Questionnaire [61]. Other items related to physical work—the frequency of domestic work and water fetching and carriage—and psychosocial work—perceived difficulty of performing domestic tasks—were also developed. A third primary area was also developed related to social support. A questionnaire consisting of new and modified items from validated surveys was created. Cognitive Interviewing, Pretest, and Survey Administration: After the questions were developed, cognitive interviews were carried out among ten women to assess whether participants interpreted the meaning of the questions as intended and to determine if the response options were appropriate [62]. Modifications were made on ambiguous and culturally irrelevant items. The revised survey was subsequently pretested among 40 women of similar demographic characteristics as the study participants. The undeclared pretest (respondents not aware it was a pretest) was conducted to ascertain whether respondents were interpreting questions correctly, question variation, difficulty, flow, order, and time spent per question. After the pretest, modifications were made on ten questions related to physical work, water insecurity, and water carriage.Socio-Demographic Factors: The sociodemographic factors included in this study were age (18–25, 26–30, 31–35, 36 years and above), income (lowest quartile, median, and highest quartile), education (primary, secondary, and tertiary), household size (0–3 people, 4–6 people, >6 people), age of youngest child (above or below five years), youngest child is walking (yes or no), pregnancy status (yes or no), occupation (no paid work, semi-skilled labor, and skilled labor), and hours of paid work per week (continuous, but categorized using quartiles for analysis purposes). Physical Factors of DWE:Frequency of common domestic work: Participants were asked to select how frequently they perform each of 14 common domestics tasks using a 5-point Likert scale (never/not me, rarely (three times/month), sometimes (two or three times/week), every day, more than once a day).Time (h/week) on domestic tasks: Response entry for the total time spent on domestic tasks per week was categorized into three groups: lowest third (<23 h/week), middle (23–30 h/week), and the highest third (>30 h/week).Lifting and Loading: Six items were used to measure ‘water fetching and carriage practices’ during domestic tasks. These items included: responsibility for water collection labor (only me, me and others, others only); number of water trips per collection period recoded into four ordered categories (None, <5 trips, 5–8 trips, >8 trips); quantity of water carried per trip (None, <25 L, 25–30 L, >30 L); where carried water was placed (head loading, on hands, use of assistive device); and how water containers to be carried were lifted (no lifting, assisted, unassisted lifting). Apart from water containers, the frequency of carrying children while performing household tasks (never–always), where children are placed on the body (on the back, hip, arm), and the frequency of lifting other loads (>25 pounds) per week were assessed.Proximity to Water Source: Three items were used to measure water-source proximity. They included: ‘time taken to reach water collection point’ (none, ≤5 min walk, 6–10 min walk, >10 min walk), ‘time taken to complete a water collection trip’ (not applicable, <10 min, 10–20 min, >20 min), and ‘where water source/collection point is located’ (inside dwelling, around the compound, elsewhere).Experience of Water Insecurity: Three items were adapted and modified from the household water insecurity scale [56]. Participants were asked to rate how frequently they were worried, angry, or frustrated that they did not have enough water, or rationed water usage in homes in the past four weeks on a 5-point Likert scale (never to always).Frequency of common domestic work: Participants were asked to select how frequently they perform each of 14 common domestics tasks using a 5-point Likert scale (never/not me, rarely (three times/month), sometimes (two or three times/week), every day, more than once a day).Time (h/week) on domestic tasks: Response entry for the total time spent on domestic tasks per week was categorized into three groups: lowest third (<23 h/week), middle (23–30 h/week), and the highest third (>30 h/week).Lifting and Loading: Six items were used to measure ‘water fetching and carriage practices’ during domestic tasks. These items included: responsibility for water collection labor (only me, me and others, others only); number of water trips per collection period recoded into four ordered categories (None, <5 trips, 5–8 trips, >8 trips); quantity of water carried per trip (None, <25 L, 25–30 L, >30 L); where carried water was placed (head loading, on hands, use of assistive device); and how water containers to be carried were lifted (no lifting, assisted, unassisted lifting). Apart from water containers, the frequency of carrying children while performing household tasks (never–always), where children are placed on the body (on the back, hip, arm), and the frequency of lifting other loads (>25 pounds) per week were assessed.Proximity to Water Source: Three items were used to measure water-source proximity. They included: ‘time taken to reach water collection point’ (none, ≤5 min walk, 6–10 min walk, >10 min walk), ‘time taken to complete a water collection trip’ (not applicable, <10 min, 10–20 min, >20 min), and ‘where water source/collection point is located’ (inside dwelling, around the compound, elsewhere).Experience of Water Insecurity: Three items were adapted and modified from the household water insecurity scale [56]. Participants were asked to rate how frequently they were worried, angry, or frustrated that they did not have enough water, or rationed water usage in homes in the past four weeks on a 5-point Likert scale (never to always).Psychosocial Factors of DWE:Psychosocial Appraisal of Domestic Work Stress: Six items were used to measure respondents’ appraisal of domestic work. Three items each measured women’s positive and negative cognitive appraisal of domestic work responsibilities on a 5-point Likert scale (strongly agree to strongly disagree).Time Demand–Control: Two item-statements (freedom to choose when and how to perform tasks) were modified from the ‘decision authority’ and one item-statement (time pressure) was modified from the ’psychological job demand’ subscales of the Job Content Questionnaire (JCQ) [61], constructed on a 5-point Likert scale (strongly agree to strongly disagree).Psychosocial Appraisal of Domestic Work Stress: Six items were used to measure respondents’ appraisal of domestic work. Three items each measured women’s positive and negative cognitive appraisal of domestic work responsibilities on a 5-point Likert scale (strongly agree to strongly disagree).Time Demand–Control: Two item-statements (freedom to choose when and how to perform tasks) were modified from the ‘decision authority’ and one item-statement (time pressure) was modified from the ’psychological job demand’ subscales of the Job Content Questionnaire (JCQ) [61], constructed on a 5-point Likert scale (strongly agree to strongly disagree).Social Factor of DWE:Social Support: Three items measuring how frequently respondents ask for or receive assistance from household members were constructed on a 5-point Likert scale (never to always). All the variables included as measures were all potential independent variables that could be utilized in predicting health or mental wellbeing outcomes. The complete survey tool can be found in (Supplementary Materials File S1).Social Support: Three items measuring how frequently respondents ask for or receive assistance from household members were constructed on a 5-point Likert scale (never to always). All the variables included as measures were all potential independent variables that could be utilized in predicting health or mental wellbeing outcomes. The complete survey tool can be found in (Supplementary Materials File S1).The finalized survey containing the above items and demographic questions was administered by interview to study participants in the four selected communities. Data were collected in August and September 2019 from women residing in households in selected communities. The interviewers were six graduate students from the State University who were fluent in the local language and experienced at carrying out community surveys. Two interviewers were designated to each community. Interviewers approached prospective participants from every fifth house on streets in the communities, introduced the research, ascertained the woman’s eligibility by asking three questions (age, engagement in domestic work, and presence of any chronic illness), and obtained informed consent before proceeding with survey administration. Of the 365 women who agreed to participate in the study, questionnaire data were collected from 356 women, yielding a completion rate of 98%. Missing values in all variables of interest did not exceed 4%. Examination of missingness using Little’s MCAR test (Little, 1998) was not significant (χ2 p-value = 0.19). The study met the required regulatory requirements for the protection of human research participants. The study was approved by the Ethics Review Committee of the Oyo State Ministry of Health, Planning, Research, and Statistics department (approval ID:AD/13/479/223) and by the University of Iowa Institutional Review Board (IRB ID: 201904718). Before the recruitment of study participants, the community heads and village heads were approached for support. Informed consent was obtained from all participants. Women not engaged in domestic work, women over the age of 49 years (reproductive ages), and those with chronic illness and disabilities were excluded. Descriptive and Preliminary Analyses: Categorical variables were summarized using frequencies and percentages. Normally distributed continuous variables were summarized using means and standard deviations. Non-normally distributed continuous variables were summarized using medians and interquartile ranges. Normality assumptions were examined among continuous variables measuring DWE. Polychloric (inter-item) and polyserial (item–total) correlations were also estimated to assess the interrelationships among variables in the DWE domains. Data Screening and Management: The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy [63,64] and the Bartlett’s test of sphericity were used to assess the adequacy of performing factor analysis on the data [65]. The criterion for an acceptable KMO estimate was <0.70. Variables with individual measures of sampling adequacy <0.50 were excluded from factor analysis for each item. The Bartlett’s test (p < 0.05) of sphericity was used to examine if the correlation between items was greater than expected by chance. Other criteria used to determine the a priori exclusion of items from the final confirmatory factor analysis (CFA) model included (1) >20% missing responses, (2) insignificant factor loadings <0.35, (3) items with negative error variance, (4) high cross-loading, (5) low communalities (R2 < 0.40) per item, (6) weak inter-item and item–total correlation (<0.35), and (7) high influence on reduction in Cronbach’s alpha coefficient. Reliability was determined using Cronbach’s alpha <0.7 and the mean inter-item and item–total correlation <0.35 as acceptable cut-offs for each measure of DWE [66]. Confirmatory Factor Analysis: CFA uses item–factor relationships and model fit indices to evaluate whether the covariance matrix of the observed data matches the covariance matrix of a hypothesized model. Based on previous literature and the conceptual model created, CFA models were used to assess the inter-relationships among the DWE measures (observed variables) and their latent constructs. Since all items were categorical (ordinal data), a robust weighted least square estimation with mean and variance adjustment (WLSMV) was used to estimate model parameters—the residual variances among items, and covariance between factors, and the estimated factor loadings. The goodness-of-fit for CFA models was determined by the comparative fit index (CFI), the Tucker–Lewis index (TLI), the root mean square error of approximation (RMSEA), and the Kline method for assessing model fit using the χ2 statistic. As a rule of thumb, CFA and TLI values of ≥0.9 signifies very good model fit. For RMSEA, values of ≤0.06 indicate good model fit, values ≤0.08 indicate moderate model fit, and values ≥0.1 indicate poor fit [67]. Lastly, a non-significant χ2 or a ratio of χ2 to degree of freedom (df) <3 demonstrate a good model fit for the Kline method [68]. The cut-off value used to determine item loading on a related factor was 0.40 with p-value < 0.05. Modification indices were added to improve model fit while the factor structure essentially remained the same. Factors were allowed to correlate, and the variance of each factor was set to 1.0 in all models.Convergent validity was assessed by examining the strength of the average variance extracted (AVE > 0.5) of each construct, i.e., the average R2 per factor was >0.5. Construct-level discriminant validity was ascertained if the squareroot of AVE for each construct was greater than the correlations of the construct with other constructs. Raw mean scores (per factor) were generated and presented by demographic variables. Standardized regression-based factor scores were estimated for all observations by multiplying the inverse of the item’s correlation matrix value by the correlation matrix value of its factor loadings. All analysis, including CFA analysis was performed using the Lavaan and semPaths package in R [69].Pearson’s correlation was used to examine the relationship among latent factor scores of DWE (within and across domains). Multiple linear regression models (one model per latent factor) were used to examine the extent to which each regression-based factor score of DWE changed across the categories of sociodemographic variables. Statistical significance was determined at an alpha level of 0.05.More than half (62%) of the participants reported having started or completed secondary school education. Most (84%) were employed in semi-skilled, informal labor such as petty trading, tailoring, and hairdressing. The mean age was 30.8 years (SD = 6.5) and median monthly household income was NGN 15,000/USD 40. Approximately half lived in households ranging from four to six members. A tenth of the women was pregnant (11.2%), half had children under five years of age, and more than half (66%) relied on dug wells for water supply. Head loading was the most prevalent (68%) means of transporting water carriage. Women spent an average of 29 h/week and 41 h/week on domestic and paid work, respectively. After evaluating item suitability for factor analysis, 21 items were dropped during preliminary analysis. Four items relating to the ‘frequency of common domestic tasks’ (grinding, pounding food, gardening, fetching firewood) were dropped because of low variability in participants’ responses. Twelve items relating to ‘posture and movements’ during domestic tasks were dropped because of high missing responses and low variability in participants’ responses. For example, most participants (260) indicated sitting as usual posture even for tasks that typically do not require sitting, such as sweeping and washing clothes. Three items measuring ‘positive appraisal of domestic work’ were dropped because they had low Cronbach alpha, mean, and item–total correlation. Internal consistency for the remaining 31 items of DWE measure were then re-assessed using the Cronbach’s alpha. All items demonstrated good sub-scale reliability, except for ‘stress-appraisal’ and ‘demand/control constructs’ (α = 0.68 and 0.67) (see Supplemental Table S1). The Kaiser–Meyer–Olkin measure of sampling adequacy was 0.71, which is above the recommended cutoff of 0.6, and the Bartlett’s sphericity test was significant (χ2(378) = 3283.07, p < 0.001). All these value indicators suggested that all 31 items could be used in our CFA models. Standardized parameter estimates (factor loadings, robust standard errors, variances, and R2) from each item are presented in Table 1. All six measures of DWE demonstrate strong theoretical fit. The factor loadings were all significant (p < 0.001) and higher than the set cut-off (>0.4). The AVEs of each construct were all higher than the acceptable value of 0.40 and greater than the correlation coefficients among factors, which shows strong discriminant validity. The goodness-of-fit statistics of the three CFA models are presented in Table 2. First, a model consisting of seven factors and 30 items was tested. Items with negative variances (items 23 and 31), low communalities or R2 (items 1, 12, 16, 22, and 27), high cross-loadings with other items (items 1 and 31) and low factor loadings (item 1) were excluded from the model. The remaining 24 items shared at least 40% of their variance with their designated factor (R2 ≥ 0.4), indicating adequate convergent validity. Next, a second model consisting of seven factors and 24 items was tested. Finally, a third model of six factors (combined proximity to water source and water carriage into one factor, because of high inter-factor correlation) was tested in which three correlation terms were included between highly correlating items (items 7 and 8, 17 and 18, 19 and 21) from the same factor. All models met the criteria for an acceptable model as all fit indices were well within the recommended cut-offs. However, the third model had the best fit (CFI = 0.98; TLI = 0.97; RMSEA; χ2/df =1.5) and was selected as the final model. The final DWE measures consisted of: physical factors, which included 16 items across three latent factors (frequency of domestic tasks, water sourcing and carriage, and experience of water scarcity); psychosocial factors, which included six items across two latent factors (stress appraisal and demand and control); and social factors, which included three items in one latent factor (social support) (see Table 1 for final items by factor; Figure 2 for CFA Path Diagram; see Supplemental Table S2 for the DWE tool).Frequency of Domestic Work (never to always): The most frequent common domestic tasks women engaged in every day or more than once a day were cooking (93%), bathing and dressing children (90%), sweeping (87%), washing dishes (90%), fetching water (75%), carrying children (75%), and washing clothes (64%). Most women reported almost never or rarely engaging in fetching and carrying firewood (75%), pounding food (79%), grinding food (74%), or gardening/planting (76%). The mean score for all items in the factor was 3.07, meaning most women engaged in many of the listed tasks every day/sometimes (two or three times/week). Respondents ranked sweeping (63%), washing dishes (21%), and cooking (20%) as the top-three easiest domestic tasks and ranked fetching and carrying water (45%), manually washing clothes (29%), and cleaning toilets and bathroom (24%) as their top-three most physically strenuous tasks.Water Fetching and Carriage: Approximately 25% of the women had on-plot water services within their household, 21% had a water source (mostly well source) located around their compound, and the remaining half had a water source located outside their dwelling/compound (53%). About one-quarter (23%) did not have to fetch water (0 min walking distance), 25% walked around their compound to access water (≤5 min walking distance), 38% spent between 6 and 10 min, and 15% spent more than 10 min. About 22% did not have to complete a water trip, 24% of the women spent less than 10 min completing a water trip from their compound to their household, 35% spent 10–20 min, and 19% spent more than 20 min completing a water trip. Regarding the number of water trips, approximately 21% did not have to complete a water trip. Another 21% completed fewer than five trips per collection period, 46% completed between five and eight water trips per collection period, and 12% completed more than eight trips per collection period. Experience of Water Scarcity (never to always): Almost half of the respondents reported that they never experienced water scarcity consistently across the three items. The remaining half reported experiencing some form of scarcity, whether rarely, occasionally, or often. Women reported feeling worried often (10%), sometimes (20%), and rarely (25%) about having enough water in the past 30 days. Women reported rationing water often (6%), sometimes (21%), and rarely (31%) in the past 30 days. Lastly, women reported feeling angry/frustrated often (12%), sometimes (21%), and rarely (21%) about not having enough water to complete domestic tasks in the past 30 days. Stress Appraisal (Strongly Agree to Strongly Disagree): Of the four stress appraisal statements, women most often responded “strongly agree” or “agree” to: “Felt drained after completing domestic tasks” (28%); “Doing household tasks requires a lot of physical effort” (40%); and “Caring for children requires a lot of physical effort” (35%).Demand and Control (Strongly Agree to Strongly Disagree): Of the four demand and control statements, women most often responded “strongly agree” to: “Have adequate time to complete domestic tasks for the day” (51%); “Have adequate time for hobbies and other meaningful activities” (44%); and “I can choose not to do domestic work when tired or exhausted” (46%).Social Support (Never to Always and Yes/No): A total of 64% of the women reported that they ask for assistance from their family members. Women then rated their frequency of asking for and receiving assistance from family members. They mostly responded “always” (68%) or “often” (33%) to ‘asking for assistance with domestic tasks from family members’ and mostly responded ‘always’ (33%) or ‘often’ (32%) to “getting help with housework from family members”.Within-Domain Relationships: Within the physical domain of DWE, high frequency of domestic work scores was significantly correlated with increased experience of water scarcity (r = 0.31; p < 0.01), but not with water sourcing and carriage (r = 0.10; p > 0.15) (Supplemental Table S3). Increased water sourcing and carriage scores were associated with an increased experience of water scarcity (r = 0.20; p < 0.01). In the psychosocial domain, increased stress appraisal was significantly associated with increased demand and control (high demand and low control) scores.Across-Domain Relationships: High frequency of domestic work scores were significantly associated with increased stress appraisal (r = 0.36; p < 0.01) but decreased social support (r = −0.25; p < 0.01) and demand and control (r = −0.32; p < 0.01). Water sourcing and carriage had non-significant relationships with stress appraisal and demand and control, but a high score was correlated with decreased social support (r = −0.17; p < 0.01). Experience of water scarcity was positively associated with stress appraisal (r = 0.13; p < 0.01), but had no significant relationships with demand and control and social support. Increased stress appraisal (r = 0.17; p < 0.01) and increased demand and control (r = 0.14; p < 0.01) were associated with decreased social support.Women’s Age: Women in older reproductive age groups (36 years and above; 31 to 35 years) had significantly lower (p < 0.05) demand and control and higher water sourcing and carriage scores (Table 3) when compared to women in the youngest reproductive age group (18 to 25 years). Women in older reproductive age groups (31 years and above) had higher social support scores (p < 0.01). There was no significant difference in frequency of domestic work, experience of water scarcity, or stress appraisal scores by women’s age.Pregnancy Status: Women who were pregnant had significantly higher demand and control scores but lower frequency of domestic tasks scores (p < 0.05) when compared with non-pregnant women. There was no significant difference in water sourcing and carriage, experience of water scarcity, stress appraisal, or social support scores by pregnancy status.Household Income: Lower household income was significantly associated with higher (p <0.01) frequency of domestic tasks, water sourcing and carriage, and social support, but lower demand and control scores. There was no difference in experience of water scarcity or stress appraisal by income level.Level of Education: Low (primary or no formal education) education was significantly associated with higher mean and regression-based water sourcing and carriage scores. There was no significant association between level of education and other DWE measures.Household Population: Increased household size population (more than six people) was significantly associated with lower mean and regression-based frequency of domestic tasks and demand and control scores, but higher social support scores. Child’s Age and Walking Status: Having a child under five years of age was not associated with scores from any DWE measures but having a child who was not walking yet was significantly associated with increased frequency of domestic tasks (p < 0.01) and experience of water scarcity (p < 0.05).This study documented DWE conditions among rural Nigerian women and developed a measurement tool that accounted for the psychometric properties of, and relationships among, measures of DWE. Rather than focusing upon one or two DWE indicators, we developed the measures using established behavioral, work psychology, and gender theories. We tested our hypothesized framework of relationships using CFA by analyzing the internal consistency and predictive, convergent, and discriminant validity. Results indicate that the final CFA model was robust, and the latent factors derived were reliable and valid measures of DWE of women in rural Nigeria. This new measurement is important because there are no standardized or validated indicators to assess domestic work burden or experiences among women in the literature [31]. The DWE measures reflect three main domains: women’s interaction with their physical environment and the physical aspect of domestic work (physical domain); psychosocial appraisal of domestic work responsibilities (psychosocial domain); and how social support influences work burdens (social domain). When examining the relationships within each domain, there were significant inter-relationships within the physical domain. Timely management and performance of daily domestic work responsibilities, such as manual laundry, cooking, and childcare, depended on women’s access to water, which is supported by prior research [70]. Thus, water insecurity, water carriage, and domestic work are not separate entities as they had been addressed in the prior literature. Women were primarily responsible for performing multiple domestic tasks every day [71], and they ranked manual laundry—a water-dependent task—and water-fetching/carriage as the most difficult tasks. This result reinforces the economic and health benefit of investing in water-supply technologies over the past decades and the need to increase the number of households with access to improved water sources.Certain physical-domain factors also varied across sociodemographic characteristics. Frequency of engaging in domestic tasks significantly differed across levels of household income, household size, pregnancy status, and childcare-giving status. Low-income, non-pregnant women from large households and those taking care of young children (<12 months) most frequently engaged in domestic tasks. The practice of water carriage decreased as women advanced in age but increased among low-income, less-educated women. This result could be because children support their mothers, assisting with domestic responsibilities as both the women and children get older. The level of a woman’s education was directly related to her paid and household income. Education predicted the capacity to acquire wealth-based assets such as on-plot water infrastructure that can reduce domestic work demands.Within the physical domain, women’s experience of water scarcity was also related to increased water carriage and sourcing for water outside of homes. In this study, women’s experience of water scarcity was lower than expected. Increased population density in Ibadan metropolis has influenced the rate of home ownership in rural areas. In the last 20 years, urban-to-rural migration of middle-income families has increased because of ease of land and home ownership [72]. This phenomenon of urban–rural migration may have influenced the sociodemographic characteristics, DWE, and coping strategies among women in these rural areas. Many women live in abject poverty (as evident from the average household income), and few have on-plot water infrastructure. Nonetheless, few women experience water scarcity or rely on an unimproved water supply because of the communal water-sharing strategies adopted in the environment. Communal and household water sharing have been identified as adaptive buffers used to deal with water insecurity in LICs [55,73,74]. There were also inter-relationships within the psychosocial domain; women with high stress appraisal scores also had high demand and control scores. Women who perceived their domestic work to be stressful (stress appraisal) also reported that they were time-pressured, did not have time for leisure activities, and had less control over the completion of tasks (demand and control). Psychosocial stress results from the interaction between external demands, such as domestic work responsibilities, and women’s internal cognitive appraisal or perception of their capacity to cope with work demands. Demand and control significantly differed across age categories, levels of household income, and household size. In agreement with previous studies, demands of work—including perceived control and frequent engagement in domestic tasks—significantly reduced among high-income, older women and those living in larger households (more than six people) [47,49,53]. Living with limited resources and without mechanical domestic appliances or public-provided utilities to lessen domestic work burden, women often rely on support from their grown children [75]. The reverse was observed among pregnant women, whose perceived demand and stress appraisal increased despite the reduced frequency of engaging in domestic tasks. It is possible that stressors from paid work, in addition to pregnancy-induced stress, may have influenced their perceived domestic work demands [44]. For the social domain, spousal and children’s assistance with domestic work responsibilities (social support) increased as women aged, as income increased, and as household size increased. Strong social support systems have been found to positively influence a myriad of women’s health outcomes [76] in the literature. These include improved psychological resilience to violence [77], food insecurity [78], and water insecurity [74,79]; improved access to maternal health services [70]; improved psychosocial health [20]; and improved musculoskeletal health [50,51]. Thus, poor social support may reduce the opportunities for responsibility-sharing and division of labor. This study took a multi-factorial approach to systematically measure women’s DWE in rural Nigeria and provided the groundwork for further research into the inter-relationships between factors contributing to the DWE in similar populations. However, some of the items in the tool may not apply to the context of work in other populations. These measures need to be validated and contextualized among other populations. Future research in other populations could pilot and adapt this survey tool (Supplementary Materials File S1) using this same approach to identify relevant items and measures for that context, which would provide more information on the validity of these measures across populations. The tool presented in this study mainly focused on the negative experience of work and was limited in capturing the coping strategies or resilience mechanisms women utilize to cope with domestic work. This tool’s test–retest reliability analysis was not performed, which could have given more information on the stability of the relationships between the measures over time and the ability to account for seasonal variation in DWE. Some items from the final CFA model were removed because of poor internal consistency, low unique and negative variances, and low or cross-factor loadings. Question rewording and ordering, and further validation studies will be required to vigorously assess if the items contribute to the latent factors of DWE measures.Furthermore, self-report biomechanical risk factors of domestic work could not be accurately assessed, such as awkward postures, repetitive movement, bending, and squatting from respondents. A systematic review found that the ability of respondents to correctly recall and assess their posture and movement using questionnaires across studies was low [80]. Posture and movement items probably had low variation in responses because of a low level of literacy or lack of understanding of the response options. As a result of this, our DWE measure may not have fully captured the physical factors/demand of domestic work. Another limitation was that our cross-sectional design measured a snapshot of women’s DWE at one time. Since variations in exposures and DWE can be time, season, and life-stage dependent, the study design, and by extension the tool, may be limited in determining with precision the physical and psychosocial demands of domestic work. Future studies need to take a longitudinal, mixed-methods approach in assessing and understanding women’s DWE as well as combine the strengths of other methods (observation and instrument-based methods) in quantifying the physical demands of domestic work. This study demonstrated that multiple factors contribute to rural women’s DWE, as evident from the inter-relationships among measures across and between domains. This study revealed the importance of taking a multi-factorial approach when measuring rural women’s DWE, rather than relying solely on frequency/time measures. Water insecurity and carriage particularly contributed to and influenced women’s DWE because they were linked to increased work demands, lack of social support, and decision authority over task completion. Engaging in demanding domestic labor is the occupational reality of many rural Nigerian women. Using standardized and contextually appropriate tools and measures such as the DWE in understanding and quantifying the impacts of domestic labor, relevant behavioral, infrastructural, and ergonomic interventions can be developed to advocate for and reduce the physical and mental burdens from domestic labor. This tool can be used to evaluate what aspect of women’s domestic labor and experiences contribute most to adverse health outcomes (e.g., musculoskeletal or psychosocial health outcomes). The measure can also help recognize and advocate for potential areas of influence that should be targeted to achieve gender equality. For example, the measures can be incorporated in the gender-specific sustainable development goals (SDG) indicators targeting gender inequality (those assessing gender gaps in unpaid care/domestic work). Finally, the measure will be useful in developing appropriate interventions that can reduce the burden of domestic work on women. The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111043/s1. Table S1: Descriptive and reliability analysis of the DWE Measures. Table S2: The final domestic work experience survey questions. Table S3: Correlation between and across DWE regression-based factor scores. File S1: The Domestic Work Experience Questionnaire (Original Tool).Conceptualization, A.O., K.K.B. and W.T.S.; methodology, A.O., K.K.B. and W.T.S.; software, A.O.; validation, W.T.S., K.K.B., B.J. and N.B.F.; formal analysis, A.O.; investigation, W.T.S., K.K.B.; resources, A.O., J.A.I.; data curation, A.O., J.A.I.; writing—original draft preparation A.O.; writing—review and editing W.T.S., K.K.B., B.J., N.B.F. and J.A.I.; visualization, A.O.;supervision, K.K.B.; project administration, A.O., J.A.I.; funding acquisition, A.O. All authors have read and agreed to the published version of the manuscript. This research was funded by The University of Iowa Stanley Award for International Research, Advancing Graduate Student Success Award, and the Graduate and Professional Student Research Grant.The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the University of Iowa Institutional Review Board (IRB ID: 201904718) and by the Ethics Review Committee of the Oyo State Ministry of Health, Planning, Research, and Statistics Department, Nigeria (approval ID: AD/13/479/223). Informed consent was obtained from all subjects involved in the study.The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical reasons.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.The Domestic Work Experience Model.Six-factor model of DWE showing the final confirmatory factor analysis model. ‘It’are variable items included in the final model, 2–8 (Factor 1, Frequency of domestic tasks); 9–11 (Factor 2, Stress appraisal); 13–15 (Factor 3, Demand and control); 17–21 (Factor 4, Water sourcing, and carriage); 24–26 (Factor 5, Experience of water scarcity; 28–30 (Factor 6, Social support). Values to the left of the numbered items are error variances; values within the lines are factor loadings of each observed variable on the latent factors; 1.00 indicates that the variance of each factor was set to 1; r = correlation specified between Factor 5 and 6.Parameter estimates and factor loadings from confirmatory factor analysis, Model 1.Standardized factor loadings: SE = standard error; αtotal = average Cronbach’s alpha per construct; AVE = Average variance extracted after items are excluded; ** items are excluded from final model.Final confirmatory factor analysis models and their fit indices.CFI = comparative fit index, TLI = Tucker–Lewis Index, RMSEA = root mean square error of approximation.Factor scores from domestic work experience measures across socio-demographic variables.Numbers are mean (SD); * p < 0.05; ** p < 0.001. # Indicates reference categories for sociodemographic variables.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11044.txt

@@ -0,0 +1,18 @@
+Controlling soil erosion is beneficial to the conservation of soil resources and ecological restoration. Understanding the spatial distribution characteristics of soil erosion helps find the key areas for soil control projects and optimal scale for investing in a soil and water conservation project at the lowest cost. This study aims to answer the question of how the spatial distribution of soil erosion in Hubei Province changed between 2000 and 2020. Moreover, how do the effects of natural factors and human activities on soil erosion vary over the years? What are the differences in landscape pattern characteristics and the spatial cluster of soil erosion at multiple administrative scales? We simulated the spatial distribution of soil erosion in Hubei province from 2000 to 2020 by the Chinese Soil Loss Equation model at three administrative scales. We investigated the relationship between soil erosion and driving factors by Geodector. We explored the landscape pattern and hotspots of land at different levels of soil erosion by Fragstat and hotspot analysis. The results show that: (1) The average soil erosion rate decreased from 2000 to 2020. Soil erosion is severe in the mountainous areas of western Hubei province, while it is less severe in the central plains. (2) Land-cover type, precipitation, and normalized difference vegetation index are the most influencing factors of soil erosion in 2000–2010, 2015, and 2020, respectively. (3) The aggregation index values at the town scale are higher than those at the city and county scales, while the fractal dimension index values at the town scale are lower, which indicates that soil erosion projects are most efficient when the project unit is ‘town’. (4) At the town scale, if the hotspot area (6.84% of the total area) is treated as the protection target, it can reduce 50.42% of the total soil erosion of Hubei province. Hotspots of soil erosion overlap with high erosion zones, mainly in the northwestern, northeastern, and southwestern parts of Hubei province in 2000, while the hotspots in northwestern Hubei disappear in 2020. In conclusion, land managers in Hubei should optimize the land-use structure, soil and water conservation in slope land, and eco-engineering controls at the town scale.Soil erosion is one of the biggest ecological problems in the world, as it leads to water pollution, reduced land productivity and water storage capacity, and deterioration of the ecological environment, which ultimately threatens human survival [1]. According to the National Soil and Water Conservation Plan (2015–2030) approved by the State Council of China, the funds will be allocated to those areas where soil erosion is severe and clustered, where a soil conservation project is urgently needed, and where the implementation of the large-scale ecological restoration project is possible [2]. To formulate land-management policies that meet local conditions and alleviate regional soil erosion, land managers and policymakers need to understand the spatial distribution characteristics of regional soil erosion and determine where soil erosion is more serious and where land with serious soil erosion is more clustered to determine the key areas for soil restoration projects. They need to understand what factors affect local soil erosion and develop targeted treatments to reduce the cost of ecological restoration [3,4,5].Soil erosion is influenced by both natural factors and human activities, including regional topography and geomorphology, non-linearly varying rainfall erosion, soil erodibility, vegetation cover, and artificial protection measures [6,7]. Humans can influence soil erosion by changing land use and land cover [8,9]. Compared with natural influencing factors, human interference is more controllable, and understanding the main factors affecting regional soil erosion can help in developing effective ecological restoration measures [10]. Therefore, a systematic understanding of the factors influencing soil erosion over time, especially the interaction between human disturbance factors and soil erosion rates, is needed to improve soil protection and control measures [4,5,9].Empirical models have been used to analyze the spatial distribution of soil erosion [11], with the Universal Soil Loss Equation (USLE) model or the Revised Universal Soil Loss Equation (RUSLE) model being the most widely used [12]. The modeling results can show the spatial variability of soil erosion, but the landscape heterogeneity of the spatial distribution of soil erosion should be further quantified [13]. The landscape pattern analysis method developed for landscape ecology can effectively reveal the complexity of soil erosion patterns [14]. Due to the influence of landforms on precipitation distribution and the nonlinear and stochastic nature of climate change accelerating extreme precipitation, the spatial distribution pattern of soil erosion tends to change at different scales of observation, thereby affecting the mechanism of soil erosion and its spatial and temporal variability characteristics [15,16,17].Therefore, measuring multi-scale landscape pattern indices has important scientific significance for the rational allocation of land resources [16,18]. Some scholars have used landscape pattern indices to analyze the dynamic evolution characteristics of soil erosion areas at different scales, and an increasing number of scholars have begun to explore the multi-scale evaluation and scale effects of soil erosion [19,20]. However, most of the current studies on scale changes lack consideration of actual administrative divisions and only consider the changes of soil erosion under different scales of grid changes, while in reality, the management of human activities is based on administrative divisions and soil conservation projects are usually funded and implemented by the government on an administrative unit [21]. The results of studies based on different grid scales cannot be directly applied to land-use management planning, and few studies have been conducted to understand the spatial and temporal characteristics of soil erosion based on different administrative scales.Identifying critical areas of soil erosion and implementing targeted management interventions can help control erosion effectively and economically [22]. Recently, hotspot analysis has been used to identify the most important areas of soil erosion and provide optimized and cost-effective management options to reduce soil erosion [23]. As soil erosion processes vary with scale, the amount of soil erosion in hotspot areas fluctuates [3]. However, the current hotspot analysis research mainly focuses on the distribution of hotspots at a single grid scale and the indication of ecosystem service protection [23]. How does the spatial distribution of soil erosion hotspot areas at different administrative scales differ? At what scales and in which areas can soil and water conservation projects be carried out at low cost and high efficiency? Few studies have answered these questions.To address the abovementioned issues, this study assessed the multi-scale spatial and temporal variation of soil erosion losses in Hubei Province in 2000, 2005, 2010, 2015, and 2020 using the Chinese version of the RUSLE model, Chinese Soil Loss Equation (Figure 1). Compared with previous studies, this study innovatively examined the scale effects of soil erosion based on different administrative scales and identified effective control types and areas by integrating landscape patterns and hotspot analysis to suggest effective erosion control in regional landscape planning. The main objectives of this study were to: (1) Reveal the spatial characteristics of soil erosion; (2) Quantify the contributions of different influencing factors; (3) Identify the optimal control units and key areas.Hubei province is located in central China (Figure 2). Of the total area of the province, mountains account for 56%, hills for 24%, and plains and lakes for 20%. Except for the high mountain areas, most of the province has a humid subtropical monsoon climate and is the most serious area in China in terms of soil erosion, which is mainly located in northwest Hubei, southwest Hubei, northeast Hubei, and southeast Hubei. Severe soil erosion has led to the siltation of rivers, lakes, and reservoirs in Hubei, aggravating flooding, destroying land resources, reducing the productivity of arable land, deteriorating the production and living conditions of rural people, restricting economic development, increasing poverty, and making it difficult to grow land, obtain water, and increase income [24].The data sources of this paper are described in Table 1. All the data were resampled at a 1 km × 1 km resolution.We used the CSLE as the following equation:(1)A=R⋅K⋅L⋅S⋅B⋅E⋅T
+        where A is soil loss in t·ha−1·yr−1. The calculation steps of other factors in this formula are as follows. R is rainfall erosivity in MJ·mm·ha−1·yr−1. K is soil erodibility in t·h·MJ−1·mm−1. L and S are dimensionless topographic factors of the slope length and the slope steepness. B is the dimensionless vegetation cover factor of biological practices for trees, shrubs, and grasslands. E is the dimensionless factor of engineering practices. T is the dimensionless factor of tillage practices. The major difference between CSLE and USLE is that soil conservation practice factors of crop management (C-factor) and erosion-control (P-factor) used in the USLE are described by three erosion-control factors: biological (B-factor), engineering (E-factor), and tillage (T-factor) according to Chinese soil and water conservation classifications. Based on the modeling results, the national soil erosion classification standard table [2], and a similar study [30], the soil erosion zones with different erosion levels were classified into six categories (Table 2). See Appendix A, Appendix B, Appendix C, Appendix D, Appendix E and Appendix F for details of CSLE.Geodetector is a statistical tool to measure the spatial stratified heterogeneity (SSH) and to explore the determinants of the spatial heterogeneity (SH). The q value of Geode-tector was used to detect how much of the SH of the soil erosion value was explained by a given factor or by two factors [31,32]. The formula of the q value is as follows:(2)q=1−∑h=1lNhσh2Nσ2=1−∑h=1L∑i=1Nh(Yhi−Yh¯)2∑i=1N(Yi−Yh¯)2
+        where N means the number of units that composed the study area; the study area is stratified into h = 1, 2,…, L stratum; stratum h is composed of Nh units; Yi and Yhi denote the value of unit in the population and in stratum, respectively; the value range of q is [0,1]. A larger value of q indicates a stronger explanatory power of the independent variable X for attribute Y, and vice versa.The input data for Geodetector for Y (dependent variable) are as follows: the soil erosion value, the input data for Geodetector for X (influencing factor) are as follows: altitude, lithology, rainfall, slope, NDVI, land-use type. We classified the continuous datasets based on the requirement of Geodetector and previous research. The altitude, rainfall, and NDVI are divided into nine strata using the natural break method. The slope was divided into grades from first to sixth: <5, 5–10, 10–15, 15–20, 20–35, and >35, respectively.We used the aggregation index (AI) to measure aggregation levels of soil erosion spatial patterns. AI is class-specific and independent of landscape composition, and has better performance than other landscape indices when measuring clusters of spatial patterns [33]. We used fractal dimension (FRAC) to represent shape aspects of patches at different soil erosion levels. In landscape ecological research, patch shapes are frequently characterized via the fractal dimension of the object. The fractal dimension index is appealing because it reflects shape complexity across a range of spatial scales [34]. We analyze landscape patterns at the class level by Fragstats 4.2 software. A brief description of the two indicators is as follows.(3)AI=gjjmax→gjj(100)
+          where gjj is the number of like adjacencies (joins) between pixels of patch type (class) i based on the single count method. max-gjj is the maximum number of like adjacencies (joins) between pixels of patch type (class) i based on the single-count method (0 ≤ AI ≤ 100). Given any patch, AI equals 0 when the focal patch type is maximally disaggregated; AI increases as the focal patch type is increasingly aggregated and equals 100 when the patch type is maximally aggregated into a single, compact patch [33].(4)FRAC=2ln(0.25pij)Inaij
+          where pij is perimeter (m) of patch ij, aij is area (m) of patch ij, 1≤ FRAC ≤ 2. FRAC approaches 1 for shapes with very simple perimeters such as squares and approaches 2 for shapes with highly convoluted, plane-filling perimeters [34].In this paper, we used Getis–Ord Gi∗ statistic to find out where soil erosion with either high or low values clusters spatially. A significant hotspot is a highly soil-eroded area surrounded by other highly soil-eroded areas. The cold spot indicates a low soil erosion area surrounded by a low soil erosion area. The formula of Gi∗ statistic is as follows:(5)Gi∗=∑jnwijaj∑jnaj
+        (6)Z=Gi∗−EGi∗VarGi∗
+        where Gi∗ is the cluster index of cell i; Z is the significance of G∗; wij is the spatial weight; aj is the soil erosion value of cell j. E(Gi∗) is the expectation value of Gi∗ and Var(Gi∗) is the variance of Gi∗. The absolute value of z-score is positively correlated with the degree of cluster of cold and hotspots.The CSLE model estimated that the average soil erosion rate in Hubei was 3301.81 t·ha−1·yr−1 from 2000 to 2020 (Figure 3). Land with slight and light soil erosion levels accounted for 77% of the total area of Hubei Province. At the same time, land with severe and very high soil erosion levels covered 8.63% of Hubei and was mainly concentrated in mountainous areas. The total amount of soil erosion on severely eroded land was high, although the area of land with severe erosion was relatively small. The year with the highest average soil erosion was 2010, followed by 2020. The year with the lowest average soil erosion was 2005, followed by 2015. Spatially, from 2000 to 2020, the high, very high, and severely eroded areas in the northwestern, southwestern and northeastern regions of Hubei Province showed fluctuating trends, with the largest area in 2010.From 2000 to 2020, soil erosion rates showed fluctuations, with the highest soil erosion in 2010 and the lowest soil erosion in 2005; the highest value was 302% of the lowest value, and the area of slight erosion increased from 42.78% in 2000 (Figure 3a) to 64.80% in 2020 (Figure 3e), and the most severely eroded area increasing from 1.52% of the total area in 2000 to 2.05%, but severe erosion declining from 8.94% of the total area to 5.17% in 2020. They all showed fluctuating changes rather than linear increases or decreases, and the area of slight erosion was the erosion class that fluctuated the most during the study pe-riod. In addition, soil erosion rates did not show dynamic changes corresponding to dif-ferences in scale. At the city, county, and grid scales, the severe, very high, and high soil erosion areas have increased from 2000 to 2010 and decreased from 2010 to 2020. The town scale (Figure 3f–j) showed a fluctuating decreasing trend from 2000 to 2005, an increasing trend from 2005 to 2010, a decreasing trend from 2010 to 2015, and then an increasing trend from 2015 to 2020.The area of land with severe soil erosion in the plains decreased gradually, and these lands are dispersed as the scale decreased. At the city scale (Figure 3p–t), the high erosion rate areas were mainly located in the mountainous areas of western Hu-bei, of which the Shennongjia Forest Area had the highest soil erosion among the cities in Hubei Province from 2000 to 2020. In addition, the soil erosion in Shiyan city was more severe over time, mainly because the city is located in a mountainous area with steeper topographic slopes and weaker soil conservation capacity. In contrast, cities in the central and southern parts of Hubei Province have lower erosion rates than those in the northern, eastern, and western parts.From 2000 to 2020, severe erosion zones were clustered in the southwest and northwest Hubei, mainly including Shiyan city, Enshi city, and Shennongjia Forestry District (Figure 3). Overall, soil erosion in Hubei province has shown a decreasing trend, with the emergence of various traditional and modern conservation structures leading to an increase in vegetation cover and a significant reduction in soil loss. During this period, a series of national policy measures have been taken to reduce soil erosion, one of which is the return of cultivated land to forests program. The Grain for Green project has achieved significant effects in soil erosion management. In 1999, Hubei Province started a pilot project to return farmland to forest, and in 2002, it was fully rolled out. Since 1999, Hubei has returned a total of over 1.33 × 104 km2 of farmland to forest, including 4193.33 km2 of reforestation of sloping farmland and 7240 km2 of reforestation of barren mountains and wastelands. The project areas are mainly concentrated in the Three Gorges Reservoir, Danjiang Reservoir and Wuling Mountains, Qinba Mountains, Dabie Mountains, and Makufu. The project area is mainly clustered in the Three Gorges Reservoir Area, Danjiang Reservoir Area and Wuling Mountain Area, Qinba Mountain Area, Dabie Mountain Area, Moufu Mountain Area, and other ecologically important areas. Until 2019, the forest coverage rate in Hubei province increased by 7 percent, and the living wood accumulation increased by more than 30 million cubic meters [35]. The ecological benefits are reflected in various aspects, such as reduced soil erosion, increased soil fertility, and reduced wind and sand erosion.The factor analysis of soil erosion in Hubei Province using Geodetector showed that the height, slope, soil, NDVI, rainfall, and land use type had a significant influence on soil erosion (Figure 4a), and soil erosion was influenced by both natural and human activities. From 2000 to 2015, land use type and precipitation had a greater influence on the soil erosion rate than other factors, and land use type and precipitation had a greater impact on soil erosion rates than other factors from 2000 to 2015. In 2000, the NDVI had the second lowest influence on the soil erosion rate among the six factors; however, its influence gradually increased to become the most influential factor on the soil erosion rate in 2020. The degree of influence of land use type on soil erosion tended to fluctuate downwards and was surpassed by slope and the NDVI.In addition, the combined effect of the two factors on soil erosion had a greater influence than that of the single element (Figure 4b). The slope and NDVI, slope and land use, rainfall, and land use all had a greater influence on the rate of soil erosion together than when they were single elements. In the future, land management policymakers should optimize the land use structure and pay attention to soil conservation projects on sloping land and other ecological greening constructions.In terms of the interannual variation in the mean AI values of the different soil erosion zones (Figure 5a), between 2000 and 2020, the highest AI values were in the slight, light, and severe soil erosion zones. The AI values in the slight soil erosion zone and light soil erosion zone fluctuated upwards, while the AI values in the high soil erosion zone and the more se-verely eroded zones fluctuated and decreased. The AI values showed that the land with slight, light and severe erosion levels was more clustered than the land with other soil erosion levels (Figure 5a) and was concentrated in the central and eastern plains of Hubei Province; the AI values of the high and very high soil erosion zones were the lowest, indicating that the distribution of the more eroded soils became more dispersed. The lowest AI values were in the high and very high soil erosion zones, indicating that the land with higher soil erosion was scattered, mainly in the mountainous areas of western Hubei Province and the hills of eastern Hubei. This phenomenon indicates that the land with high soil erosion was relatively difficult to modify in Hubei.The change in the average FRAC values for the different levels of soil erosion areas between 2000 and 2020 (Figure 5b) showed that the FRAC index for the light erosion zones was higher than that for the high soil erosion zones in 2000. The fluctuating FRAC values in light erosion zones decreased, while the FRAC values in severe erosion zones fluctuated and increased until 2010, when the complexity of soil patches in light erosion zones started to be lower than that in high erosion zones. The higher the FRAC index was, the more complex the shapes of the patches. Therefore, the higher the FRAC value in a patch, the more difficult it was to implement soil conservation measures. The FRAC values in the severe erosion zones were still high (Figure 5b,c). In conclusion, it is still difficult to carry out soil erosion protection projects in Hubei Province.In terms of the average AI values at three different scales, both the AI values of the very high soil erosion zone and the AI values of the severe erosion zone were greater at the town scale than at the city scale. If the aggregation of land with serious soil erosion is low, it is difficult to concentrate soil and water conservation work, and there is no cluster bene-fit, which increases the cost of soil and water management. The FRAC values of the very high erosion areas varied greatly among the three scales, with the highest average FRAC values in cities and the lowest average FRAC values in towns, while the FRAC values of the severe erosion areas varied greatly among the three scales. Therefore, erosion control projects can be carried out on a town-by-town basis. In addition to selecting the scale for erosion project management, hotspot analysis should also be used to select key conservation sites for soil and water conservation projects.The soil erosion hotspots at 99% confidence in 2000, 2010, and 2020 account for about 12.43%, 11.79%, and 5.73% of the town overall, while the 99% confidence cold spot areas account for 6.07%, 12.63%, and 0% of the town overall (Figure 6), respectively. At the city scale, hotspots were clustered in the northwestern mountainous areas, over-lapping with the severely eroded soil areas. The average rate of soil erosion in the hotspot area at the town scale with a significance of 90% and above was 8179.50 t·hm−2·yr−1, which is a very high soil erosion level. Compared with the analysis of the spatial distribu-tion characteristics of land with different soil erosion levels, hotspot analysis of soil ero-sion revealed the cluster of land with the same soil erosion level. In this study, the hotspot analysis revealed that there was an area of high soil erosion in northwestern Hubei Prov-ince. It would be more cost-effective to carry out land conservation projects in hotspot are-as. In contrast, the cold spot patches clustered in the central and southeastern plains in 2000, gradually decreased from 2000 to 2015 and disappeared by 2020. The continuity of areas with relatively low soil erosion rates compared to areas with higher erosion rates is related to the topographic characteristics and land use structure of Hubei Province. Relatively large-scale soil conservation measures need to be implemented.Compared to the city scale (Figure 6a–d), two hotspots, the northeast and southwest, have appeared at the county scale. They both decreased gradually from 2000 to 2010. The cold spot at the town scale formed a patch in the central plains that decreased from 2000 to 2015 and disappeared in 2020. It is worth noting that this cold spot area did not overlap with the cold spot area at the city scale.The hotspot area at the town scale included not only the northwestern and south-western mountainous areas but was also scattered in the northeastern and southeastern mountainous areas, decreasing from 2000 to 2020 (Figure 6k–o). The cold spots were located in the central, southern, and eastern plains. From 2005 to 2020, the hotspot areas gradually decreased, and the hotspot distribution map showed a trend towards fragmentation. The distribution of high soil erosion areas became dispersed. The average soil erosion rates in the hotspot areas decreased at both the county and town scales, with decreases of 53.68% and 43.92%, respectively. During the study period, guided by the Chinese central government’s soil and water conservation policies and program, the Hubei provincial government actively participated in the Grain for Green program [36,37], a which may slow soil erosion in Hubei. As the scale changed from the county scale to the town scale, land with high soil erosion rates in southwestern Hubei Province became dis-persed hotspots in 2010.The hotspot area at the city scale was only 84.29% of the hotspot area at the town scale, but the average soil erosion rate of soil erosion hotspots in Hubei Province at the town scale was 12.17% higher than that at the county scale. The hotspots at the town scale were more clustered and severe, making soil conservation measures less costly and more effective. For town-scale hotspot areas (6.84% of the total area of Hubei Province in 2020), 50.42% of the total soil erosion was reduced. Town-scale hotspots were mostly located in the mountainous and hilly areas of Hubei Province. To control soil erosion in town-scale hotspots, land managers should combine erosion control projects with ecological restora-tion projects on sloped land to optimize the regional ecological layout [38].We found that the areas of severe soil erosion in Hubei Province were mainly in the mountainous areas of western Hubei Province, including Enshi, Shiyan, and Yichang, which is consistent with the results of Zeng’s research [39]. This indicates that the western mountainous areas of Hubei Province have been the areas with high soil erosion from 1980 to the present. By comparing the remote sensing survey data of soil erosion in Hubei Province, 52.41%, 56.17%, and 77.76% of the total eroded area in 2006, 2011, and 2019 in Hubei Province [24], it can be seen that soil erosion is on a decreasing trend, and all these trends are consistent with the results of this paper. Our study shows that the soil erosion area in Hubei province decreases, with light erosion accounting for 42.79%, 53.61%, and 77.02% of the total erosion area in 2005, 2010, and 2020(Figure 7). However, there are differences in the evaluation results because of different evaluation methods and different evaluation years.We found that soil erosion in Hubei Province is more correlated with land use, slope, and vegetation cover. During the study period, key projects for soil erosion control in Hubei Province include the National Key Construction Project of Soil and Water Conservation, the Comprehensive Management Project of Soil and Water Erosion on Sloping Arable Land, the Soil and Water Conservation Project in Danjiangkou Reservoir Area and Upstream, the Soil and Water Conservation Project for Consolidating the Results of Returning Cultivated Land to Forests, and the Soil and Water Conservation Project for Comprehensive Management of Rock Desertification in Karst Areas [24]. These ecological restoration projects reduce soil erosion in Hubei Province by increasing vegetation cover and reducing erosion loss through grass retention and tree planting, and reducing the topographic slope of the area to reduce the flow rate of water on the slope and weaken the erosion effect of flowing water on the ground. This analysis is similar to the findings of Xiao Wang et al.’s research [40].Due to the lack of erosion data in the region, the soil erosion results estimated by the CSLE model in this study were analyzed using the observations of river sand transport monitoring stations at the outlet of each watershed of typical rivers (11 river sand transport monitoring stations in 2010 and 2015) from the Hubei Soil and Water Conservation Bulletin for correlation analysis. The results showed that the overall correlation coefficient was 0.79 in 2010 and 0.82 in 2015, indicating that the soil erosion estimation results for these two years were satisfactory. This could partially reflect the accuracy of the soil erosion results. Besides, most of the existing studies on soil erosion in China are on the Loess Plateau in northern China [30], the Beijing–Tianjin–Hebei region [41], and the Yunnan–Guizhou Plateau in southern China [20,42,43], but few studies on the spatial distribution characteristics, landscape patterns, and influence mechanisms of soil erosion have been conducted in central China. Few studies are available for comparison of soil erosion in Hubei province.Because of the inconsistent resolution of precipitation data, DEM data, soil data, LULC data, and NDVI data, the accuracy of the calculation results are relatively rough. Besides, the R-factor could be calculated in various ways due to the different data resources available in different locations, based on annual precipitation data [44], monthly precipitation data, and daily precipitation data, and it is difficult to compare the results of similar research with similar topics since different calculations are used for soil erosion analysis [45,46,47]. As high-resolution data increases, future research should use a higher resolution to reduce calculation errors.In this study, the multi-scale spatial-temporal variations in soil erosion loss in 2000, 2005, 2010, 2015, and 2020 were evaluated using the CSLE model in Hubei Province. In contrast to previous studies, this study innovatively examines the effect of soil erosion characteristics based on different administrative scales and identifies effective control types and regions by integrating landscape patterns and hotspot analyses to propose effective soil erosion control suggestions in regional landscape planning. The results show that soil erosion fluctuated in Hubei over the study years. Scaling effects existed in the spatial characteristics of soil erosion. The landscape pattern and hotspots of soil erosion in each year changed with scale. The slope, NDVI and land use had a greater influence on the rate of soil erosion than other factors. The impacts of human activities increased over time and as the scale decreased. The town scale was the best control scale based on the scale effect analysiss.The sustainability of human societies depends on the wise use of natural resources. Soils contribute to basic human needs. To make regional policies and plans for soil conservation, it is necessary to identify where the soil erosion problems are, which means knowing where soil erosion rate is exceeding the soil loss tolerance. In the future, researchers could explore the trade-offs between soil erosion and other ecological services at multiple scales, which would be valuable references for effective and sustainable management and policy decisions for minimizing trade-offs and maximizing synergies of ecological services. Land managers in the Hubei government need to optimize the land-use structure, pay attention to sloping land improvement [48], increase vegetation coverage [49,50], and increase ecological restoration projects in the west of Hubei. Besides, land policymakers should consider the scale effect of soil and water conservation projects when making land-use plans.Q.L. performed the modeling and data analysis and formed the raw manuscript. Y.Z. funded and supervised this study. L.W., Q.Z. and X.S. performed the modeling and data analysis. S.Y., J.L., X.S., T.X. and Y.J. revised the manuscript. All the authors were involved in manuscript writing. All authors have read and agreed to the published version of the manuscript.This work was supported by the Special Foundation for National Science and Technology Basic Research Program of China (2021FY100505); The National Natural Science Foundation of China (42171061;41271534) and the China Scholarship Council (201906770044).Not applicable.Not applicable.Not applicable.The authors declare no conflict of interest.The equations of average annual rainfall erosivity and 24 ratios of half month erosivity to annual erosivity calculation using daily erosive rainfall data were represented as follows by referring Liu’s research [51].
+        (A1)Ryear=∑j=124Rj
+        (A2)Rj=1N∑i=1N∑k=0m(α⋅Pi,j,k1.7265)
+        (A3)RRj=Rj/Ryear
+        where Ryear is average annual rainfall erosivity in MJ·mm·ha−1·h−1·yr−1, j represents the half month sequence of 1, 2, …, 24 ineach year, Rj is half month rainfall erosivity in MJ·mm·ha−1·h−1, i is time series of daily rainfall from 1981 to 2010, k is days of daily erosive rainfall (rainfall equal or greater than 10 mm) within each half month period, Pi,j,k is 0 when no erosive rainfall occurs in the half month, and a represents calibrated parameter values of 0.3957 for warm months from May to September and 0.3101 for cool months from October to April. RRj is the ratio of the average erosivity of half month j to the average annual erosivity. The annual rainfall erosion value was interpolated using the Kriging interpolation method in Arcgis 10.7 software to obtain the R map in Hubei province.The equations of K factor (t·h·MJ−1·mm−1) in this study was referred Williams et al.,’s [52]
+        (A4)K=0.1317×{0.2+0.3×exp−0.256×SAND1−SILT100}×SILTSILT+CLAY0.3×1−0.25×TOCTOC+exp3.72−2.95×TOC×1−0.7×SN1SN1+exp22.9×SN1−5.51
+        where, SAND, SILT, CLAY, TOC is the content for sand, silt, clay, and organic content of soil(%), SN1=SAND/100.L and S are dimensionless topographic factors [53].
+        (A5)LS=L×S
+        where L is slope length, L=γ22.1α, α=β/1+β,β=sinθ/0.0893×sinθ0.8+0.56, θ is the slope value derived by DEM using Slope function ArcGIS 10.7, α is the slope exponent and β is the ratio of fine gully erosion to surface erosion [54], S is slope steepness, the calculation formula is as following.
+        (A6)S=10.8sinθ+0.03θ<5°16.8sinθ−0.055°≤θ<10°21.9sinθ−0.9610°≤θThe calculations of vegetation and biological practice factor B was referred Xie’s research. The value of B factor is based on Fraction Vegetation Coverage (FVC) (Table A1). FVC was derived based on Wang’s research [55] and Li’s research [56], and different B values were assigned to the land-use types under different vegetation cover degrees. Since bare land in the study area is very few (only 0.409%), we did not take them into the calculation.
+        (A7)FVC=NDVI−NDVImin/NDVImax−NDVImin
+        where FVC is the vegetation cover, the NDVI value with a cumulative frequency of 0.5% is the NDVImax value, and the NDVImin value with a cumulative frequency of 0.5% is the NDVImin value.Value for B factor under different FVC value of different land cover in Hubei province.P factor is determined as the ratio between the soil losses expected for a certain soil conservation practice. Without engineering practices recorded, the values were taken by refereeing Xu’s studies [57] (Table A2).Value for P factor under different land cover.The criteria of tillage practice indicator (T value) was referred Liu’s research [51] (Table A3).T factor value under different slope level.The values at different administration levels were extracted using the zonal statistics function in spatial analysis tool of ArcGIS 10.7 software.Workflow of this study.Location and altitude of Hubei province.The average soil erosion rate at the (a–e) gird scale, (f–j) town scale, (k–o) county scale, and (p–t) city scale from 2000 to 2020.The q values of influencing factors on soil erosion from 2000 to 2020 ((a) q statistic; (b) interaction q statistic).(a) AI and FRAC indexes of land at different soil erosion levels. (b) Mean value of AI and FRAC at three scales of very high erosion zones from 2000 to 2020. (c) Mean value of AI and FRAC at three scales of severe erosion zones from 2000 to 2020.The hotspots and cold spots of soil erosion at (a–e) city scale, (f–j) county scale, and (k–o) town scale from 2000 to 2020.Correlation analysis between simulated value (105 t) and observed value (105 t) in (a) 2010 and (b) 2015.Data description.The standard for soil erosion level.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11045.txt

@@ -0,0 +1 @@
+The purpose of this pilot study was to assess Chronic Myeloid Leukemia (CML) patients’ adherence to, beliefs about, and barriers to oral anticancer agents (OAC) using brief self-report measures in community-based cancer clinics. Patients completed a structured interview including a health literacy assessment, a Brief Medication Questionnaire, two single-item self-report adherence questions, and the Medications Adherence Reasons Scale. Of the 86 participants, 88.4% were white; 55.8% male; mean age, 58.7 years; and 22.1% had limited health literacy. Nonadherence (missing at least one dose in the last week) was reported by 18.6% of participants and associated (p < 0.003) with less-than-excellent perceived ability to take CML medications (16.3%). Black participants reported more difficulty taking CML medications than white participants (28.6% vs. 8.3%, p = 0.053). Among all participants, 43.0% reported their CML medicine was ineffective and 24.4% that taking CML pills was somewhat to very hard. The most common reasons for missing a dose were simply missed it (24.4%) and side effects (18.6%). Most patients perceived their ability to take CML medication was good to excellent, yet nearly one in five reported missing at least one dose in the last week. Brief, no-cost self-report assessments to screen CML patients’ OAC adherence, barriers, and beliefs could facilitate counseling in busy community cancer clinics.Oral anticancer agent (OAC agent) use is increasing and accounts for an estimated one-third of cancer drugs under development [1,2]. Oral antineoplastic therapies can improve survival, lower treatment-associated costs, and decrease patient burden by minimizing clinic visits or eliminating the need for infusions [3]. Despite the advantages, references [1,4,5,6,7,8,9,10] nonadherence to OAC agents has become a significant problem in modern oncology treatment.The World Health Organization (WHO) reported that nonadherence to oral medication is the single most modifiable factor in treatment outcomes and has a greater impact than improvement in treatments [10]. Optimal medication adherence has been defined as a patient taking their medication exactly as prescribed, at the exact time, the correct dosage, and for the recommended length of time [11,12]. However, one standard definition for medication adherence by which all measures are compared is lacking [13]. Oral tyrosine kinase inhibitors (TKIs) have transformed chronic myeloid leukemia (CML) from a fatal disease to a chronic illness [14]. The National Cancer Institute estimates there will be 9110 new cases of CML in 2021 with a five-year survival rate of 70.6% [15]. Survival benefits can only be realized if patients with CML consistently adhere to their medications [16,17,18]. A minimum threshold of 85% adherence has been described as critical to maintaining a complete cytogenic response in CML [18]. Yet, several recent studies have found a 25–35% prevalence of nonadherence to oral CML therapy [19]. Many of these studies defined adherence as use of 85–90% of the prescribed drug [20]. However, a recent systematic review identified nineteen studies reporting prevalence rates of oral chemotherapy adherence in adult CML patients ranging from 76% to 98% [21]. Reports of patients who are fully (100%) adherent only range from 20% to 53% [22,23,24]. The wide variation is due to differences in adherence definitions, measures, and the type of education patients received [21].Most CML adherence studies were performed in a single center using pharmacy claims data [25], investigator pill counts [26], patient treatment diary templates [27], electronic monitoring devices [17,28], or physician surveys [29]. Few studies have also included self-report assessments. Krikorian et al., used the Beliefs about Medicine Questionnaire, which had been used in chronic disease studies to assess patients with cancer or a higher risk for cancer [26]. It overestimated oral chemotherapy adherence compared to nurse or pharmacist pill counts. Daouphars et al. developed a 10-item adherence self-report and compared it to prescription refills to measure imatinib adherence [25].Some self-report studies used web-based surveys. Buzaglo et al. used a 51-item self-report [19] that assessed patient characteristics, financial burden, psychosocial distress, and the medication adherence of patients with CML participating in the Cancer Experience Registry. Geissler et al. [30] used the eight-item Morisky Medication Adherence Scale [31] to assess TKI adherence in patients participating in the CML Advocates Network. These studies did not assess adherence in underserved populations or health literacy, which is more prevalent in low-income and minority populations. Better understanding of OAC agent adherence among the underserved and those with low health literacy is needed [32]. Single item assessments (of missed pills over 7 days or 4 weeks) have been used to successfully assess nonadherence relevant to chronic disease outcomes but have not been used to assess TKI nonadherence [33,34,35]. One single item self-report question assessing missed doses over the past 7 days has been used in general medicine and cardiology outpatient clinics serving patients with limited income and literacy [33]. This multisite study found this single item question was particularly well suited for use in busy clinic settings and served as simple means of identifying at-risk patients for interventions to support adherence.The purpose of this pilot study (WF 99716CD) was to (1) assess the rate of OAC agent nonadherence among patients diagnosed with CML and treated in community cancer clinics using multiple self-report measures and (2) characterize patients’ barriers and beliefs regarding CML OAC medication adherence. This study was conducted to obtain baseline information to inform future research. We collaborated with multiple community-based oncology practices participating in the National Cancer Institute (NCI) Community Oncology Research Program (NCORP) to enhance the generalizability of study findings to future community oncology clinical settings. Cancer care delivery researchers in the Gulf South Minority/Underserved site of the National Community Oncology Research Program (NCORP) led this study. NCORP is a National Cancer Institute-funded national network to provide cancer clinical trials and care delivery studies in the community setting. The study was coordinated by the Wake Forest NCORP Research Base and conducted at five NCORP network community practices that saw at least 30 patients with CML in the past year and were interested in assessing adherence to OAC agents. Participating practices were well distributed geographically (Northeast, Great Plains, and Southeast) and of varied practice ownership types (e.g., academic medical centers and regional health systems with and without integrated health plans); two of the five practices were connected with Minority/Underserved NCORP Community Sites. The study was conducted from January through April 2017. NCORP practice study staff completed site-specific CITI training and were trained by the study team on administration of all instruments, data collection from EHRs, and data entry. The staff also piloted the protocol with other research staff before enrolling participants to ensure feasibility of implementation.NCORP practice study staff reviewed cancer center appointment schedules to identify eligible participants (English-speaking adults with a diagnosis of CML who had been prescribed OAC agents (imatinib, nilotinib, dasatinib, bosutinib, and/or ponatinib) for at least 30 days). Eligible patients were recruited during routine clinic visits during the study period. Structured interviews were conducted to assess basic demographic information, four no-cost brief self-report measures assessed adherence, beliefs about the medication, and potential reasons for nonadherence. A brief health literacy assessment was also administered. The Wake Forest Health Science Institutional Review Board approved the study, including informed consent.The study measures included medical record access, registration, and structured interviews. To assess potential correlates of OAC agent adherence, medical records were accessed to ascertain data related to patient characteristics (e.g., age, gender, health insurance coverage, co-morbid diseases, and number of non-CML prescriptions). Additional patient characteristics including ethnicity and race were asked in the brief interview developed by the authors. The session was conducted during a routine clinic visit and lasted approximately 15 minutes. We performed in-person interviews to overcome potential barriers associated with health literacy.The Brief Medication Questionnaire (BMQ) is a self-report tool for screening adherence and barriers to adherence of all medications taken during the previous week [36]. We modified the regimen screen to include only the question of how many times did you miss taking a dose of your CML medication in the previous week. We also kept the 2-item Belief Screen that asks, how well does the medication work for you (very well, ok, not well) and does the medicine bother you and a 1-item Recall Screen about potential difficulties remembering to take pills (very, somewhat, not at all). We characterized nonadherence as any score above 0. Single-Item Adherence Measure 1. This question asked, “Over the past seven days, how many times did you miss a dose of your CML medication?” This measure (and similarly worded measures) has been used in previous chronic disease studies [33]. We characterized nonadherence as any score above 0. Single-Item Adherence Measure 2. The Self-Rating Scale Item is a single question that uses a five-point scale to measure adherence to medication over the last 4 weeks [37,38]. It has demonstrated low patient burden and the ability to predict adherence-related clinical outcomes in HIV patients as good as or better than other adherence measures [37]. This item asks: “Thinking about the past four weeks, please rate your ability to take your CML medicine as prescribed” and it is scored on a five-point Likert scale (Excellent, Very Good, Good, Fair, and Poor). This question has been shown to significantly correlate with other measures of self-reported adherence and a medication event monitoring system (MEMS) [39] in the HIV literature. We considered those self-reporting excellent adherence as adherent and those self-reporting less than excellent as nonadherent. The Medications Adherence Reasons Scale (MARS) is an 11-item validated survey that identifies potential risk factors associated with medication adherence [34,35]. Items are based on the frequently reported reasons for nonadherence and are scored on a 5-point Likert scale. It was designed to assess medication adherence in patients with multiple chronic conditions. Health Literacy has previously been identified as a barrier to adequate prescription medication adherence but has not been studied with OAC therapy. Health literacy was assessed using the Rapid Estimate of Adult Literacy in Medicine short form (REALM-SF), a validated commonly used 7-item health word reading recognition test [40]. Raw scores range from 0 to 7 and can be converted to reading grade levels: a score of 7 indicates 9th grade or higher which can be interpreted as adequate health literacy. A score of 6 or below is reading below a high school level and can be considered limited health literacy. Results are presented as mean (standard deviation), median, or number (percentage) where appropriate. Differences in demographic and clinic variables between adherent and nonadherent participants were assessed using the Kruskal-Wallis test for continuous variables and the chi-square or Fischer’s exact test (where cell counts were low) for categorical variables. A p-value of less than 0.05 was considered statistically significant. All analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA) [41]. A total of 86 participants were enrolled. Enrollment ranged from 2 to 30 participants per site. The mean age was 58.7 years, 88.4% were non-Hispanic white, 55.8% were male, 22.1% had limited health literacy (read below a 9th-grade level), and 75.6% were privately insured (Table 1). More than 69.8% had at least one comorbid disease and all participants took at least one non-CML prescription drug; 29.1% took six or more non-CML medications. There was no difference in adherence by demographics; however, participants who reported nonadherence had higher mean BMIs compared to those that reported missing no pills (35.58 vs. 31.31, p = 0.05).Nearly one in five (17.9%) participants reported being nonadherent (missing at least one dose of CML medication in the last week) (on the BMQ—Table 2). On single-item measure one, 18.4% reported being nonadherent (missing at least one dose in the last week); on single-item adherence measure two, 16.3% reported their ability to take their CML medicine was less than excellent in the last 4 weeks. When comparing the two single item adherence questions, there was no significant difference by race for 7-day adherence. However, these results differed marginally between blacks/others who reported more difficulty taking their medicine as prescribed compared with whites (28.6% vs. 8.3%, p = 0.053) (Table 3). There was no significant difference regarding medication adherence by literacy level for the single item questions. Due to the small sample size, differences across sites were not assessed. Participants who reported that their ability to take CML medication in the last four weeks was excellent were less likely to report they had missed a dose than those who did not report excellent ability (87.5% vs. 50.0%, p = 0.003).Almost half (43.0%) of participants felt their CML medicine was ineffective and 24.4% reported taking CML pills was somewhat to very hard. The two most common reasons for missing a dose were simply missed it (24.4%) and the side effects it caused (18.6%). Compared to participants who reported missing no doses of CML medication in the last 7 days, participants who missed at least one dose were significantly more likely to report that taking their medication was somewhat to very hard (50.0% vs. 18.6%, p = 0.008). They were also more likely to indicate they had missed doses because of side effects (50.0% vs. 11.4%, p = 0.001) or because of a busy schedule (37.5% vs. 2.9% p = 0.004). We assessed self-reported OAC agent nonadherence among participants with CML across five community cancer clinics and characterized patient barriers to CML OAC adherence. To our knowledge, this is the first study to assess patient-reported medication adherence in CML across multiple community-based cancer centers. Our findings indicated almost all participants perceived their ability to take their CML medication was good to excellent, yet nearly one in five participants reported missing at least one dose in the last week. Of concern, one in four reported difficulties remembering to take their CML medication, and the main reason they reported missing doses was that they simply forgot, or they did not think their medication was working for them.Our study findings were consistent with previous studies of CML adherence, with nearly 18% missing at least one dose of their CML medication in the previous 7 days. This puts patients at risk for falling below the threshold of 85% adherence necessary for a complete cytogenetic response. The nonadherence rate in this pilot study was consistent with that found by Ibrahim et al., which used the medication event monitoring system (MEMS) to measure adherence to imatinib in 87 participants [18,39]. They found that 26% of participants taking imatinib were less than 85% adherent. Moreover, compared to participants with an adherence rate of over 85%, less adherent participants had a higher probability of losing their cytogenetic response at two years (27% vs. 2%) and a lower probability of remaining on imatinib (65% vs. 92%). The authors concluded that poor adherence is the principal factor contributing to the loss of cytogenetic response and treatment failure in patients on long-term imatinib therapy. Efficace et al. [42] found only about half (53%) of CML participants taking TKIs reported strict adherence. Participants reported both intentional and unintentional reasons for nonadherence; the most common reason for each was dealing with side effects and forgetfulness. These were the two most common reasons given in our study with “simply missed it” or “missed it because of busy schedule” being almost twice as common as experiencing side effects from the medication.There are varied reasons why patients may have suboptimal adherence to imatinib therapy. Some studies cited poorer patient understanding and knowledge of CML and its treatment, forgetfulness, concomitant drug burden, lower level of social support, depression, and financial burden as common reasons for medication nonadherence in CML patients [18,19,21,26,43,44]. In a recent qualitative study in Spain, Talens found barriers to OCA adherence included the impact of side effects on patients’ work, leisure time, and quality of life [45]. A Canadian survey found patients and providers have different perceptions of barriers to OAC agent adherence [44]. Most providers, but few patients, reported comprehension (92% vs. 1%), cost (91% vs. 25%), regimen complexity (88% vs. 4%), and interactions with other medications (76% vs. 21%) as barriers. Interestingly, almost all providers believed that patients reported adverse effects some or most of the time but 30% of patients indicated they never or rarely reported adverse effects [29,42,46]. Previous studies identified low patient health literacy as a barrier to patients’ clear understanding and adequate use of medication [47,48,49,50,51]. We anticipated this would be the case in our study, but this was not the case likely due to the small sample size. However, our finding that one in five participants had limited health literacy is not surprising. According to the only U.S. national health literacy survey to date, 14% of adults have below basic health literacy skills and another 22% have only basic skills [52]. Numerous studies have found limited health literacy is more prevalent among low-income patients cared for in community clinics [49,50,52,53,54,55,56,57,58,59,60]. This indicates the possibility that patient education and medication counseling could be enhanced by employing health literacy principles [61,62]. Those principles include use of plain language, easy-to-understand written materials focused on the benefits of taking their prescribed OAC medication and the risk of suboptimal adherence and asking patients to “teach-back” key information to confirm understanding. After testing the two single questions, we found only minor differences between the single-item questionnaires (assessing 1-week or 4-week adherence). Based on our previous medication adherence studies with HIV patients who had trouble remembering missed doses over 4 weeks [63], we believe it may be easier for patients to be asked to recall medication adherence over the past 7 days than recall the last 4 weeks and therefore more reliable. However, patient reports of behavior may be attributable to other factors such as taking medications prior to appointments to appear adherent.This multisite pilot study demonstrates that the use of brief questionnaires is an easy no-cost method of assessing CML OAC adherence, beliefs, and barriers in busy community cancer clinics. Self-report, though not as specific as using pill counts, MEMS caps, or pharmacy fill data, is practical and efficient for use in busy clinics. A measurement of adherence that is complicated, expensive, intrusive, or time-consuming is not ideal in clinical settings [33]. Community cancer clinics having a simple means of regularly identifying suboptimal adherence could help identify at-risk patients for counseling. The single item adherence question “over the past seven days how many times did you miss a dose of your medication” has been used in previous chronic disease studies. Wu and colleagues found in a multisite randomized controlled trial that this single item self-report of medication adherence question predicted hospitalization and death over a year in patients with heart failure [33].Thus, clinic staff could routinely ask patients the single item missed dose question to rapidly screen for adherence. If patient has missed a dose in the last week, staff could personalize the two belief questions from the BMQ ( how well does the CML medication work for you and how difficult is it to remember to take you CML medicine) and then to further tailor counseling give the 11-item medication adherence reasons scale to identify reasons for nonadherence. This study had some limitations, including the sample size and inclusion of English-speaking participants only. However, the study sites were geographically distributed, despite the study population being mostly white. To better understand medication adherence and barriers to medication-taking, more research is needed in community clinics that care for a greater number of low-income patients and patients from racial and ethnic minority groups who are more likely to have low health literacy [62]. Adherence was assessed by self-report and not by more stringent, but costly, methods such as the Medication Event Monitoring System (MEMS) caps [39], an objective measure of adherence that may not be suited for busy community oncology clinic settings. In addition, this pilot study was not designed to assess the validity and reliability of the assessment questions included in this study. That is an important area for future research. We also did not assess whether the type of health care provider (physician, nurse, pharmacist, medical assistant, etc.) administering the assessment questions would affect patient reporting and adherence findings All participants were taking at least one other prescription medication, but adherence to these medications was not assessed. Adherence to OAC agents prescribed for the treatment of CML is essential to maximize treatment effectiveness and clinical outcomes. We found adherence was suboptimal using a simple self-report question. The most common barriers to taking CML medication were “simply forgot” and side effects. If a patient reports missing a dose, clinic staff should administer the MARS to identify risk factors for nonadherence. Although more stringent methods of medication adherence may be ideal, they may not be feasible for busy community oncology clinics. The importance of these findings provides clinics with actionable insight to quickly identify patients at risk for nonadherence and screen for medication beliefs and barriers to personalize education and counseling. Future studies should assess adherence to all prescription medications taken by CML patients and expand the study population to include a greater number of participants, particularly minority patients. Future studies should assess adherence to all Rx meds taken by CML patients, expand the study populations to include a larger number of participants, particularly minority patients, and determine adherence relative to expected use among nonadherent patients to determine whether the level of nonadherence could potentially affect clinical outcomes (e.g., complete cytogenetic response). Conceptualization, T.C.D., C.L.A., G.M., G.J.L., R.S. and K.E.W.; methodology, T.C.D., C.L.A., G.M., G.J.L., R.S. and K.E.W.; formal analysis, W.M.B.; investigation, T.C.D., C.L.A., G.M., G.J.L., R.S. and K.E.W.; writing—original draft preparation and review and editing, all authors contributed; supervision, T.C.D., C.L.A., G.J.L., K.E.W. and P.A.P.; project administration, T.C.D., C.L.A., G.M., G.J.L., R.S. and K.E.W.; funding acquisition, T.C.D., C.L.A., G.J.L. and K.E.W. All authors have read and agreed to the published version of the manuscript.This research was funded by the National Cancer Institute, grant number 1UG1CA189824 Wake Forest NCORP Research Base; and the National Cancer Institute, grant number 2UG1CA189854 Gulf South Minority/Underserved NCORP. This work was also supported in part by the National Institute of General Medical Sciences of the National Institutes of Health [2 U54 GM104940-02], which funds the Louisiana Clinical and Translational Science Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Wake Forest University Health Sciences (IRB00039112 on 9 September 2016).Informed consent was obtained from all subjects involved in the study.The datasets generated during and/or analyzed during the current study are not publicly available due to the data set containing PHI but are available from the corresponding author on reasonable request.We appreciate the opportunity to partner with the Catholic Health Initiatives NCORP, Delaware/Christiana NCORP, Geisinger Cancer Institute NCORP, Feist-Weiller Cancer Center, Gulf South Minority Undeserved NCORP, and Sanford NCORP of the North Central Plains.The authors declare no conflict of interest.General demographics and medication adherence among patients with chronic myelogenous leukemia (CML) who were prescribed oral chemotherapy.* Significant p < 0.05.Medication Adherence Single Item Questions: past 7 days among patients with chronic myelogenous leukemia (CML) who were prescribed oral chemotherapy.* Significant p < 0.05.Medication Adherence Single Item Questions: past 4 weeks among patients with chronic myelogenous leukemia (CML) who were prescribed oral chemotherapy.* Significant p < 0.05.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11046.txt

@@ -0,0 +1 @@
+Population growth and water scarcity necessitate alternative agriculture practices, such as reusing wastewater for irrigation. Domestic wastewater has been used for irrigation for centuries in many historically low-income and arid countries and is becoming more widely used by high-income countries to augment water resources in an increasingly dry climate. Wastewater treatment processes are not fully effective in removing all contaminants, such as antimicrobial resistant bacteria (ARB) and antimicrobial resistance genes (ARGs). Literature reviews on the impact of wastewater irrigation on antimicrobial resistance (AMR) in the environment have been inconclusive and mostly focused on treated wastewater. We conducted the first systematic review to assess the impact of irrigation with both treated or untreated domestic wastewater on ARB and ARGs in soil and adjacent water bodies. We screened titles/abstracts of 3002 articles, out of which 41 were screened in full text and 26 were included in this review. Of these, thirteen investigated irrigation with untreated wastewater, and nine found a positive association with ARB/ARGs in soil. Out of thirteen studies focused on treated wastewater, six found a positive association with ARB/ARGs while six found mixed/negative associations. Our findings demonstrate that irrigation with untreated wastewater increases AMR in soil and call for precautionary action by field workers, their families, and consumers when untreated wastewater is used to irrigate crops. The effect of irrigation with treated wastewater was more variable among the studies included in our review, highlighting the need to better understand to what extent AMR is disseminated through this practice. Future research should assess factors that modify the effect of wastewater irrigation on AMR in soil, such as the degree and type of wastewater treatment, and the duration and intensity of irrigation, to inform guidelines on the reuse of wastewater for irrigation.Consequences of an ever-growing global population, such as water pollution, climate change, and unevenly distributed water resources, have led to limitations in accessing clean freshwater, driving the need for the reuse and recycling of water resources. Agriculture is the largest user of freshwater and accounts for almost 75% of water use [1]. With the world’s population estimated to reach 10 billion within the next 30 years, agricultural production is predicted to increase by 70%, putting further strain on freshwater resources [2]. Almost 50% of the world’s population uses polluted water sources for agricultural irrigation, and 20 million hectares are estimated to be irrigated with wastewaters [3]. Wastewater has been used in agriculture for centuries in many cities around the world that have a historically low accumulation of rainwater. It is also an increasingly critical alternative source of water in countries that are most impacted by water scarcity, especially those which rely on agriculture for income. For many low-income countries, reusing untreated wastewater is one of the few affordable alternatives to the advanced processes that occur in most wastewater treatment plants in high-income countries [4]; however, increasing stress on water resources has also led high-income countries to reuse domestic wastewater. For example, the U.S. reuses 4% of its treated wastewater, and some states rely on treated wastewater extensively, such as California and Florida, which use approximately half of their treated wastewater for agriculture [5]. China uses reclaimed wastewater for multiple applications, with one-third of its reclaimed wastewater going towards agricultural irrigation [6]. Irrigating crops with wastewater can also be beneficial as it supplies nutrients to the soil, reducing the need for farmers to purchase fertilizer [6]. In addition to agriculture, irrigation with treated wastewater is also used for landscaping and urban parks.Wastewater can also contain high concentrations of heavy metals, pathogens, pharmaceuticals, plastic additives, and other contaminants. Contaminants can adversely affect plant growth when wastewater is applied to crops [7]. Human exposure to wastewater contaminants can also be harmful, and agricultural reuse of wastewater has been associated with health risks. Exposure to wastewater through agricultural irrigation has been linked to enteric diseases such as salmonellosis, shigellosis, cholera, giardiasis, amoebiasis, hepatitis A infections, and viral enteritis among farmers, their families, those living close to wastewater irrigation areas, and consumers of crops irrigated with wastewater [8]. Farmers working in fields that use untreated wastewater for irrigation have also reported experiencing skin irritation, rashes, and dermatitis [8]. Adequate treatment of wastewater prior to agricultural application can alleviate some of these health concerns. However, wastewater treatment processes are not fully effective in removing all contaminants. Contaminants of particular concern include pharmaceuticals, personal care products and antibiotic residues, as well as antimicrobial resistant bacteria (ARB) and antimicrobial resistance genes (ARGs) [9]. Antibiotics are detected in treated wastewater effluent [10] and ARB/ARGs can withstand or even proliferate at treatment plants [11]. Wastewater irrigation can lead to continuous exposure of the irrigated fields to a variety of antibiotics, which can prompt the emergence of resistant strains (Figure 1). ARB in wastewater deposited onto soils by irrigation can also elicit the transfer of ARGs between wastewater bacteria and native soil communities [12]. Crops planted in soil irrigated with wastewater can take up ARB/ARGs [13] and pose a risk of spreading AMR to consumers [14,15]. There is also the potential of ARB/ARG contamination in water bodies that are adjacent to wastewater-irrigated soils [16].Antibiotics are considered persistent organic pollutants of emerging concern due to their known lasting effects on aquatic environments [17]. Effects of antibiotics on other environmental media are not as well known. Previous non-systematic reviews have provided mixed results on how wastewater irrigation affects AMR in soil. One review concluded that soils irrigated with treated wastewater do not demonstrate an increase in ARB and ARGs [11]. Two other reviews had inconclusive results [12,18]. These reviews were mostly focused on irrigation with treated wastewater and included few studies on irrigation with untreated wastewater. We conducted a systematic literature review to assess the effect of irrigation with both treated and untreated wastewater on the prevalence and abundance of ARB and ARGs in soils and adjacent water bodies. While wastewater from animal sources is often also used for irrigation, and both manure and municipal fecal sludge are used for soil amendment, we focused our review on irrigation with municipal domestic wastewater, either alone or combined with other waste streams.We searched the PubMed, Web of Science, CAB Direct, and Agricultural and Environmental Science databases and conducted a search for gray literature in Science.gov. We developed search terms to denote treated and untreated wastewater (e.g., wastewater, sewage, effluent, reclaimed wastewater), agricultural processes that use wastewater (irrigation, agriculture), outcomes of interest (antimicrobial/antibiotic resistance), and environmental reservoirs of interest (e.g., soil, field, surface water, groundwater; Table S1). More detailed information on the PubMed search string can be found in the supplementary information (Text S1). The search was conducted in November 2020. References identified in the database search were imported into Covidence software, where duplicates were removed. Titles and abstracts of the articles were screened using our inclusion and exclusion criteria. For any review articles identified during the title/abstract screening, we screened the bibliographies to identify additional relevant studies. Articles short-listed during the title/abstract screening were reviewed in full text to determine eligibility.We included studies if they detected or quantified ARB or ARGs in soils irrigated with wastewater or in water bodies adjacent to wastewater-irrigated soils. Studies were included if any form of treated or untreated domestic wastewater (alone, mixed with industrial/other waste streams, or diluted in an ambient water body after discharge) was used for irrigation. We excluded (1) studies that focused on application or amendment with sewage sludge or biosolids and with solely non-human waste (agriculture waste, dairy waste, piggery waste, manure), (2) studies that covered AMR in aquaculture and marine environments, and (3) experimental lab studies or studies conducted with irrigation water artificially spiked with antibiotics and/or bacteria that would be found in wastewater.We extracted relevant data from eligible full-text articles using Microsoft Excel. Extraction was conducted by two reviewers (SS, CN) independently to ensure accuracy, and any discrepancies were resolved by discussion. We reported data on study location, study design, type and treatment of wastewater, duration of irrigation, type of samples tested, analytical methods, prevalence and abundance of ARB and ARGs detected, and additional relevant environmental factors (e.g., pH, soil moisture, seasonality). We qualitatively synthesized data from all eligible articles to separately assess the effect of irrigation with treated and untreated municipal wastewater on the prevalence and abundance of specific ARB/ARGs in soil and adjacent water bodies. We also investigated the effect of additional environmental variables on AMR in soil and water. Our PRISMA checklist can be found in the supplementary information (Table S2).We screened the titles/abstracts of 3002 studies and reviewed the full texts of 41 studies. Based on our inclusion/exclusion criteria, a total of 26 studies were eligible to be included in the review (Figure 2). The eligible studies were conducted between 2003 to 2020. Eight studies were conducted in high-income countries (Israel, Spain, Australia, Germany, US) while eighteen studies were based in low-and middle-income countries (Nigeria, Cameroon, Burkina Faso, Egypt, India, Mexico, China). Most studies focused on agricultural fields while four studies investigated urban parks and one study focused on a recharge basin. Twenty-five studies focused on soil samples and only one study investigated subsoil pore water samples.Thirteen studies focused on irrigation with untreated municipal wastewater and another thirteen studies focused on treated municipal wastewater. The studies included irrigation both directly with wastewater and indirectly using water from an ambient water body that receives wastewater or treated wastewater effluent. Soils irrigated with freshwater, rainfed fields, and pristine soils from remote areas (e.g., national parks) were used as a comparison group. The duration of wastewater irrigation ranged from 1.5 years [18] to 100 years [19,20,21]. Studies used a mix of culture-based and molecular methods and reported the prevalence of antimicrobial-resistant isolates and both absolute abundances of ARGs and relative abundances normalized by 16S rRNA gene counts. The most commonly detected ARB included fecal bacteria and native soil bacteria, such as Escherichia coli, Enterococcus, Azotobacter chroococcum, Pseudomonas spp., and Flavobacteria. Studies assessed resistance to a wide range of antibiotics, including tetracycline, ampicillin, and ciprofloxacin. Commonly investigated ARGs included tetracycline resistance genes, sulfonamide resistance genes, quinolone resistance genes, and beta-lactamase genes. Some studies focused on detecting genetic elements involved in horizontal gene transfer, such as class 1 integrons (intl1).Out of thirteen studies focused on untreated wastewater, one studied solely domestic wastewater, seven studied wastewater that was a combination of domestic and industrial, hospital, agriculture, market or slaughterhouse waste, and five referred to municipal wastewater without specifying the content (Table 1). Three studies had no comparison group to allow assessment of associations but detected ARB in soils irrigated with untreated wastewater [22,23,24]. Of the ten studies with a comparison group, nine found that wastewater irrigation was associated with increased ARB/ARGs in soil (Table 1). In one of these nine studies, the wastewater came indirectly from a waterbody [25]. Additional details of the studies are provided in the Supplemental Information (Table S3).Four studies were conducted in Mezquital Valley in Mexico, one of the world’s largest wastewater irrigation systems, where untreated wastewater from Mexico City has been used to irrigate farmlands for 100 years [26]. All of these studies found a positive association between wastewater irrigation and ARB/ARGs in soil. Studies at this site also took advantage of the long history of wastewater irrigation to assess whether ARB/ARGs in soil increase with increasing duration of irrigation. One study found substantially more isolates resistant to at least one antibiotic in wastewater-irrigated fields (51%) than in rainfed fields (6%) and a higher prevalence (25%) of isolates resistant to ≥2 antibiotics in wastewater-irrigated fields than in rainfed fields (6%) [20]. Another study found the absolute abundance of sul1 genes to be 150–1500 times higher and sul2 genes 50–520 times higher in wastewater-irrigated soils than in rainfed soils; the relative abundance of both genes was also higher in wastewater-irrigated soils. While the absolute abundance of both genes increased with increasing years of irrigation, the relative abundance did not; soils irrigated for 100 years did not contain more sul1 and sul2 genes on the relative scale compared to soils irrigated with wastewater for 1.5 years [19]. A similar study at this site showed significant positive correlations between absolute gene abundance and years of irrigation for intl1, korB, tetW, aadA, and qacE + qacEΔ1 (quaternary ammonium compound resistance) genes while the relative abundance of these genes did not vary with duration of wastewater irrigation [21]. A fourth study from Mezquital Valley compared a field that has been irrigated with untreated wastewater for over 80 years to a rainfed field that had never been irrigated. Soil samples from the wastewater-irrigated field had an absolute abundance of 3.3 × 106 gene copies of sul1 genes per g of soil compared to 3.1 × 105 gene copies per g in samples from the rainfed field while sul2 genes were only detected in the wastewater-irrigated field [27]. In a further experiment in the same study, where soil cores from both fields were irrigated with wastewater with and without sulfamethoxazole and ciprofloxacin, the relative abundance of sul1 genes in soil from the rainfed field increased by up to 3 orders of magnitude after the irrigation experiment, while it increased by <1 order of magnitude in soil from the wastewater-irrigated field.Other investigations of irrigation with untreated wastewater included an additional study in Mexico and studies conducted in Egypt, China, Cameroon, Burkina Faso, and India. In Mexico, water from a river that receives discharges of untreated domestic wastewater from the city of Chihuahua was used to irrigate two agricultural fields. Irrigation with wastewater-impacted river water stopped on one of the fields 14 years prior to the study but continued on the other. The field continuing to receive wastewater-impacted river water showed a higher number of multidrug-resistant bacteria compared to both the field that no longer receives water from the river and a control field that was rainfed [25]. In a study in Egypt, the incidence of plasmids was 25–50% higher in isolates from wastewater-irrigated soil than from soils irrigated with canal water, and >50% of isolates carrying plasmids were resistant to ampicillin and kanamycin while >25% were resistant to tetracycline [28]. A study in China compared agricultural fields, one irrigated with untreated domestic wastewater for over twenty years and a second irrigated with fishpond water, to a field that was not used for cultivation. While the soils irrigated with fishpond water had higher tet and sul relative gene abundances than the wastewater-irrigated fields, ARGs were not detected in the field not used for cultivation [29]. In Cameroon and Burkina Faso, a study researched the impact of irrigation with raw sewage receiving input from homes, hospitals, agriculture, markets, and slaughterhouses compared to non-irrigated soils. Transferable ARGs conferring resistance to trimethoprim, aminoglycosides, beta-lactams, amphenicols, tetracyclines, sulfonamides, macrolides, quinolones, phosphonic antibiotics, and nucleoside antibiotics were 27% more abundant in wastewater-irrigated soils than in non-irrigated control soils [3]. An additional publication from the same study investigated different AMR mechanisms in both fields, including the presence of genes encoding antibiotic inactivation enzymes, antibiotic target replacement, antibiotic target protection and efflux pumps. The study found the number of ARGs encoding antibiotic inactivation enzymes to be lower in the non-irrigated fields compared to the wastewater-irrigated fields, and the number of ARGs encoding other resistance mechanisms were slightly higher in wastewater-irrigated fields [30].In four studies in India, fields were irrigated for at least a decade with wastewater that came from factories and domestic sewage. When compared to a groundwater-irrigated field, Pseudomonas spp. isolated from the wastewater-irrigated field had higher resistance towards sulphadiazine, ampicillin, and erythromycin [31]. The other three studies investigated wastewater-irrigated fields in the same area but did not report results from a comparison field. All three studies detected various ARB in wastewater-irrigated fields. These included free-living Azotobacter chroococcum isolates resistant to nitrofurantoin (92%), polymyxin-B (86%), co-trimoxazole (81%) and a total of six antibiotics (41%) [22], bacterial isolates resistant to tetracycline (75%), doxycycline (58%), ampicillin (50%), and nalidixic acid (50%) [23], and Pseudomonas spp. isolates resistant to cloxacillian (100%), methicillin (58%) and a total of four antibiotics (25%) [24].Out of thirteen studies focused on treated wastewater, the wastewater effluent was secondary-treated in three studies, a mix of secondary- and tertiary-treated in one study, tertiary-treated in three studies and biologically treated with a wetland system in one study (Table 2). The remaining five studies did not report the extent of treatment. In three studies, the wastewater effluent was diluted prior to irrigation by discharging into an ambient waterbody. Eight studies focused on agricultural fields, four on urban parks and one on a water storage basin recharged with treated wastewater. One study investigated pore water samples while the rest investigated soil. One study did not utilize a comparison field for assessing associations but found ARB in wastewater-irrigated fields. Of the twelve studies that had a comparison site, six found that wastewater irrigation was associated with higher ARB/ARGs in soil while four studies found mixed associations, and in two studies, wastewater irrigation was associated with lower or similar ARB/ARGs in soil compared to irrigation with freshwater, groundwater, and non-irrigated fields. Additional details of the studies are provided in the Supplemental Information (Table S3).The six studies that found a positive association were conducted in China, Australia, Mexico and Germany. In China, a study compared a field irrigated with treated wastewater (either directly or indirectly from a river receiving discharge), a field that was irrigated with untreated wastewater until 6–7 years ago and with rain- and groundwater since then, and a third field that is non-irrigated. The study found that the relative abundance of sulfadiazine-resistant bacteria was highest in the field previously irrigated with untreated wastewater, with no other differences in the relative abundance of ARBs between the fields. The relative abundance of tetA, tetC, tetE, tetG, tetS, sul1 and sul3 genes as well as the sum of the relative abundances of tet and sul genes were significantly higher in currently and previously wastewater-irrigated soils than in non-irrigated soils [32]. There was no difference in relative abundance of ARGs between currently and previously wastewater-irrigated soils. Two other studies in China focused on public parks irrigated with reclaimed wastewater but did not report the type and degree of treatment. In one of these studies, there was a higher diversity and abundance of ARGs encoding beta-lactam, FCA (fluoroquinolone, quinolone, florfenicol, chloramphenicol, amphenicol), and aminoglycoside resistance in soil from urban parks irrigated with reclaimed wastewater than in control soils from urban parks in the same cities not irrigated with wastewater [33]. The abundance of ARGs was 99–8655 times higher in wastewater-irrigated parks while the abundance of transposase genes was up to 2959 times higher compared to control soils. The second study also found higher diversity and absolute abundance of sul1 genes (1.69 × 108 copies per g dry soil) and intl1 genes (7.62 × 107 copies per g dry soil) in soil irrigated with reclaimed water than in pristine soils from national parks (9.08 × 107 and 2.61 × 107 copies per g dry soil) [34].A similar study in Australia compared urban parks irrigated with tertiary-treated wastewater, urban parks irrigated with potable water and remote national parks. Wastewater-irrigated parks had a higher number of different ARGs than both other sites. The abundance of ARGs in soil from wastewater-irrigated parks, conferring resistance to all major classes of antibiotics, except for erythromycin and vancomycin, was 815–4300 times higher than soil from a national park. Urban parks without wastewater irrigation, on the other hand, had 150–1240 times higher prevalence of ARGs than soil from the national park [35]. There was no difference in the relative abundance of intl1 and tnpA genes between sites. In a study in Mexico, soil from recreational parks irrigated with tertiary-treated wastewater had a higher number of multi-drug resistant bacteria in parks closer to the wastewater treatment plant compared to parks further away [36].Finally, a study in Germany compared ARGs in subsoil pore-water in fields irrigated with secondary-treated wastewater during periods of different irrigation intensity and a period with no irrigation. The relative abundance of sul, tet, qnr, bla and intl1 genes was higher during high-intensity irrigation compared to the irrigation break, and the relative abundance of several ARGs increased with increasing irrigation intensity [37]. A lab study was set up to replicate the field study and confirmed that the relative abundance of ARGs was higher in soils irrigated with treated wastewater versus freshwater [37]. Additionally, a study in Nigeria investigated soil irrigated with secondary-treated wastewater. While the study did not use a comparison site, 100% of E. coli isolates from wastewater-irrigated soils were resistant to ≥5 antibiotics [38].The six studies that found mixed or negative associations between wastewater irrigation and ARB/ARGs in soil were conducted in Spain, Israel and the US. Two studies in Spain investigated fields irrigated with wastewater from a channel that received up to 92% effluent from 10 wastewater treatment plants versus fields irrigated with rain- or groundwater. In the first study, the relative abundance of tetM, mecA, qnrS1 and blaOXA-58 genes was higher in wastewater-irrigated fields, but the relative abundance of blaCTX-M-32 was higher in the groundwater-irrigated areas [39]. The second study also investigated a third field irrigated with wastewater-impacted river water, where wastewater effluent made up <18% of the water flow. The abundance of intl1 genes was higher in soil irrigated with groundwater but the highest abundance of blaTEM was found in the soils irrigated with river water containing <18% wastewater effluent, while the abundance of qnrS1 genes was higher in both wastewater-irrigated fields [40].In Israel, a study compared fields irrigated with secondary-treated wastewater to fields irrigated with freshwater, including groundwater from an aquifer recharged with secondary-treated wastewater. The relative abundance of ARB was similar or higher in the freshwater-irrigated soils. Absolute gene copy numbers for ARGs tested (sul1, sul2, ermB, ermF, tetO, and qnrA) were similar or higher in the freshwater-irrigated soils at three out of four study sites while they were higher in wastewater-irrigated soils at the remaining site. Similarly, the relative abundance of ARGs was higher in the freshwater-irrigated soils at three sites and higher in wastewater-irrigated soils at the fourth site [41]. Notably, one of the comparison sites in this study was irrigated with groundwater from an aquifer that is recharged with secondary-treated wastewater. In a second study in Israel, commercial agriculture fields irrigated with secondary- and tertiary-treated wastewater were compared to fields irrigated with surface water, groundwater, or desalinated water. The study also examined an experimental orchard and lysimeters irrigated with tertiary-treated wastewater and freshwater. Wastewater-irrigated soil in lysimeters had higher relative and absolute abundance of intl1 genes compared to freshwater-irrigated lysimeters. However, almost all ARGs were below detection limits in all tested soils, even after irrigation with treated wastewater [42]. A third study in Israel compared soils irrigated with greywater treated by constructed wetlands to soils irrigated with freshwater, with no difference in the abundance of tetracycline-resistant bacteria between the two types of soils [43].Finally, a study in the U.S. investigated Enterococcus from sediments of a basin recharged with tertiary-treated wastewater for more than 20 years and compared it to enterococci isolated from soils and sediments in a groundwater-filled pond. A higher proportion of bacteria isolated from the groundwater-filled pond was resistant to 4–6 antibiotics (25%) than bacteria from the wastewater-recharged pond (9%), and a smaller proportion of bacteria from the groundwater-filled pond was susceptible to all antibiotics tested (7%) than bacteria from the wastewater-recharged pond (36%) [44].Other environmental factors besides wastewater irrigation had impacts on the abundance and diversity of AMR in soil. Multiple studies noted the impact of soil moisture, precipitation, temperature, pH and soil depth. A study in Israel found soil moisture had a significant positive correlation with bacterial resistance to tetracycline and ciprofloxacin [41]. A study in Germany found that the relative abundance of sul1 and plasmid-borne qnrS genes in subsoil pore water increased with increasing temperature, and the relative abundance of sul1 genes was positively correlated with precipitation, but there was no correlation between ARGs and humidity [37]. Similarly, in Mexico, the prevalence of ARB in wastewater-irrigated soils was lower during the dry sampling period compared to the rainy period [36]. Evidence on the effect of soil pH was mixed. A study in Australia found ARG abundance in soil to increase with soil pH [35] while conversely, in another study, higher soil pH was negatively correlated with the abundance of ARGs and the intl1 gene [34]. In a study in Mexico, there was no association between soil pH and ARG abundance [21]. Most studies investigated top soils (0–30 cm depth) and some studies assessed the effect of soil depth on ARB/ARGs. In Mexico, the prevalence of multi-resistant bacteria was not affected by soil depth, comparing samples collected at 0–15, 15–30 and 30–50 cm depth [25]. Similarly, In China, the relative abundance of ARGs was not significantly different between soil depths of 0–10 cm and 10–20 cm [29]. Aggregation of agricultural soil may also play a role in the dissemination of AMR in wastewater-irrigated fields. A study in China found no difference in ARG abundance between rhizosphere, non-rhizosphere and wetland samples [34]. In a study in Mexico, untreated domestic wastewater used to irrigate soil cores was dyed before irrigation to visualize water flow paths. The dye stained a greater volume and deeper in the soil cores collected from wastewater-irrigated fields (80%) than those in the more compacted rainfed fields where the dyed water followed the root system rather than penetrating a larger area of the soil core (50%). The abundance of sul1 and sul2 genes was higher in stained soil compartments along the flow path than in unstained compartments, suggesting that water flow paths could be an area of concern with high levels of resistance genes [27].This review summarizes results from 26 studies on the impact of wastewater irrigation on the prevalence and abundance of ARB and ARGs in soil and water. Our review indicates that an important determinant of the presence of AMR in wastewater-irrigated soil is whether the wastewater used for irrigation was treated. We found evidence of a positive relationship between irrigation with untreated wastewater and both the presence and abundance of ARB/ARGs in soil, where nine out of ten studies that had a comparison group (e.g., fields irrigated with freshwater) showed an increase in ARB and ARGs in wastewater-irrigated soils. In contrast, studies that investigated irrigation with treated wastewater had heterogeneous findings. Out of the twelve studies in this category that had a comparison group, wastewater irrigation was associated with more abundant ARB/ARGs in soil in six studies, while the remaining six studies found mixed or negative associations.Our review also revealed that studies examining ARB and ARGs in water bodies due to wastewater irrigation are currently limited. Only one study in our review studied sub-pore water, and we identified no studies investigating AMR in underlying groundwater aquifers or surface water bodies adjacent to wastewater-irrigated fields. Wastewater irrigation has been associated with the detection of pathogens, nitrates, and antibiotics in surface- and groundwaters [45]. Future research should investigate whether ARB/ARGs are detected in waters impacted by wastewater irrigation.Our findings highlight the need to further investigate the drivers of heterogeneity to identify settings and factors that modify the risk associated with wastewater irrigation. Notably, the studies focused on untreated wastewater exclusively came from middle and low-income countries while eight out of thirteen studies on treated wastewater came from high-income countries. AMR carriage is significantly higher in low-income countries, which has been attributed to unregulated antibiotic use and poor sanitary conditions [46,47]. Therefore, wastewater used for irrigation in low-income countries is more likely to contain ARB/ARGs. The extent and effectiveness of wastewater treatment also differs between high- and low-income countries. The six studies that investigated irrigation with treated wastewater and found mixed or negative effects on ARB/ARGs in soil were conducted in high-income countries with presumably effective and well-operating wastewater treatment systems, while the majority of the studies that found an increase in ARB/ARGs in soil from irrigation with treated wastewater came from low-income countries. Therefore, differences in ARB/ARG loads in wastewater and removal efficiency for ARB/ARGs in wastewater treatment plants between high- and low-income countries could explain why studies on irrigation with untreated wastewater found an increase in AMR in soil while studies on irrigation with treated wastewater had heterogeneous findings.Differences in the types of wastewater treatment steps employed would also be expected to affect the presence of antibiotics, ARB, and ARGs in the treated effluent and consequently the impact on soils. However, studies in our review that focused on secondary vs. tertiary-treated wastewater had similarly mixed findings. Among the three studies that investigated tertiary-treated wastewater, two found a positive association between wastewater irrigation and ARB [36] and ARGs [35] while the third found a negative association with ARB [44]. Whether or not the wastewater or treated effluent was diluted via discharge into an ambient waterbody prior to irrigation also did not appear to influence the effect of wastewater irrigation on ARB/ARGs in soil.It is possible that ARB from irrigation with wastewater could take several years to accumulate in soil [48]. The duration of wastewater irrigation prior to sampling varied immensely (1.5 to 100 years) across the studies in our review, and nine out of 26 studies did not report the duration. Based on studies in Mezquital Valley, duration of irrigation has implications for the dissemination of ARB/ARGs within wastewater-irrigated soils [19,21]. Therefore, mixed findings between studies could be due to differences in the duration of wastewater irrigation. In addition, when ARB and ARGs are detected in wastewater-irrigated soils, it is unknown whether and how long they persist, either through the survival of the host bacteria or as free naked DNA [18]. Few studies in our review reported the time elapsed between the last episode of wastewater irrigation and collection of samples, and in most studies, soils appeared to be sampled concurrently with ongoing wastewater irrigation. Three studies investigated fields where wastewater irrigation was discontinued and had mixed findings. In Spain, the field currently irrigated with untreated wastewater had the highest abundance of multi-resistant bacteria while the field previously irrigated with untreated wastewater and the rainfed control field had similar abundance of multi-resistant bacteria [25]. In a study in Germany, the relative abundance of ARGs was higher during periods with active irrigation compared to after a 4-month irrigation break [37]. In contrast, in China, there was no difference in the relative abundance of ARB and ARGs between fields currently vs. previously irrigated with wastewater [32]. Other facets of wastewater irrigation, such as the origin of wastewater, and the intensity, frequency and volume of irrigation can also modify the effect of wastewater irrigation on AMR in soil; these factors were only partially reported by studies in our review.Mechanisms for AMR naturally exist in native soil communities [49]. When determining the impact of wastewater irrigation on ARB and ARGs, it can be difficult to assess the respective contribution of wastewater due to the natural bacteria and resistance already present in the soil [41]. DNA can exist in soil for long periods of time so when researchers use molecular methods of detection, such as polymerase chain reaction (PCR), they might detect pre-existing native bacteria that have been in the soil for many years [12]. Many studies in our review included soil samples that were not wastewater-irrigated, allowing a comparison to isolate the impact of wastewater. However, comparison soils can also be contaminated with AMR elements if they are close to wastewater-irrigated sites (via aerosols and dust) or if they have received soil amendment with manure or biosolids [19]. Only a few studies in our review used “pristine” comparison soils from remote areas with less anthropogenic activity.Selection of which ARB/ARGs were investigated can also lead to heterogeneous findings across studies. It is also important to note that due to limits of detection and quantification, PCR can fail to detect or quantify ARGs that are present in low levels, which may still have a biological impact [18]. Levels of ARGs can be expressed as a gene ratio, comparing the gene copy numbers of the ARG to those of a common gene such as 16S rRNA. These ratios are used to define the relative abundance of the ARG and can be too low to be interpreted or compared between samples [18].Finally, it is important to include details of soil properties and environmental characteristics in future studies. Multiple studies in our review indicated associations between the abundance of ARGs and soil pH, soil moisture, precipitation and temperature. Additionally, the fate and transport of antibiotics in soil and the DNA extraction efficiency from soil samples can vary with varying soil properties [41]. Reporting soil and environmental characteristics in future studies could help identify factors that may modify the effect of wastewater irrigation on the presence and abundance of ARB/ARGs in soil.Studies have attempted to estimate the human health risk from exposure to antibiotics through wastewater irrigation [15,50], but the health effects of potential exposure to ARB and/or ARGs due to wastewater irrigation are unclear [18,51]. Individuals can be exposed to these through contact with soil or consumption of crops that have taken up ARB/ARGs from wastewater-irrigated soil, potentially leading to gut colonization with resistant bacteria. However, environment-to-human transmission of AMR remains poorly understood [52]. Studies using advanced molecular techniques such as whole genome sequencing have shown genetic overlap between ARB isolated from humans, animals and the farm environment, suggesting transmission between these reservoirs and hosts through farming practices such as soil amendment with manure [53,54]. Similar risks could exist for farmers as well as consumers due to environment-to-human transmission and spread of AMR when untreated wastewater is used for irrigation [55]. Low-income countries, where most wastewater remains untreated and is also more likely to contain ARB/ARGs due to high community carriage rates, are a particularly high-risk setting for further emergence and spread of AMR via wastewater irrigation. Novel resistance mechanisms that emerge in such hotspots have been shown to quickly spread globally [56,57]. Irrigation with untreated wastewater could therefore pose risks beyond the countries where it is practiced.Given scarce water resources, climate change and population growth, wastewater irrigation is increasingly common, in both low- and high-income countries. Through a systematic review and synthesis of the available literature, we demonstrate the diverse impact that domestic wastewater irrigation can have on the presence of AMR in soil. Our findings indicate a clear relationship between untreated wastewater irrigation and increasing prevalence and abundance of ARB and ARGs in soil. While there are no studies on the magnitude of human health risks associated with exposure to AMR via irrigation with untreated wastewater, our findings warrant precautionary action by field workers, their families, and consumers, particularly in low-income countries where use of raw sewage for irrigation is common. Studies should also investigate whether irrigation with untreated wastewater leads to contamination of adjacent water sources with ARB and ARGs. In our review, the evidence on whether irrigation with treated wastewater increases the prevalence and abundance of AMR in soil was mixed. Future research should explore factors that can explain the heterogeneity in the effect of irrigation with treated wastewater on ARB and ARGs in soil, such as the extent of wastewater treatment and the intensity of irrigation, to inform guidelines on wastewater reuse for irrigation.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111046/s1, Table S1: Search terms, Text S1: PubMed search string, Table S2: PRISMA checklist, Table S3: Characteristics of studies included in review.Conceptualization, S.S. and A.E.; methodology, S.S. and A.E.; investigation, S.S. and C.N.; data curation, S.S.; writing—original draft preparation, S.S. and A.E.; writing—review and editing, S.S, C.N., G.A., S.T. and A.E.; visualization, S.S.; supervision, A.E.; project administration, S.S. and A.E. All authors have read and agreed to the published version of the manuscript.This research received no external funding.Not applicable.Not applicable.The authors declare no conflict of interest.Role of wastewater irrigation in the emergence and spread of antimicrobial resistant bacteria (ARB) and antimicrobial resistance genes (ARGs) in the environment.Flowchart of literature review and screening.Characteristics of studies on irrigation with untreated wastewater.WWI: Wastewater irrigation; AMR: Antimicrobial resistance; ARG: Antimicrobial resistance gene. a No specific target organism, DNA extracted directly from soil.Characteristics of studies on irrigation with treated wastewater.WWI: Wastewater irrigation; AMR: Antimicrobial resistance; ARG: Antimicrobial resistance gene. a No specific target organism, DNA extracted directly from soil.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11047.txt

@@ -0,0 +1 @@
+The aim of this study was to analyse the inclusion of a gender perspective (GP) in scientific production on interventions for a reduction in psychological distress in children who have experienced parental gender-based violence (CEXPGBV). To achieve this, a review of publications was carried out in the Web of Science, EBSCOhost, ProQuest and Cochrane Library databases. A total of 3418 records were found, and 44 items of research selected. For GP analysis, the questionnaire “Gender perspective in health research” (GPIHR) was applied and relationships with the terminology of violence were analysed, as well as the definition of term used, references to violence by men or received by women and the instruments used to assess these. Generally, the assessed studies do not contain a GP, since 70% of the GPIHR items were answered negatively. Likewise, 89% of research used general terms to refer to violence without referring to gender. These results show the importance of considering instruments such as GPIHR in both the planning and development of future research in order to avoid possible gender bias.Gender perspective (GP) is a theoretical approach with the main aim of analysing gender inequalities, with some of its central categories being socialisation and gender roles/practices, and power relations or domination systems—subordination between sexes [1]. As Rohlfs et al. [2] have stated, the application of GP in health research has become a requirement of good practice in different fields and professions. Both national and international organizations stress the need to include GP in health research to advance scientific quality, avoid bias, reduce inequality and advance equity in people’s health [3,4,5]. Likewise, as Vázquez-Recio [6] concludes after analysis of scientific production, the inclusion of GP has become an ethical position in research. In recent decades, despite an increase in promoting the advantages of including GP, its application remains scarce [7]. One mechanism for increasing methodological rigor and advancing science is criteria ensuring inclusion of GP in research. For this purpose, we found the validated questionnaire “Gender perspective in health research” (GPIHR) by Tomás et al. [8].In the area of gender-based violence, including this perspective in the research context has become inevitable, as its own legal definition reflects the “manifestation of discrimination, the situation of inequality and the relations of power of men on women” [9]. This is reflected in the latest data from the macro-survey on violence against women 2019 [10], where one in every two women (57.3%) residing in Spain has suffered violence during their lives for the simple fact of being a woman. Likewise, 89.6% had children when the violent episodes took place, and 51.7% reported that these had witnessed or heard the violence against the mother and had suffered violence at the hands of the partner. According to estimates of this macro-survey, 1,678,959 children live in homes where the woman is currently suffering some form of violence from their partner (physical, sexual, control-based, emotional, financial or fear-based).Regardless of the high number of children of women in gender-based violence contexts, there is no specific term to define this situation. In the scientific literature, we find this population defined as “exposed to” “intimate partner violence” or “domestic violence”. Nevertheless, childhood and adolescence development in contexts where this type of violence occurs, besides affecting children’s well-being, also affects their construction of gender dynamics, and helps to perpetuate the relationship dynamics of gender inequality. Therefore, we propose identifying this population as “Children with experiences of parental gender-based violence” (CEXPGBV), since we believe this identifies and expresses the relational complexity that this experience entails for these children. Likewise, this conceptualization considers relational aspects on the part of the children related to attachment theory, development theory, social learning theory, emotional security theory, and the Eco-Biological Theory of Development [11,12]. The parental/referential distinction is made since, apart from the fact the person attacked is always the mother, the aggressor may have different profiles. Most commonly it is the father; however, we also find aggressors toward the mother who do not fulfil a parental role, but do, however, act as a referential (mother’s partner).A key factor when analysing scientific production on gender-based violence and children is language. English is the predominant vehicular language due to quality standards for scientific publication [13]. Despite the adoption of the term “Gender-Based Violence” in the Declaration on the Elimination of Violence against Women of the United Nations in 1994 [14], and the current support of the European Council and European Commission to refer to this problem, in the research context, the terminology is diverse, nonspecific and varies over time. In analysis by López-Cepero et al. [15], it was found that, among different terms available to refer to gender-based violence in intimate partner relationships, “domestic violence” is most used in this research area. Likewise, Reed et al. [16], after analysing the terminology in scientific production, concluded with the need to recognise intimate partner violence as a gender problem in research, in order to progress toward reducing this highly prevalent form of abuse.Including GP in research is one strategy which can enable this problem to become more visible. According to the compilation by Ferrer-Pérez and Bosch-Fiol [17], the recognition of these is:“(a) between men and women there have existed and still exist historical inequalities and discrimination that cause gender gaps, (b) certain power relations are established, which are generally favourable to men as a social group and discriminatory towards women and (c) these relations have been socially and historically constructed, they condition life and the roles played by women and men, cross the entire social fabric and are articulated with others (such as those derived from social class, ethnicity, age, sexual preference or religion).” (p. 72)Likewise, another study [18] suggests introducing GP in psychological research in situations of gender-based violence through critical analysis of the following points: (a) theoretical models based on gender-based violence, its causes and mechanisms; (b) sample selection method; (c) instruments selected for assessment; and (d) carrying out meta-analytical reviews to learn the real scope of differences between abused and non-abused women and between abusers and non-abusers.In spite of the difference between “sex” and “gender”, both concepts tend to be used interchangeably in research. In the review by Ritz et al. [19] this conceptual distinction is clearly stated:“Sex can be thought of as a biological attribute (such as those characteristics relating to genetics, physiology, anatomy, or reproduction) used to classify sexually reproducing animals (typically as males or females), while gender refers to the social processes that collectively influence the social roles, relationships, behaviors, power, or other traits that are culturally accorded to those classified as women/girls and men/boys.” (p. 4)At a behavioural level, gender is expressed through practices or roles where the individual demonstrates a series of socially expected characteristics, whether for men or women [20]. In this regard, certain behaviour associated with the feminine, such as caring being expected from women, and self-sufficiency from men. However, there can also be a “crossed” function (a man with feminine characteristics or a woman with masculine) and “androgynous” functions (both masculine and feminine characteristics) [21,22]. It must be noted that these practices are not static and vary with time and culture, though power dynamics are perpetuated.Therefore, using the term “gender” as a synonym for “sex” can cause bias in results interpretation, as the sample’s socialization factors are not being considered. In CEXPGBV, this becomes a crucial element in research, given that they are immersed in a family environment with maximum expression of male dominance over women. Along these lines, different studies show that in gender-based violence dynamics, the aggressors usually fulfil the male gender role [23,24,25] and use parenting models which perpetuate gender stereotypes [26,27].It should be mentioned that in the present work, binary terms are used (man–woman; masculine–feminine), though this dichotomization is artificial as these are social constructs, and do not reflect diversity of sexes and possible genders derived from interactions between biological and social factors [19].There are several systematic reviews and meta-analyses highlighting the psychological distress caused to children who have witnessed gender-based violence towards their mother in the primary family nucleus [28,29,30]. Data from these studies are consistent with the latest report by the Government Delegation for Gender Violence (2015) “The invisible victims of gender violence” in the Spanish population [31], where we found that these children present statistically significant difficulties related to internalizing responses (concordant with anxiety processes, depression, somatic complaints), externalizing (breach of norms/limits, aggressive behaviour) and post-traumatic stress disorder (PTSD), with several comorbidities among these, defined as Complex Trauma or Developmental Trauma [32,33].As for differences in the impact on this population according to the variable “sex”, at the start of the 21st century, externalizing symptoms were associated with boys, and internalizing with girls [34]. However, based on subsequent reviews, the data are inconclusive. On the one hand, we found meta-analyses where results express a greater association of externalizing symptoms in boys, but not in terms of internalizing symptoms in girls [28,35]. On the other hand, analyses by Kitzmann et al. [36] concluded that sex did not play a moderating role in internalizing or externalizing symptoms, and similarly Vu et al. [37] did not find that sex was associated with exposure to gender violence and adaptation problems in childhood. Finally, in the review by Fong et al. [38], they conclude with a trend linking externalizing/aggressive behaviour and the sex of the children with the sex of the aggressive parent. It must be noted that in all studies mentioned, sex and gender were used as synonyms despite being conceptually different study variables.There is less scientific production in research that takes gender into account. In fact, we found no meta-analyses or systematic reviews where differences in the involvement of CEXPGBV with GP are analysed. We did find studies by Smagur et al. [39] where the effect of gender roles was assessed as a predictor of internalizing and externalizing behaviour problems in this population at 4 years of age. In the starting hypotheses, behavioural problems were expected according to gender roles; however, it was found that girls with gender roles typified as “feminine” showed a risk of externalizing, but not internalizing, behaviour problems. These results are explained due to the detrimental effects of assuming femininity as inferior. Likewise, in their analysis they presented possible bias in these results as mothers and fathers have difficulty in observing internalisation problems in this age group [40,41,42].Since the 1980s, several studies have been published where different types of child–adolescent psychological interventions are presented and used to reduce psychological distress in children who have witnessed violence towards their mothers [34]. In the last two decades, several systematic reviews and meta-analytic analyses on interventions for CEXPGBV have been published. Analyses of these are mainly directed toward methodology used in interventions (individual, family, group and joint). As for gender analysis, in the meta-analysis by Romano et al. [43] where five meta-analyses and seven systematic reviews were analysed, they state a lack of data (or inadequate data) which limits examination of the weight of basic demographic variables such as sex or variables that might be relevant (e.g., attitudes toward family violence) as key moderators in identifying the long-term effects of intervention.We found a recent meta-analysis by Latzman et al. [44], which included randomised controlled trials (RCTs) considered of low or moderate risk of bias, finally totalling eight studies for analysis (n = 924). In this study, no mention was made of sex or gender in analyses regarding children, nor its possible moderating effect on results. It only referred to the children’s relationship with the aggressor and parent who had been attacked, stating that, in 7 out of 8 selected studies, the assaulted person was the mother; in the remaining study the assaulted parents were fathers and mothers. It is striking that in analysis of this situation, the only study whose sample of assaulted parents is mixed is what the author terms the most “inclusive”, adding no further reference to the matter. However, no reference is made to the percentage of assaulted mothers and fathers in the sample of this article, nor consideration that in 87.5% of selected studies the attacked parent is the mother (therefore, at the population level, this study rather than being “inclusive” is actually damaging to the sample), nor is there analysis explaining the reasons for difference between abused fathers and mothers, or whether there exist gender factors which explain this.The aim of this study was to analyse the inclusion of a gender perspective in scientific production available until 2020 on interventions to reduce psychological distress in children who have suffered parental/referential gender-based violence.The aims are: (1) to identify published experimental, quasi-experimental, non-experimental and single-case studies of intervention in the reduction of psychological distress in CEXPGBV; (2) quantify and describe variables related to gender and violence in research (terminology, definition, assessment); (3) assess inclusion of a gender perspective in published studies using the GPIHR instrument; and (4) identify possible relationships between variables described and GPIHR dimensions.A systematic review was performed to achieve these aims; the writing of this research was carried out following recommendations from the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement [45] and considerations for professional practice by Sánchez-Meca and Botella [46].Given the socio-sanitary nature of the present study, the Web of Science, EBSCO host, ProQuest and Cochrane library meta-databases were used to search for research. A total of 46 databases were selected with a health and/or social nature, as well as general research. The Mendeley bibliographic manager was employed to manage the search information. The selection of keywords for each search element (see Appendix A Table A1) was through a conceptual bibliographic review in terms of terminology used in research on gender-based violence, prevalence of psychopathological involvement in CEXPGBV and interventions aimed at this population.The search was restricted to studies published up to 2020, compiling all previous published material. Inclusion criteria were: (1) experimental studies (RCTs), quasi-experimental studies, non-experimental and single-case studies; (2) the attacked parent being a woman; (3) intervention aims to counteract the effects of violence on the mental health of the children, regardless of modality; (4) the study had to be written in English. Research was excluded where: (1) the intervention was directed at mothers/fathers without assessing the impact on children; (2) children were in child protection centres; (3) if the aim of intervention was different to counteracting the consequences of experiencing parental/referential gender-based violence.For each study, the term used to describe violence was compiled and categorised as “general” if not referring to gender (or associated factors) or as “specific” if doing so. Likewise, it was analysed whether the term of violence used was described, whether it refers to such violence by men or received by women and whether any instrument is used to assess it, either in frequency, intensity or duration. In addition, year of publication of study and decade were gathered to analyse temporal evolution.GP analysis in studies was performed using the questionnaire “Gender perspective in health research” (GPIHR) by Tomás et al. [8], where items were adapted to childhood research (Appendix A Table A2). This comprises 10 items: 3 assess introduction, 1 aims, 3 methodology and 3 the research aim. The GPIHR has three factors revealing different levels of GP inclusion in research projects:-Factor 1: Gender sensitivity. Referring to differences in health between men and women and the relationship between gender factors and the health issue addressed in the research project.-Factor 2: Feminist research. Gathering all necessary conditions for research to have a gender perspective and a feminist purpose, i.e., investigating the causes of inequality in order to try and change them.-Factor 3: Sex difference. Reflecting disaggregation of data by sex and age group, enabling identification of differences in health.Factor 1: Gender sensitivity. Referring to differences in health between men and women and the relationship between gender factors and the health issue addressed in the research project.Factor 2: Feminist research. Gathering all necessary conditions for research to have a gender perspective and a feminist purpose, i.e., investigating the causes of inequality in order to try and change them.Factor 3: Sex difference. Reflecting disaggregation of data by sex and age group, enabling identification of differences in health.Completion of the GPIHR questionnaire was performed in a dichotomous “yes/no” manner for each question. It was considered “Yes” when the question could be answered in the affirmative with the information provided in the assessed research. It was considered as “No” when the evaluated research did not provide the necessary information to answer the question, when the information provided was ambiguous or directly led to a negative answer.All GPIHR questions are focused on the inclusion of GP-related factors in the research. Therefore, the greater the number of “Yes” responses in the questionnaire, the greater gender perspective the research is considered to contain.Results analysis was carried out through the statistical program IBM SPSS by descriptive analysis, the Kolmogorov–Smirnov test was used to verify the assumption of normality of data in the GPIHR, the Kruskal–Wallis test for analysis of categorical variables with more than two categories, the Mann–Whitney U for categorical variables with two categories, and Spearman’s rank-order correlation coefficient to quantify the relationship between ordinal variables. When a statistically significant result was obtained with the Kruskal–Wallis test, two-to-two posterior comparisons were made with the Mann–Whitney U test. In this case, in order to control inflation of Type I error rate, the Bonferroni correction was applied on probability values to achieve a statistically significant result at a nominal significance level of 5%, and p value associated with the U test result had to be equal to or less than 0.017.A total of 3418 records were obtained, 89% identified through database searches, 11% other sources (Figure 1). Forty-four records were finally included [47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90]. It should be noted, of 48 records excluded in the last screening phase, 39.6% related to the ambiguity of describing as “parent” the main caregiver of the children who had been assaulted (n = 10), since the groups of people attacked were a mixture of fathers and mothers (n = 9). In all studies where groups of the abused parent were mixed, the vast majority were women, with percentages ranging between 81% and 98% [91,92,93,94,95,96,97,98]. Only one study [99] does not provide specific data. However, it reports that in 65% of cases the physical aggressor was the father, 2.5% the mother and 32.5% both parents. Furthermore, it reports that in 87% of cases the mother reported the child’s state. Likewise, of all selected studies, in only one was the intervention aimed at the father and assessed consequences in the children.The systematic review showed that, from 44 studies selected, 42% were RCTs (n = 19), 29% quasi-experimental (n = 13), 27% non-experimental (n = 11) and 2% single cases (n = 1). Table 1 shows frequency analysis. As for the terminology of violence used in the research, general terms were most used (89%). Intimate Partner Violence and Domestic Violence were those most used, in 39% and 32% of studies, respectively, the remainder (19%) being general terms with few repetitions (marital violence, domestic abuse, family violence, violent homes and traumatic violence). A total of 11% of studies referenced some factors related to gender (woman abuse, battered woman and wife abuse), of which two used the term gender- based violence.In the Appendix A (Table A3), we found compliance of GPIHR items for each selected study. For frequencies of compliance with the GPIHR questionnaire in selected studies (Table 2), items were generally answered favourably 30% of the time. By factor, 18% of items related to “Gender sensitivity”, 8% to “Feminist research” and 68% to “Sex difference” were answered favourably. For analysis of the normal distribution of data in the GPIHR, the Kolmogorov–Smirnov test was employed, showing a significance of 0.005; therefore, distribution was not normal.Not having normal distribution meant the non-parametric Kruskal–Wallis test was used to analyse categorical variables of study type if the concept of violence used refers to gender, if the study expresses the gender component in the violence and if the violence and decade in which it was published are assessed. As shown in Table 3, significant differences were seen regarding whether the study assesses violence in the “Sex difference” dimension and in the decade of the study both in the general score of the GPIHR and in the “Feminist research” dimension.For the variables in which significant values were found, the Mann–Whitney U test was employed, comparing whether there were differences 2 to 2. Of all the a posteriori comparisons made, statistically significant differences were found in the decade of study, both in total score of GPIHR and in the “Feminist research” dimension. As regards the total score, significant differences were found between the studies published up to 2000 and those between 2001 and 2010 (p = 0.023), with a higher average of scores in the first period. Likewise, significant differences were found between the decade from 2001 to 2010 and from 2011 to 2020 (p = 0.036), the mean score being higher in the most recent decade. Additionally, as for the “Feminist research” factor, significant differences were found when comparing the categories “Up to 2000” and “From 2001 to 2010” (p = 0.008), the average range being higher in the studies published up to 2000 than in those between 2001 and 2010. Finally, significant differences were also found in the scores in the “Sex difference” factor depending on whether the study assessed violence towards the mother (p = 0.035), finding higher mean scores in the studies where violence was assessed.Spearman’s correlation coefficients were calculated (Table 4) to assess if there were correlations between ordinal variables. Significant positive correlations were seen between year of publication of study and that the study defines the term of violence used and assesses violence towards mothers that children have experienced. Likewise, a significant positive relationship was observed between the variables “definition of the term of violence used” and “expression in the investigation of the sex component in intimate partner violence”. Additionally, a significant negative correlation was found regarding assessing violence and the experimental model. The latter was coded with value 1 for RCT, 2 for quasi-experimental, 3 for non-experimental and 4 for a single case. In this way, the more current the study, the greater the tendency for the methodology used to be considered of greater statistical validity.As for interactions of GPIHR scores with the rest of the selected variables, a significant positive correlation was only found for the “Sex difference” dimension with the variables “Year” and “Assessment of violence”. In other words, it was observed that the higher the scores in “Sex difference”, the higher the assessment of violence received by the mother.Likewise, it has been found that assessing violence towards the mother is significantly related to more recent studies, with higher methodological quality and with the variable “Sex difference” in the GPIHR. These results point to a tendency in current studies to consider the severity and typology of experiences of aggression towards mothers, as well as the quality of internal validity in the experimental design, with variables such as age, sex and sex difference in results.Finally, positive correlations were observed in all GPIHR dimensions with each other, except for “Sex difference” and “Feminist research”. This shows a trend where higher scores in any dimension make them more expected in other dimensions, except for “Sex difference” and “Feminist research”.The fact that the study generally takes gender in violence into account, that the term of violence used includes a gender factor, or that it assesses violence is not related to the GPIHR instrument. We only found a significant direct correlation with the variables “Year” and “Assessment of violence” with the GPIHR dimension “Sex difference”. However, this correlation between the two suggests it might be due to methodological factors inherent to the study rather than factors related to gender perspective.Among elements which help to explain these results, we find the low GP scores of the studies (explaining why there is no normal distribution of results) and the number of studies where the assessed variables were taken into account. By dimension, we found that from items of the “Gender sensitivity” dimension, 18% of occasions were answered positively, in the “Feminist research” dimension only 8% and in “Sex difference”, 68%. Nevertheless, for study variables, we found that only 11% (n = 5) used a term of violence with a gender component, 41% (n = 18) presented the sex component in intimate partner violence (more assaulted women or men who were aggressors) and 64% (n = 28) assess violence between parents in some way.Bearing in mind the content of items of the factor with the highest completion rate (“Sex difference”), we find this refers to processes related to basic methodology factors. Thus, we found high compliance rates of 84% and 95% due to the stratification of the sample by sex and age, respectively. However, only 25% of studies show or mention analysis by sex in their study results. It should be noted that there are no significant differences between study methodology and performance of these analyses. These data agree with the findings of the meta-analysis by Romano et al. [53], where the lack of data (or inadequate data) do not allow us to analyse the weight of differences in the intervention in basic variables such as sex.Despite low compliance in “Gender sensitivity” (18%), analysing item by item, the highest compliance is in the introduction referring to magnitude of the problem in boys and girls (30%). Next, 23% of research on methodology highlighted the relationship between the problem (internalizing, externalizing responses, etc.) and some gender factors. Among the most assessed variables, we found attitudes and beliefs about violence toward children; however, knowledge about abuse dynamics, behaviour in the face of violence towards their mother and self-blame were also seen. It should be noted that the same variable can be considered a gender factor in some studies, but not in others. One example is that of Muela et al. [77], where it is specified that they study possible relationships between sex and externalizing responses from a gender perspective and as an intervention mediator. Nonetheless, in the rest of the studies that take type of response and sex into account, the response was not considered a gender factor as it was not examined from the child’s socialisation. Similarly, the parenting variable was included as a gender factor in analysis by Jouriles et al. [66], given that it is seen as a moderating factor in results from aspects related to care, gender and impact of violence. In other selected studies, despite assessing aspects regarding parenting, a gender factor was not taken into account as it was not considered from a socialisation viewpoint.The “Gender sensitivity” factor item referring to the search for association between the health problem studied and some gender determinant through aims or hypotheses, was only met in 14% of studies (n = 6). We found that, based on the literature, Graham-Bermann et al. [56], hypothesised that the moderating variable “disclosure” of traumatic experiences in therapeutic groups would occur more in girls than in boys. Likewise, Muela et al. [77] hypothesised that boys would present greater clinical symptoms than girls, particularly externalizing. On the other hand, one hypothesis of Jouriles et al. [66] is that sex would moderate the relationship between frequency of contact with the mother’s aggressor and behavioural problems, predicting a greater relationship in girls than in boys. In the case of Hiltz-Hymes [60] and Pernebo et al. [79], they state among their aims the exploration of sex as a moderator in interventions, and Jaffe et al. [63] described their objective as exploring stereotypes and myths related to the male/female sex in the sample. Finally, the item regarding mention in the introduction of the study as to whether there was scientific literature with a gender perspective only appeared in 4% of research (n = 2). Specifically, Hiltz-Hymes [60] refers to research stating that psychopathology together with boys’ early exposure to violence can increase the likelihood that a person accepts and/or uses patriarchal ideologies to rationalise or justify their abusive or violent behaviour. On the other hand, Jouriles et al. [66] presents different studies explaining how socialization factors cause girls to be more affected than boys, deriving from continuous contact with their mother’s violent partner.The factor with the lowest compliance was “feminist research” (8%), which brings together items aimed at researching the causes of inequality with the aim of changing it. Only three studies consider gender category in the introduction as a determinant in the mental health of these children (item 3). Hiltz-Hymes [60] expresses how the ecological approach considers the influence of patriarchal structures, parental characteristics, sex, age and other factors as an explanation for children’s responses to parental gender violence. In the case of Jaffe et al. [63] the explanation of externalizing responses (violence as appropriate conflict resolution, sexism, gender roles in power inequality, etc.) is given from an approach based on modelling and identification with their mother or father. Jouriles et al. [66] explain the higher rate of externalizing responses in girls than boys due to the gender socialization process. They show how the fact that their socialization is “oriented towards others” (care, emotional recognition, etc.), making them more aware of the consequences of violence, and therefore, they adopt more aggressive and oppositional behaviour.Items 9 and 10 refer to the global analysis of the study. Specifically, the first addresses whether the study helps to increase knowledge of the mental health of girls or boys and diversity in its expression, with a success rate of 7% [56,63,66]. In the last item, research assesses if it helps to highlight changes in gender structure that may affect equality or equity in mental health between boys and girls who have experienced parental gender-based violence. In this case, the success rate is 9% (n = 4). Hiltz-Hymes [60] expresses the importance of working on gender roles in interventions with these children, aiming to, without relying on rigid gender roles, develop respect for themselves and others, as well as helping to deconstruct the negative self-image brought about from experiences of violence. On the other hand, Jaffe et al. [63] propose training the various agents who may have contact with these children (teachers, police, etc.) for detection and development of preventive programs, and express the need to continue research to assess the differential impact of group therapy in boys and girls. Suderman et al. [84] consider that, given the results obtained, their intervention is a way of changing attitudes and beliefs about abuse of women, abuse of partners and other forms of violence. Finally, Jouriles et al. [66] highlight how the visits with the aggressor variable can have several consequences in the development of behaviour problems, especially in girls, and the importance of reducing aggression in contacts. Among the different preventive measures, they suggest training of mothers, fathers, social agents (e.g., courts) and limitation of contacts with the aggressor.From the total number of studies analysed, only two used the term “gender-based violence” to refer to violence exerted on mothers [77,80]. In most cases (89%), a generic term was used, specifically “intimate partner violence” in 39% of cases and “domestic violence” in 32%, the rest being general terms with few repetitions. These data agree with the results of López-Cepero et al. [15], where the use of general terms was also found to refer to gender violence in the partner; in their case, “domestic violence” was most used. Another aspect to take into account is that although all research included in our analysis includes mothers and their children, only 41% state that violence is received mostly by women or exercised mainly by men. In other words, most studies still make the sex and gender component in violence invisible, either by action (by using general terms) or omission (by omitting information).Other factors detected showing the invisibility of gender in intimate partner violence are ambiguity when describing the sex of the attacked parent as “parent” and conducting research with mixed groups of fathers and mothers. Such studies comprised around 40% of records excluded in the last screening phase of the study (n = 19). It is striking that, of studies excluded by mixed samples of fathers, the percentage of assaulted mothers ranges between 81% and 98%. Nevertheless, exposure to violence is made from general terminology without mentioning gender. Only one item of research stated that the sample is mixed but did not provide data on the number or percentage of fathers and mothers [99]. However, data were provided on the frequency of the perpetrator of intimate partner violence, 65% by the father, 2.5% the mother and 32% both parents (not specifying whether in defence or the frequency of each parent). It is noticeable that in 87% of cases, the mother reported the child’s condition. By not providing data on the number of fathers and mothers, it is possible that this inequality is due to a high number of mothers in the intervention, which would evidence the invisibility of the number of women. In the case of an equal number of mothers and fathers, it would be explained how gender roles put the mother in charge of the care of the children, highlighting one inequality factor.Of the total number of studies selected, only one was found where the intervention was with the father and assesses the impact of the intervention on the children’s mental health. This is the quasi-experimental study by Satyanarayana et al. [81], where an intervention was carried out to reduce the father’s alcohol consumption in heteronormative families where gender violence occurred. In their hypotheses, they consider alcohol consumption a gender violence moderator causing distress to the children. Thus, they observe that lower alcohol intake will cause a reduction in gender-based violence and therefore children’s distress. In their results, they found statistically significant differences in the reduction in partner violence between the group receiving the intervention and the group that did not; however, this did not produce any change in the children’s distress. It should be noted that this study did not score on any of the GPIHR items; therefore, interpretation of its results lacks a gender perspective. Likewise, other studies were found where the father received the intervention, but these were discarded because results focused on recurrence of assaults on women, and not on the implications that this intervention might have on children’s mental health.An explanation of the difference in scientific production regarding evaluation of the consequences of intervention on children to a greater extent in mothers than in fathers may have to do with gender roles. Care is a traditionally feminine area, where the mother is expected to provide for the children. In the same regard, the tendency to not include men in the caregiving role, and only link them to aggression, alcohol consumption and violence, ignores the pedagogical impact of the father in the lives of children. Obviously, interventions must be different for the aggressor and the victim; however, it is often clear that both are parents.If we look at the temporal evolution of the studies in the analysis, we find a direct relationship between research methodology factors and publication timing. In other words, a significant relationship has been found regarding the greater timeliness of the publication, the greater frequency of definition of the term of violence used, and the more often that term is defined, the more the gender component in intimate partner violence is exposed. Therefore, it appears there is a certain tendency to operationalise concepts; however, the terms used remain general.When analysing the possible existence of a relationship between temporality and the gender perspective by decade of publication, we found significant differences both in the total scores of the GPIHR and in the factor “Feminist research”. Generally, these results indicate a temporal evolution of GP in “U”-shaped investigations. In other words, there is a downward trend from publications from before 2000 to the period 2001 to 2010, and subsequently a change with a significant increase from 2011 to 2020 compared to the previous decade. Nonetheless, we must treat these data with caution since, overall, most items answered positively are related to the methodological quality of the study. Therefore, the fact that there are significant differences in the increase at a general level from the decade of 2000 to 2020 does not imply that the investigations generally have GP. In addition, after comparing the “Feminist research” factor, we found significant differences in research prior to 2000 with those carried out between 2001 and 2010, with those prior to 2000 having higher scores.These results show that the methodological quality of the studies on interventions in minors with experiences of parental gender violence has increased significantly over the years; nevertheless, factors related to the gender perspective have not. This is of vital importance, given that it is a research context with a high gender component, and not having this perspective may bias interpretations of the data.Following assessment and analysis of the scientific literature published up to 2020 on interventions with CEXPGBV, we conclude that most studies do not contain a gender perspective. A total of 70% of GPIHR items were answered negatively, and items with the highest success rate related to basic research methodological aspects. Likewise, the lack of a gender perspective in studies and low rates of variables related to gender (term of violence used, specify prevalence of violence against women, etc.) hindered establishing any relationship between the two.On the other hand, we found a clear difference in compliance of GPIHR between the aspects related to sex and gender. In general, research considered sex at the descriptive level of the sample; however, in most, factors regarding the socialization of gender have been omitted. Likewise, in most studies evaluated, the term gender has been used as a synonym for sex.In the scientific literature, the operationalization of variables is vital, which is why the indiscriminate use of the terms sex and gender and the use of generic terms to describe gender-based violence hinder inclusion of GP in research in this population. Thus, at present, research faces the challenge of operationalizing social and socialization factors related to gender in childhood, while at the same time continuing to assess components related to the mental health of children. These difficulties might explain the clear trend of the studies towards methodological improvement and paralysis of the evolution of GP in the studies. However, we must bear in mind that methodological quality is not opposed to the gender perspective, both being complementary.We believe it important to express the methodological limitations of researching the efficacy of interventions in this population. Despite recommendations for the use of RCTs to control variability in the researched condition (focused on internal validity), in CEXPGBV healthcare contexts, this methodology is hardly applicable. Given the high prevalence of intrusive variables (visits with the aggressor, violent episodes during the intervention, changes of address, school, interrupted parental consent, etc.), we found a gap between experimental practice and care practice, since gender-based violence cannot be isolated in the laboratory, nor does it have markers. Faced with this situation, Barkham et al. [100] suggest a complement to the evidence-based practice (characterised by the use of RCTs), practice-based evidence (PBE). In PBE, external validity is prioritised with data from healthcare practice through the routine measurement of responses that generate distress in the person being attended to, in which measurement can be administered pre and post intervention, at repeated intervals or session to session.Finally, we consider that the GPIHR instrument offers basic aspects for evaluating gender perspective in health research. Among the notable factors, we find: (a) its length of 10 items enables ease of application; (b) applicability in most of the usual research sections; (c) its three dimensions provide clarity on basic components to consider that a study contains GP; (d) ease of adaptation to the population. Likewise, it is not a demanding questionnaire as regards GP in research, but rather offers the basic keys to consider that the research contains GP. Therefore, it is vitally important that future research related to children with experiences of parental gender-based violence take each item of the GPIHR instrument into account both in planning and in elaboration of study.Despite attempting to achieve the highest representativeness of the published research, we found various limitations. First, heterogeneity of sample. Gender-based violence, being so widespread and varied, means that among included studies we find variations in the recruitment of the sample (women’s shelters, community services, mental health, NGOs). Second, when trying to obtain the largest number of studies by including experimental, quasi-experimental, non-experimental and single-case studies, we found limited control of variability of studies. Third, it has not been analysed whether tests used both in evaluation of distress of the children and of the aggressions in the family context had GP. Fourth, one of the limitations of applying the GPIHR questionnaire is having prior knowledge regarding gender perspective.B.P.-R., C.L.-S. and M.V.A.L. contributed to the conception and design of the study. B.P.-R., C.L.-S. and M.V.A.L. organised the database. B.P.-R. performed the study search, gender perspective assessment, and statistical analysis and wrote the first draft of the manuscript. C.L.-S. and M.V.A.L. reviewed the process of systematic review, statistical analysis, and writing style. All authors contributed to the review of the manuscript, read and approved the submitted version. All authors have read and agreed to the published version of the manuscript.This research was funded by the University of Murcia (Q3018001B) and the Association for the Development of Mental Health in Children and Youth “I want to grow” (G73618357).Not applicable.Not applicable.Publicly available datasets were analyzed in this study. This data can be found here: Web of Science: https://www.webofscience.com; EBSCOhost; https://search.ebscohost.com; ProQuest: https://www.proquest.com; Cochrane Library databases: https://www.cochranelibrary.com.We thank the Mare Nostrum Campus of the University of Murcia and the Association for the Development of Mental Health in Children and Youth “I want to grow”.The authors declare no conflict of interest.Search elements for systematic review.Note. a = In Title.Items adapted from the questionnaire “Gender perspective in health research” (GPIHR) [8].Note. The questionnaire “Gender perspective in health research” (GPIHR) [8] comes from an open access article under the CC BY-NC-ND license.Selected Studies and GPIHR Scores.Flow diagram.Study characteristics.Percentage of compliance with GPIHR items.Probability levels (p) associated with the Kruskal–Wallis test.Spearman’s correlation coefficients.Note. * = The correlation is significant at the 0.05 level (bilateral); ** = The correlation is significant at the 0.01 level (bilateral).Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11048.txt

@@ -0,0 +1 @@
+The longitudinal quality of life (QoL) of COVID-19 survivors, especially those with post-acute sequelae (PASC) is not well described. We evaluated QoL in our COVID-19 survivor cohort over 6 months using the RAND SF-36 survey. From July 2020–March 2021 we enrolled 110 adults from the United States with a positive SARS-CoV-2 nasopharyngeal polymerase chain reaction (PCR) into the Northern Colorado Coronavirus Biobank (NoCo-COBIO). Demographic data and symptom surveillance were collected from 62 adults. In total, 42% were hospitalized, and 58% were non-hospitalized. The Rand SF-36 consists of 36 questions and 8 scales, and questions are scored 0–100. A lower-scale score indicates a lower QoL. In conclusion, hospitalization, PASC, and disease severity were associated with significantly lower scores on the RAND SF-36 in Physical Functioning, Role Limitation due to Physical Health, Energy/Fatigue, Social Functioning, and General Health. Long-term monitoring of COVID-19 survivors is needed to fully understand the impact of the disease on QoL and could have implications for interventions to alleviate suffering during recovery. The impact of the SARS-CoV-2 infection on physical and mental health, as well as social and emotional well-being, merits standardized evaluation and international research attention. Many COVID-19 survivors suffer from persistent dyspnea, fatigue, insomnia, and anxiety, amongst a range of other symptoms [1,2,3,4,5,6,7,8,9]. COVID-19 survivors also report low quality of life (QoL) 1–3 months after infection, in addition to significant impairment in physical and psychological functioning [10,11,12]. Physical health is closely related to QoL, and patients with measurable decreases in pulmonary function following COVID-19-related pneumonia also reported lower QoL [13]. Moreover, hospitalized patients who survived COVID-19 reported a decrease in short-term QoL [14,15,16]. The longitudinal QoL in survivors of COVID-19, especially those with post-acute sequelae of COVID-19 (PASC) lasting 6 months or more, needs further attention. The persistence of physical and mental COVID-19 symptoms have further challenged the concept of QoL [3] and likely contribute to the loss of QoL. We aimed to evaluate for differences in QoL in those with and without PASC. For this purpose, we chose to administer the RAND version of the SF-36. The Rand SF-36 consists of 36 questions and eight scales: Physical Functioning, Role Limitations due to Physical Health, Role Limitations due to Emotional Problems, Energy/Fatigue, Emotional Well-being, Social Functioning, Pain, and General Health. Questions are scored 0–100, and the higher the score, the better the level of functioning. This instrument has been validated for use in health related QoL, such as chronic headaches, Amyotrophic Lateral Sclerosis, traumatic brain injury, and stroke. In addition, the time to administer is minimal (approximately 10–15 min) compared to the WHOQOL-100, which is an internationally used QOL instrument that takes anywhere from 30–90 min depending on the literacy level of participants. Colorado State University (CSU), in conjunction with University of Colorado Health (UC Health), established the Northern Colorado Coronavirus Biorepository (NoCo-COBIO), an observational, longitudinal cohort of individuals infected with SARS-CoV-2 [17]. One objective of the biorepository is to collect information on the QoL of COVID-19 survivors using the RAND version of the SF-36 in the convalescent phase of the disease. The NoCo-COBIO includes multiple visits during the first 180 days from the initial polymerase chain reaction (PCR)+ diagnosis and involves a one year follow up. We hypothesized that participants with severe and moderate disease would have a decreased QoL compared to those with mild disease. We also hypothesized that those who experienced hospitalization and post-acute sequelae of COVID-19 (PASC) would have decreased QoL scores compared to those who were not hospitalized and did not develop PASC.An observational, longitudinal cohort design was used to recruit COVID-19 survivors to participate through the University of Colorado Health (UC Health) and Colorado State University (CSU) networks, as previously described [17]. Inclusion criteria for this study were a positive SARS-CoV-2 PCR test and ≥18 years. Pregnancy was an exclusion criterion. Each participant consented to undergo clinic visits at 4 different time points: at enrollment, and 1 month, 3 months, and 6 months after enrollment. Pre-existing conditions were collected from hospital electronic records and self-reported at the clinic visits. Body mass index (BMI) was calculated by self-reported height and weight for the non-hospitalized participants (n = 36), and hospital records were used for height and weight from hospitalized participants (n = 26). The National Institute of Health’s body mass calculator was accessed on 2 January 2021, (https://www.nhlbi.nih.gov/health/educational/lose_wt/BMI/bmicalc.htm) to calculate BMI. The RAND SF-36 was administered at least 15 days after infection and will be re-administered at 6 months and one year of follow-up for this biorepository cohort. All enrolled participants provided written informed consent. The study met the Helsinki Declaration guidelines. This research study, from the Northern Colorado Coronavirus Biobank (NoCo-COBIO), has been approved by CSU’s Research Integrity and Compliance Review Office Internal Review Board (IRB; protocol ID 20-10063H), as well as UC Health’s IRB (Colorado Multiple IRB 20-6043), and is registered with ClinicalTrials.gov (NCT05603677). The cohort was examined for the relationship of disease severity, hospitalization, and PASC with QoL using the RAND SF-36 survey [18]. The complete study design is illustrated in Supplemental Figure S1. Participants were categorized as having mild, moderate, or severe disease based on the Yale Impact Score which used oxygen requirements during the acute phase of COVID-19: no oxygen use was categorized as mild, a 1–5 L/min oxygen requirement was considered moderate, and a greater than 5 L/min oxygen requirement was documented as severe disease [19]. Demographic data and hospitalization status were obtained from participants at their clinic visits. Symptom surveillance was administered to evaluate new or persistent symptoms since the initial COVID-19 diagnosis at each follow-up visit. PASC was defined as having at least one of the following symptoms reported during the longitudinal surveillance period: fatigue, dyspnea, joint pain, chest pain, or cognitive dysfunction, at any or all of the follow-up visits [17]. Cognitive dysfunction was defined as absent-mindedness, forgetfulness, confusion, or problems with concentration. All data were de-identified and password-protected for inclusion in the database and prior to analysis. The Rand SF-36 consists of 36 questions and 8 scales measuring QoL [18]. The scales are: Physical Functioning, Role Limitations due to Physical Health, Role Limitations due to Emotional Problems, Energy/Fatigue, Emotional Well-being, Social Functioning, Pain, and General Health. All questions are scored 0–100, and a combination of questions make up the scales. The higher the score on a question, and ultimately a scale, the better the level of functioning [10,20]. Considerable evidence was found for the reliability of the SF-36 (Cronbach’s alpha greater than 0.85, and reliability coefficient greater than 0.75 for all dimensions except social functioning). Construct validity was confirmed by distinguishing between groups with expected health differences. To demonstrate validity, the SF-36 was able to detect low levels of ill health in participants who had scored 0 (good health) on the Nottingham health profile, the reverse scoring of the SF-36 [21,22]. For this study, we created an electronic version of the RAND SF-36 which was completed in person with the study team and which automatically calculated scores for the 8 scales, saving time and ensuring accuracy. Either Collaborative Institutional Training Initiate (CITI)-trained biomedical students, the study coordinator, team MD, and/or the principal investigators verbally administered the Rand SF-36 QoL questionnaire privately to the participants at the end of the study visits. Demographic data were presented as Mean ± SD or Frequency and percent (%). Independent T-tests or ANOVA (with a Tukey–Kramer p-value adjustment to determine which groups differed) were used to test the differences among characteristics of COVID-19 participants on all 8 scales of the Rand Version of the SF-36. ANCOVA was performed using body mass index (BMI) and age as covariates. All analyses were performed using SAS 9.4 (Cary, NC, USA). p < 0.05 was considered significant without adjustment for multiple tests. Data were collected on 62 participants from the NoCo-COBIO cohort with a mean age of 51.8 ± 16.6 years (Table 1). Most participants were non-Hispanic (53; 85%) and female (38; 61%) with no pre-existing medical conditions. Many participants were obese (27; 44%) or overweight (15; 24%), with a mean BMI of 30.3 ± 7.9. In all, 58% of participants had mild COVID-19 at the time of acute illness, 19% had moderate disease, and 23% suffered severe disease. A total of 26 (42%) participants were hospitalized for COVID-19, and 32 (51.6%) had PASC (Table 1). In total, 62 participants underwent the RAND SF-36 QoL surveillance. All participants were at least 15 days post-PCR+ test results at time of questionnaire. Mean day post PCR+ was 125.3 ± 71.8 days. Participants requiring hospitalization were compared to those who did not, and we found that hospitalized participants scored significantly on the following scales: Physical Functioning (p < 0.0001), Role Limitations due to Physical Health, Role limitations due to Emotional Problems, Energy/Fatigue, Social Functioning, and General Health (p < 0.05) (Table 2, Supplemental Figure S2). All hospitalized participants suffered either moderate or severe disease, as shown in Table 3. As expected, those with moderate and severe disease scored significantly lower than participants with mild disease on the same five scales (Physical Functioning, Role Limitation due to Physical Health, Energy/Fatigue, Social Functioning, and General Health) Table 3.SF-36 scores were compared in participants with and without PASC, and those with PASC had significantly lower scores in Physical Functioning, Role Limitation due to Physical Health, Role Limitation due to Emotional Problems, Energy/Fatigue, Social Functioning, Pain, and General Health (Supplemental Figure S3 and Table 4). SF-36 scores were compared between those with and without pre-existing medical conditions (Table 5). Participants with chronic conditions had significantly lower scores in the following: Physical Functioning, Role Limitations due to Physical, Energy/Fatigue, and General Health (Table 5). Additionally, SF-36 scores were evaluated for differences due to participant sex, ethnicity, and days post-SARS-CoV-2 PCR positive test. Females had a significantly lower Pain scale score than males (Supplemental Table S1). There was no difference between Hispanic and non-Hispanic participants in any of the scales (Supplemental Table S2). No significant differences were found in any of the eight scales between those who were 15–44-days post PCR+ and those who were 175+ days post PCR+ (Supplemental Table S3).Lastly, SF-36 scores were compared across body weight categories based on CDC definitions for BMI (≤19.9 = underweight, 20–24.9 = normal, 25–29.9 = overweight, >30 = obese). Obese individuals had reduced measures for Physical, Role Limitation due to Physical Health, Energy/Fatigue, and General Health compared to those who were normal weight or underweight (Table 3). ANCOVA was performed with BMI and age as continuous covariates in models, testing PASC, hospitalization, and disease severity as the independent variables and with the 8 SF-36 scales as the outcome variables. BMI and age were not significantly related to PASC, hospitalization, and chronic conditions (Supplemental Table S4). This study demonstrates that severe and moderate COVID-19 has a greater effect on QoL when compared to those with mild disease. Notably, most of the moderate and severe disease patients in this cohort were also hospitalized. We found significantly decreased scores in Physical Function, Role Limitations Due to Physical Health, Energy/Fatigue, Pain, Social Functioning, and General Health scales. Pre-existing medical conditions, PASC, and BMI categories also resulted in lower QoL scores. While those who were obese had lower scores on most SF-36 scales, BMI as a continuous covariate was not significantly related to any of the SF-36 scale scores, nor was age. Physical Functioning and Energy/Fatigue scales had the lowest scores on the RAND SF-36 QoL survey [23,24,25,26,27]. These results are consistent with previous studies, showing that 50–75% of non-hospitalized patients reported fatigue after COVID-19 [27]. Previous studies have shown that QoL is diminished at 1–3 months post-SARS-CoV-2 infection [20,22,28]; however, these participants from the U.S. still had lower QoL 6-months after PCR+ diagnosis. Although some studies have found an association between obesity and severe COVID-19 disease, the BMI dropped out of the models when included as a continuous covariate with the other significant variables in this study [23]. Considerable attention has been raised by COVID-19 survivors who have lingering symptoms for months after the infection. PASC affects patients across the spectrum of disease severity [3,27,28,29,30,31,32]. This study validates the significant impact of PASC on survivors’ QoL. Further research into the mechanisms behind PASC, early identification of those who will develop PASC, and methods to prevent, treat, or support those who develop PASC is warranted. Interestingly, despite the physical impact of COVID-19, Emotional Well Being on the SF-36 was not significantly impacted in any of the patient groups, even those with PASC; however, the scale for Role limitations due to Emotional Problems was impacted in those who were hospitalized or experienced PASC. Previously, chronic conditions, such as asthma, have been associated with severity of COVID-19 and correlated with decreases in all SF-36 scores [33]. Doll et al. reported that obesity has a higher impact on physical rather than emotional well-being measures, and proposed that assessments of emotional well-being can be confounded by the presence of chronic comorbidities [23]. Although Emotional Well Being was not impacted in this sample, those with PASC had significantly lower scores for Role Limitations due to Emotional Problems. Continued investigation of patients with PASC is warranted. This study emphasizes that COVID-19 affects long-term patient health, an awareness of which can assist clinicians in identifying those who may be at risk for diminished QoL, such as the hospitalization of patients, presence of PASC, and those having underlying chronic conditions including obesity. Accordingly, the medical and scientific community can work together with these patients to improve their QoL over the long-term. Some limitations are present in this study. First, our sample was limited to participants residing in Northern Colorado within the U.S. Second, we did not have baseline SF-36 QoL scores for these adults prior to acute COVID-19 infection that could be compared to scores during the convalescence stages of the disease. Third, the assessment of PASC related to QoL may have been influenced by the definition of PASC used. The longitudinal symptom surveillance included ongoing fatigue in response to COVID-19, and the SF-36 instrument captures fatigue as one of the questions for the Energy/Fatigue scale. Fourth, physical isolation has occurred more frequently during the COVID-19 pandemic, and is associated with stress, depression, unhealthy diet, and reduced physical activity in other countries which may have affected our results [28]. Finally, there may have been a potential bias due to participants responding to a clinician asking questions, compared to other study personnel.We found that QoL in COVID-19 survivors is impacted by hospitalization, disease severity, PASC, and possibly obesity. These findings are important for clinicians across medical specialties and public health practitioners to assist COVID-19 survivors to regain QoL and prior functional status. The future of COVID-19 survivors, especially those with persistent symptoms and PASC remains unclear. As we gain more understanding of the effects of COVID-19, we can seek interventions to prevent and approach obstacles in survivors. The results of this study strongly support the continued longitudinal collection and assessment of QoL and symptom surveillance from COVID-19 survivors to help address concerns related to the identification of reduced QoL. Targeted approaches for the prevention, control, and treatment of PASC is needed to improve QoL outcomes globally. The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111048/s1, Figure S1: Study design, Figure S2: Comparison of scales of the SF-36 for those who were hospitalized and those who were not hospitalized due to COVID-19, Figure S3: Comparison of those with PASC and those without PASC due to COVID-19 and the scales of the SF-36, Table S1: Comparison of SF-36 Scores between Males and Females with COVID-19, Table S2: Distribution of SF-36 scores by ethnicity, Table S3: Distribution of SF-36 scores according to the number of days post-SARS-CoV-2 PCR positive diagnostic test results, Table S4: Comparison of SF-36 scores across body mass index categories of COVID-19 participants.E.P.R. and J.A.D. conceived, designed, and conducted the research; K.M., forma analysis; K.M., writing—original draft preparation; B.A.B., S.M.L. and T.L.W., writing—review and editing; B.A.B., S.S., S.M.L., K.B., M.T., J.H. and J.L. administered survey and data entry for analysis; E.P.R. and J.A.D., funding acquisition. All authors have read and agreed to the published version of the manuscript.The Northern Colorado Coronavirus Biorepository project was created and designed with funding and translational research infrastructure in the Colorado State University department of Environmental and Radiological Health Sciences, and with pilot funds obtained from the CSU Vice President for the Research Office to Elizabeth P. Ryan. This project was initiated in partnership with University of Colorado Health (UCH), Northern Colorado Trauma Research Department, and funding support obtained from the UCH North Foundation Funds to Julie Dunn.This study was approved by the Colorado State University Research Integrity and Compliance Review Office Institutional Review Board (IRB; protocol ID 20-10063H) and University of Colorado Health (Colorado Multiple Institutional Review Board 20-6043) and is registered with ClinicalTrial.gov (NCT05603677). The study was conducted according to the guidelines of the Declaration of Helsinki.All participants provided written informed consent to participate and this process was administered by University of Colorado Health Trauma Research Department program staff or a Colorado State University study coordinator. Participants had the option to revoke their consent at any point throughout the study and have signed additional informed consent forms with any amendments to the study protocol.All data in this manuscript is available through REDCap. Contact the corresponding author for data.The authors declare no conflict of interest.Demographics of COVID-19 adult participants that completed the Rand SF-36.Note: Disease severity was determined by oxygen use in accordance with the Yale Impact Score: mild (no oxygen required), moderate (1–5 L oxygen use), severe (<5 L oxygen use). Pre-existing conditions were collected from hospital electronic records and self-reported at clinic visits: CVA cerebrovascular accident, COPD chronic obstructive pulmonary disease, DM diabetes mellitus, HTN high blood pressure. Values are presented as mean ± standard deviation or frequency (percent). * BMI = Body Mass Index, PCR = polymerase chain reaction, PASC = post-acute sequelae of COVID-19; 3 participants were mildly underweight (BMIs = 18.8, 19.4, 19.9).Comparison of SF-36 scores in adults hospitalized for COVID-19.Values are presented mean ± standard deviation.SF-36 scores according to COVID-19 disease severity.Note * p < 0.05 compared to Mild. The Omnibus test for Social Functioning was significant, but there were no significant differences among the groups when Tukey’s post-hoc test was performed, which adjusts for multiple comparisons. Values presented are mean ± standard deviation.Comparison of SF-36 scores from participants with and without post-acute sequelae of COVID-19 (PASC).Values presented mean ± standard deviation.Comparison of SF-36 scores between adults with and without pre-existing chronic conditions.Values presented are mean ± standard deviation. Pre-existing chronic conditions were collected from hospital electronic records and self-reported at clinic visits.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11049.txt

@@ -0,0 +1 @@
+Intensive care unit discharge is an important transition that impacts a patient’s wellbeing. Nurses can play an essential role in this scenario, potentiating patient empowerment. A systematic review was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (the PRISMA Statement. Embase), PubMed/MEDLINE, CINAHL, Cochrane Central Register of Controlled Trials (CENTRAL), CUIDEN Plus, and LILACS databases; these were evaluated in May 2021. Two independent reviewers analyzed the studies, extracted the data, and assessed the quality of evidence. Quality of the studies included was assessed using the Cochrane risk-of-bias tool. Of the 274 articles initially identified, eight randomized controlled trials that reported on nursing interventions had mainly focused on patients’ ICU discharge preparation through information and education. The creation of ICU nurse-led teams and nurses’ involvement in critical care multidisciplinary teams also aimed to support patients during ICU discharge. This systematic review provides an update on the clinical practice aimed at improving the patient experience during ICU discharge. The main nursing interventions were based on information and education, as well as the development of new nursing roles. Understanding transitional needs and patient empowerment are key to making the transition easier.The number of critically ill patients has increased during the last decades. In Spain, 240,000 adults are admitted into intensive care units (ICU) each year [1]. Patients with potential life-threatening processes and vital organ dysfunction who require specialized and continuous care are admitted to the ICU [2], of whom >90% survive ICU admission [3]. Recently, ICUs have become essential in caring for seriously-ill patients admitted due to the COVID-19 pandemic [4]. The ICU plays an important role in the care process of many patients. Once patients are sufficiently stable and care can be stepped down, they can be discharged to the general ward, providing continuity of care.Discharge or transition of the patient from the ICU to a general ward is one of the most challenging, high-risk, and inefficient care transitions because patients who are among the most seriously ill are transferred from high-tech units to less acute environments, which involves many professionals in the exchange of information and responsibilities [5]. ICU discharge is therefore a complex process, and patients’ feelings and perspectives, including a sense of displacement, anxiety, and loss of autonomy, are crucial factors [6,7]. Patients feel powerless in this context, and the lack of medical knowledge and loss of control over one’s body are seen as the main factors behind these thoughts [8].Furthermore, these patients’ feelings and perceptions during ICU discharge could increase the risk of post-ICU syndrome (PICS). PICS can develop due to mental and cognitive impairments, physical disabilities, and psychological factors (anxiety, depression, and post-traumatic stress disorder (PTSD)) [9]. Effective interventions in ICU survivors are essential to decrease negative outcomes and increase the quality of life. Needham et al. (2012) suggested that effective interventions in patients to improve long-term outcomes after ICU discharge should focus on early psychological intervention, early mobility programs, post-discharge follow-up programs, ICU diaries, healing care environments, functional reconciliation, and the ABCDEFGH bundle (Airway management, Breathing trials, Coordination of care and Communication, Delirium assessment, Early mobility bundle, Family involvement, Follow-up referrals and Functional reconciliation, Good handoff communication, and Handout materials on PICS and PICS in Family (PICS-F)) [10,11] which addresses the risks factors for PICS, sedation, delirium, and immobility. Therefore, preparation of the ICU discharge process to carry it out accurately and correctly could be the cornerstone of a decreased risk of PICS afterwards.In this sense, the nurses within the multidisciplinary team of the ICU develop a fundamental role in the ICU transition planning process, as they are the ones who participate in, organize, and carry out the direct interventions of patient care during the transition [12]. Thus, it is the nurses’ responsibility to assess the needs of patients during the transition and provide adequate information and education to the patient and family. To improve the efficiency of the role of ICU nurses in patients during the transition of ICU patients, some hospitals have even introduced a new nursing role called “liaison nurse” [13,14]. The competent role of ICU nurses in planning and directing the implementation of a multidisciplinary program during ICU transition that could reduce ICU readmission and hospital mortality has also been highlighted [15].Another way in which nurses can begin to recover power to their patients is to be aware of signs and symptoms that indicate feelings of powerlessness [16]. Empowerment is a complex, multi-dimensional concept [17,18,19,20,21,22] that was introduced to allow patients to shed their passive role and play an active part in the decision-making process of their health and quality of life [20]. Successful empowerment occurs when patients come to terms with their threatened sense of security and identity, and they have a sense of control over their lives [8,23]. The benefits of improving empowerment are extensive, including decreased levels of distress and strain, an increased sense of coherence and control over the situation, and personal development and growth, together with increased comfort and inner satisfaction [24].Patient empowerment could be a useful tool to reduce the stress associated with ICU discharge. Empowerment strategies have increased over recent years, mainly self-care in chronic illness such as diabetes [25], cancer, [26,27] and other clinical scenarios [28,29]. However, their role in ICU discharge is less well known. Although this is the responsibility of the multidisciplinary healthcare team, this transition, including patient empowerment, is usually carried out by nurses [30,31].We conducted a systematic review to provide evidence on patient empowerment interventions, identify remaining gaps, and suggest directions for future research and clinical practice during ICU discharge. The main aim was to determine the effects of nursing interventions to improve patient empowerment during ICU discharge in adult patients.We performed a systematic review using the Preferred Reporting Items for Systematic Reviews (PRISMA) guidelines [32] (Supplementary Material File S1). The review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) CRD42021254377 (Supplementary Material File S2).We included randomized controlled trials (RCTs) in adults of both sexes undergoing ICU discharge. The studies included aimed to determine the effect of nursing empowerment interventions on patient wellbeing. Studies were included according to the population, intervention, comparison, and outcome (PICO) criteria (P: adults during ICU discharge; I: patient’s empowerment interventions performed by nurses; C: no intervention; O: physical and mental health symptom, patient satisfaction, and readmission). Accordingly, all studies had at least one group of patients with a nursing empowerment intervention and another with usual care during ICU discharge. The nursing empowerment intervention was defined as information, behavioral instructions, and advice on the management of ICU discharge by verbal, written, audio, or video-taped means. Studies in groups of adults at ICU discharge were also included.We included studies if they fulfilled the following criteria, (1) original research; (2) patients’ admission to the ICU; (3) reported impact of the nursing intervention; (4) full text available, without language restriction. We excluded studies with patients under 18 years of age. In addition, all observational studies, editorials, letters to the editor, review articles, systematic review, and meta-analysis, in vivo, and in vitro studies were excluded.The main outcome was nursing empowerment interventions developed for ICU discharge in ICU survivors. The secondary outcomes were the effects of the nursing interventions in this process.We reviewed four databases, including Embase, PubMed/MEDLINE, CINAHL, Cochrane Central Register of Controlled Trials (CENTRAL), CUIDEN Plus, and LILACS. The search was conducted on 17 May 2021 and was completed by selecting additional publications from the reference sections of the articles included using the search terms (Table 1). The terms selected were combined using Boolean logical operators (or, and, not). All references were analyzed using Rayyan software (http://rayyan.qcri.org accessed on 10 July 2021) a web-based tool [33]. To ensure thoroughness, we subsequently performed a cited-reference search (“reverse search”) for each article using Google Scholar and reviewed all results of each search.The review was performed independently by two investigators (CC and RTC) with experience in literature reviews. Primarily, it consisted of reviewing the titles and abstracts of all references retrieved by the database searches (CC and RTC). We searched all articles deemed potentially eligible by one or both reviewers. Secondly, the retrieved full texts were evaluated, and a decision on inclusion or exclusion was made according to the predefined selection criteria (CC and RTC). Any disagreements were resolved by a consensus, and a third reviewer was not necessary. Studies that did not fulfill the predefined criteria were excluded, and their bibliographic details were listed with the specific reason for exclusion.Two authors (CC and RTC) extracted the data independently and used standardized protocol and reporting forms in duplicate. The following information was extracted from each study included: design, population characteristics, nursing intervention, and results. If relevant data were not included in the article, the researchers searched for more information in the supplementary data, or the author was contacted to request the information.Narrative methods of synthesis were used to synthesize the included studies. The outcomes were not sufficiently similar enough to perform a meta-analysis. Each study was summarized and described with regard to participants’, characteristics of interventions, the instrument used, and critical outcome results, and this was checked by another reviewer (RTC). One table was created (Table 2).The risk-of-bias of the studies included was assessed independently using the Cochrane risk-of-bias tool [34]. To minimize bias, studies were graded independently by two reviewers (RTC and YT) and discrepancies were resolved by a consensus, and a third reviewer was not necessary.The flow chart of the study selection process is shown in Figure 1. The initial search found that 274 references and 259 studies remained after removing duplicates. After abstract and title screening, 238 studies were excluded. Twenty-one full texts remained, which were assessed for eligibility, leading to the exclusion of 13 studies due to wrong study design (n = 7), wrong population (n = 4), and wrong outcome (n = 2). Eight articles were finally included [35,36,37,38,39,40,41,42]. Therefore, the intervention of the third reviewer was not necessary. The reverse search did not return any additional references.Four studies were conducted in England [35,37,39,42], two in the USA [36,41], one in Iran [38], and one in Turkey [40]. Individual study characteristics, including the study design, sample and setting, interventions, and outcomes, are summarized in Table 2.In total, 2408 patients were enrolled in the included studies. The sample size ranged from 36 [37] to 1458 [39], and 1108 (46%) of patients received the nursing intervention. The patients’ age included in the studies ranged between 54.6 ± 7.9 [38] and 60.4 ± 15.0 [39] in the nursing intervention group. A summary of patient characteristics is shown in Table 2.There were wide variations in bias in all articles included. The randomization method and allocation concealment were adequate in most trials. In one of the eight trials, participants were blinded to the treatment allocation [37], and in two studies, there was a blinded outcome assessor [35,42]. Most trials, however, lacked blinding of participants, personnel, or outcome assessment (Figure 2). However, close to half of the authors provided insufficient information to assess whether a critical risk of bias existed for other sources of bias. Therefore, the intervention of the third reviewer was not necessary. The results of the quality assessment are shown in Figure 3.Various nursing interventions were made in the studies selected:Information/education interventions. Three RCTs included information skills training programs or educational programs [35,37,40] for patients and family and one “magic empowerment program” with information and education [38] only for patients.Discharge planning. One study assessed discharge needs and defined early discharge planning [36].ICU recovery/therapeutic environment and complex interventions. Three studies included a change in routine by the nurse-led preventive psychological intervention for critically ill patients [39], with hospital rehabilitation, comprising enhanced physiotherapy, nutritional care and information provision, case-management by a ward-based clinical team [42], and an interdisciplinary recovery program with a nurse practitioner and case manager [41].Information/education interventions. Three RCTs included information skills training programs or educational programs [35,37,40] for patients and family and one “magic empowerment program” with information and education [38] only for patients.Discharge planning. One study assessed discharge needs and defined early discharge planning [36].ICU recovery/therapeutic environment and complex interventions. Three studies included a change in routine by the nurse-led preventive psychological intervention for critically ill patients [39], with hospital rehabilitation, comprising enhanced physiotherapy, nutritional care and information provision, case-management by a ward-based clinical team [42], and an interdisciplinary recovery program with a nurse practitioner and case manager [41].Anxiety and depression: Knowles et al. (2009) [37] found that a diary with daily information about patient health reduced anxiety and depression in the experimental group. In the same line, Kuchi et al. (2020) demonstrated changes in patient attitudes toward risk-motivated behavior, and they improved physical health with information and education. Demircelik et al. (2015) found less anxiety and depression in patients receiving nursing education through multimedia interventionPost-traumatic stress disorder: Wade et al. (2019) [39] performed an RCT in 1458 adults post ICU and found no significant differences in PTSD symptom severity at six months among groups.Perceived risk score: Kuchi et al., (2020) [38] in 84 cardiovascular patients, found significant differences between the intervention and control group in the total score of perceived risk and its subscales.Patient satisfaction: Ramsay et al. (2016) [42] assessed the patient satisfaction with the PEQ and revealed significant differences between groups suggesting greater patient satisfaction in the EG.Hospital readmission: Bloom et al. (2019) [41] found that after discharge, at seven days, the readmission rate was 3.6% and 11.6%, in the intervention and control group, respectively. At 30 days, the readmission rate was 14.4% vs. 21.5% in the intervention and control groups, respectively.We aimed to study nursing interventions based on patient empowerment during ICU discharge and analyze their effects. To the best of our knowledge, this study is the first to systematically review empowerment interventions in patients during ICU discharge.Few studies to date have analyzed the impact of information and education on patients during their ICU stay and discharge, and most have limitations in the design, sample, and lack of randomization [43]. Patient empowerment studies in other fields have shown that nursing interventions improve patient stress, anxiety, and depression [44,45] (Figure 4). Patients’ emotional states should be evaluated to determine where, how, and when to intervene and ensure that the patient is emotionally prepared for the change between the ICU and the general ward. Situational control is one of the main goals of patient empowerment in this stage [21,23,46]. Patients admitted to the ICU usually feel that they have lost control of their lives, especially those with severe conditions who require sedation and mechanical ventilation, making them totally dependent and unable to decide. Faced with this situation, patients have to adapt to being dependent on others and accept how they carry out procedures, which results in a loss of control of the situation and feelings of helplessness. In addition, complications and slow recovery are, in turn, the cause of delays in transfer to the general ward, increasing daily the feeling of lack of control of the situation. Meleis et al. (2000) found that preparation and knowledge make it easier to empower people for their transition, while a lack of preparation acts as an inhibitor [47]. Therefore, it is necessary to create an environment in which returning control of the situation to the patient is prioritized and in which nurses are responsible for ensuring that patients can receive knowledge according to their expectations [48,49].In four out of eight studies, the main intervention was information and education for patients and relatives [37,38,40,50]. Various studies have demonstrated the importance of these issues [7,43] and have shown that, when they are lacking, it is more difficult for patients to participate actively during the transition [51].Four of the evaluated studies explored the impact over patients’ emotional well-being [35,37,38,40]. Knowles et al. (2009) found that a diary with daily information about patient health was beneficial. However, Bench and Day (2015) did not find that written and verbal information during discharge improved patients’ emotional state, specifically anxiety and depression. This may be due, at least partly, to the late intervention; better results might have been achieved if it had been administered early, which could have helped patients to have a more informed perspective.Other interventions in the review described the determination of patient needs through questionnaires, followed by development of an individualized recovery, and a discharge plan of care with nurse interventions specifically aimed at these needs. Constant evaluation of the needs of patients admitted to the ICU during their stay and transfer to the general ward is necessary to generate an updated and structured care plan that helps ensure the continuity of care, even though drawing up these plans takes time. Kleinpell et al. (2004) demonstrated that such an intervention was associated with patients being better prepared for both ICU and hospital discharge.Complex interventions addressed towards patient recovery have demonstrated greater patient satisfaction [42] and reduced rates of ICU readmission and mortality [41]. However, Wade et al. (2019) found that nurse-led interventions were not associated with a decrease in PTSD after ICU discharge [39]. Therefore, it is necessary to improve the evaluation and measurement of the effectiveness of these nursing interventions to determine their real benefits and how they could contribute to positive results during ICU discharge.The evaluation and follow-up of patients during ICU discharge by an advanced practice nurse in a multidisciplinary team was another of the interventions studied. This role appeared in three studies, with two different denominations, including nurse-led [39] and case management [41,42]. Although advanced practice nurses were introduced more than two decades ago as part of the multidisciplinary teams to care for patients with complex needs, they have only been involved in ICU discharge in the last few years. These new roles represent an opportunity to help patients and families regain their sense of control and to cope with the new situation outside the ICU environment.In the studies included in this review, the term patient empowerment was not explicitly used, but concepts related to empowerment were studied. This result was also found in another systematic review related to empowerment in online communities [50] where 30% of the studies did not use the term empowerment for the intervention. This may be because, despite the various existing definitions of empowerment, the elements that intervene in the concept of power/empowerment and that include control, psychological coping, legitimacy, support, knowledge, and participation, as well as highlight the need for patient empowerment researchers to broaden their perspectives from individual to structural aspects of power and empowerment [51]. In another review of the empowerment concept, the authors proposed that improving the patient’s empowerment would be necessary {Formatting Citation}. In this sense, to change the conceptual and operational ways of considering patient empowerment only as individual and interpersonal elements, but rather patients need a high level of self-efficacy and control of the health situation, and we must integrate the concept of autonomy and the perceived capacity of the patient [52].Finally, the main contribution of this systematic review is the proposal of nursing interventions to apply the empowerment of the patient during the transition from the ICU. Future studies with designs using rigorous methodologies will increase the quality and credibility of these interventions. Therefore, it is necessary to improve the evaluation and measurement of the effectiveness of these nursing interventions to determine their real benefits and how they could contribute to positive results during ICU discharge.Most interventions in this review were carried out by nurses to help ICU survivors. Nurses, when appropriately informed and educated, can apply empowerment interventions to improve the transition from the ICU to the general ward. Our results show the impact on patient empowerment during ICU discharge, with important clinical implications. It is important to detect psychological adverse effects in ICU discharge patients. We suggest that a routine evaluation of anxiety and depression in ICU patients at discharge should be mandatory, as it will permit to carry out a specific intervention to whom they will benefit and assess how beneficial it would be.However, the concept of empowerment introduced into health should be assimilated and understood by ICU nurses to use it as such and intervene in the patient. In addition, patient dependence on care and needs are not conditions those nurses must automatically and equally assume for all patients transitioning from the ICU to the general ward. Each situation should be evaluated, and each patient’s responses and expectations should be considered individually to ensure adequate care for their needs, taking decision-making and preferences into account. Likewise, the care provided by nurses on the general ward would be much easier if patients had more control of the situation and if they were informed of the changes between the ICU and the general ward [48].Our systematic review has some strength. We conducted a comprehensive search of the literature, including full-text publications, without language restrictions or filters in the search strategy. Although we included studies published between 2000 and May 2021, it is unlikely that previous relevant trials were missed. The process of the systematic review was rigorous, and all the reviewing authors were appropriately trained and have experience in reviewing manuscripts.This study has some limitations. The literature review included only articles that had the words “patient” and “empowerment” and “ICU discharge” in the title or abstract and may therefore have excluded some reported interventions on patient empowerment.Various nursing interventions during ICU discharge that focused on empowerment were carried out in the studies selected, and this included information, determination of the discharge needs and outcomes of critically ill patients, nursing care plans and assessment, and follow-up by advanced practice nurse. In almost all the studies analyzed, the main intervention was information and education of patients and families. Most of them were associated with benefits from the perspective of controlling the situation and improving negative emotional effects.Nursing interventions using patient empowerment may have positive effects during ICU discharge. This review may help other projects in a similar context to implement new nursing interventions to empower the patient during ICU discharge. In particular, it is important to identify the nursing intervention to contribute to patient empowerment in critical care and, especially, to assess its impact on the different patient dimensions and outcomes.Future research should focus on the most effective methods of information, education, and patient empowerment during ICU discharge. It would also be useful to conduct more research on interventions that aim to reduce negative effects following transfer, such as structured teaching and information programs. Further research on what the transfer experience means to critical care patients and what effects it has in the immediate post-transfer period is also required. A combination of qualitative and quantitative measures may be needed to evaluate the effect of nursing interventions on patient outcomes.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111049/s1, File S1: Preferred Reporting Items for Systematic Reviews; PRISMA checklist, File S2: Systematic review registration: International Prospective Register of Systematic Reviews PROSPERO 2021 CRD42021254377.C.C. and R.T.-C.: responsible for the protocol, conceptualization, formal analysis, methodology, reviewing procedure and data extraction, writing—original draft, writing—review and editing. Y.T.: reviewing procedure and data extraction, writing—original draft, writing—review and editing. P.C.: resolve disagreements between C.C. and R.T.-C., conceptualization, writing—original draft, writing—review and editing. I.M. and P.M.-R.: formal analysis, methodology, writing—review and editing. M.R.-G., M.A.M.-M.: writing—review and editing. G.M.-E.: funding acquisition. P.D.-H. and P.C.: conceptualization, formal analysis, methodology, supervision, writing—original draft, writing—review and editing. All authors provided critical revision of the protocol and final manuscript. All authors have read and agreed to the published version of the manuscript.This research was funded by Nursing and Society Foundation as part of the Nurse Research Projects Grants; grant number PR-248/2017.Not applicable as we only reviewed published studies.Not applicable.Not applicable.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.PRISMA Flowchart.Risk of bias assessment summary.Quality assessment of studies included.Main findings of patient empowerment strategies performed by nurses.Terms of searching in the different databases and results obtained.Main characteristics of the studies included.* Instrument used to assess the impact of the intervention. Abbreviations: BCOPE: Brief Coping Orientations to Problems Experienced, CG: Control group, DPQ: Discharge Planning Questionnaire, EG: Experimental group, HADS: Hospital Anxiety and Depression Score, HrQoL: Health-related Quality of Life, ICU: Intensive Critical Unit, ICU steps: Intensive care guide for patients and relatives, IQR: Interquartile range, SAQ: Seattle Angina Questionnaire. PEI: Patient Enablement Instrument; PEQ: patient experience questionnaire, PTSD: Post Traumatic Stress Disorder, SF-12: Short Form 12 Health Survey, STAI-6: State Trait Anxiety Inventory 6-item version, UCCDIP: User-Centered Critical Care Discharge Information Program, UK: United Kingdom.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11050.txt

@@ -0,0 +1 @@
+Total Worker Health® (TWH), an initiative of the U.S. National Institute for Occupational Safety and Health, is defined as policies, programs, and practices that integrate protection from work-related health and safety hazards by promoting efforts that advance worker well-being. Interventions that apply the TWH paradigm improve workplace health more rapidly than wellness programs alone. Evidence of the barriers and facilitators to the adoption, implementation, and long-term maintenance of TWH programs is limited. Dissemination and implementation (D&I) science, the study of methods and strategies for bridging the gap between public health research and practice, can help address these system-, setting-, and worker-level factors to increase the uptake, impact, and sustainment of TWH activities. The purpose of this paper is to draw upon a synthesis of existing D&I science literature to provide TWH researchers and practitioners with: (1) an overview of D&I science; (2) a plain language explanation of key concepts in D&I science; (3) a case study example of moving a TWH intervention down the research-to-practice pipeline; and (4) a discussion of future opportunities for conducting D&I science in complex and dynamic workplace settings to increase worker safety, health, and well-being.The Total Worker Health® (TWH) approach from the National Institute for Occupational Safety and Health (NIOSH), part of the U.S. Centers for Disease Control and Prevention (CDC), first arose in 2003. TWH is defined as policies, programs, and practices that integrate protection from work-related safety and health hazards by promoting efforts to advance worker well-being [1,2]. Interventions with a TWH focus have been demonstrated to improve workplace health effectively and more rapidly than wellness programs alone [3,4]. For example, the Wellworks-2 intervention integrated an occupational safety and health (OSH) program targeted at reducing workplace exposure hazards with a health promotion (HP) program to reduce tobacco consumption and increase healthy eating [3,5]. Results from a randomized controlled trial of Wellworks-2 demonstrated, among a number of outcomes, significantly greater smoking quit rates and reduced hazardous substance exposure ratings than an HP-only program did [3,5]. Although TWH efforts and activities have increased in recent years and have generated both national and international attention and interest [2], TWH is still an emerging area with minimal research addressing interventions appropriate to the changing nature of work in the United States and worldwide. Complex OSH challenges—including large-scale public health crises such as the COVID-19 pandemic, the rise of globalization, automation, demographic shifts, increased psychosocial hazards (including high job demands, work-related fatigue, and job stress), and the interaction of work and nonwork factors [6]—require wider and faster adoption of TWH approaches that benefit workers, employers, and society [7].To increase the impact (including representative reach to diverse worker populations), effectiveness, and integration (including the adoption, implementation, and maintenance [8,9]) of TWH activities, systematic approaches are needed to shed light on the complex processes involved in moving evidence-based TWH interventions into sustained practice [2,3,10,11,12]. Systematic and other reviews acknowledge the need for dissemination and implementation (D&I) science—defined as the study of methods and strategies for bridging the gap between public health research and practice [13]—to advance the TWH field. For example, Punnett and colleagues [12] suggest that D&I science can be utilized in TWH investigations to characterize factors affecting TWH program uptake, successful implementation, scale-up, and sustainment. Furthermore, Anger and colleagues [3] call for research in D&I science to “determine what works”, and what should therefore be included in future TWH research agendas. D&I science inquiry can be expanded not only to explain the important question of “what works”, but also to address key contextual issues such as what works, for whom, how, in what settings, and how it is sustained over time [14].A helpful, plain language tool from Curran [15] to describe key, D&I concepts posits that:The intervention is THE THING;Effectiveness research looks at whether THE THING works;Implementation research looks at how best to help people (e.g., employers and workers)/(work)places DO THE THING;Implementation strategies are the stuff researchers do to try to help people/(work)places DO THE THING as designed/intended (i.e., with fidelity), such as provide training, technical assistance, and/or incentives;Main implementation outcomes are HOW MUCH and HOW WELL they DO THE THING.The intervention is THE THING;Effectiveness research looks at whether THE THING works;Implementation research looks at how best to help people (e.g., employers and workers)/(work)places DO THE THING;Implementation strategies are the stuff researchers do to try to help people/(work)places DO THE THING as designed/intended (i.e., with fidelity), such as provide training, technical assistance, and/or incentives;Main implementation outcomes are HOW MUCH and HOW WELL they DO THE THING.Missing from the Curran definition is another critical consideration, and that is related to the context of the intervention. As mentioned above, context asks the question when, where, how, with whom, under what circumstances, and why does “the thing” work? [9,14,16,17]. Contextual factors identified in TWH studies include: (1) the legal–regulatory environment (e.g., state laws with respect to union representation); (2) employer characteristics, policies, or benefits (e.g., availability of health insurance coverage or paid sick leave); (3) work organization (e.g., shift work); and (4) social or economic factors (e.g., income or availability of community resources to support or promote health) [18]. Given the variety of contexts in which the TWH model can be implemented and studied—with variation in employers, work environments, and workers—understanding the factors that influence the effectiveness of integrated interventions is important. Currently, there is a gap in the TWH research in this area [10].In sum, applications of D&I science hold promise for addressing major OSH challenges, [19,20] including those addressed through the TWH paradigm. More investment in resources tailored to meet the needs of TWH researchers are required to build capacity in D&I science theories, models, frameworks [21,22,23,24], designs, methods [25,26], and pragmatic measures [27,28] for conducting rigorous D&I studies.The purpose of this paper is to draw upon the D&I literature to provide TWH researchers and practitioners with: (1) an overview of D&I science; (2) a plain language explanation of key concepts in D&I; (3) a case study example of moving a TWH intervention down the reseach-to-practice pipeline; and (4) future opportunites for D&I science in TWH. As more scientific knowledge has been generated to date about the methods that successfully promote implementation as compared to those that advance dissemination, the approaches discussed in this article are most relevant to the successful implementation of TWH programs. This paper fills several gaps in the TWH literature by providing an accessible overview of D&I science through a synthesis of key literature and proposing ideas for leveraging D&I to advance TWH in U.S. and global workpalces.Across many fields, the science of the systematic implementation of evidence-based interventions—broadly defined as the “7 Ps”: programs, practices, principles, procedures, products, pills, and policies [25]—lags behind the science of developing the interventions themselves [29]. Many overlapping factors related to the characteristics of interventions (e.g., high cost or poor fit with stakeholder needs), the settings where programs are implemented (e.g., resource constraints, lack of regulatory/policy support), and the characteristics of recipients (e.g., lack of buy-in for the program) lead to limited uptake, implementation, and sustained use of these programs [30,31]. In OSH specifically, this research-to-practice lag has substantial implications for the health, safety, and well-being of the global workforce [20].D&I science is a growing field of study that examines the processes by which scientific evidence is adopted, implemented, and sustained in community or clinical settings [13,30]. Although a relatively new and transdisciplinary field of study, D&I has a strong historical foundation [30,31]. The field is concerned with changing systems by understanding context, leveraging an established evidence base, documenting outcomes, and characterizing the underlying mechanisms of change so that positive results can be replicated in other community-based and especially low-resource settings [30,32].Other key characteristics and implications of D&I science, adapted from Glasgow and Chambers [33], with tailored considerations for TWH researchers are presented in Table 1.Another useful concept from Brownson and colleagues [34] is referred to as designing for dissemination, implementation, and sustainment (D4DIS). D4DIS is a process to ensure that the products of research (such as new technologies, training, and health communication messages) are developed with the needs, resources, and time frames of the target audience in mind. It is believed that these efforts will increase the dissemination, implementation, and sustainment potential of programs in real-world settings. Practical tools exist for helping researchers to plan D4DIS and engage stakeholders in these efforts (see, for example, [35] and https://dicemethods.com/tool, accessed on 13 October 2021).In summary, D&I science approaches can be used to systematically address the research-to-practice lag by actively engaging stakeholders in the design of interventions that fit the changing context and needs of end users—such as the economic climate when implementing a TWH program [36]—and examining the processes by which these interventions are adopted, implemented, and sustained in workplaces [30].The proliferation of terminology to describe D&I activities has been explored extensively in the literature [37]. In the international OSH field, the terms knowledge translation, knowledge transfer, and knowledge transfer and exchange have been used [19,37,38,39,40,41] to describe similar or synonomous processes. At NIOSH [20,42], and in the context of TWH [11], the term translation(al) research is used, and while overlapping with the term D&I, it has important differences, as described below.Translational science has been defined as, “the field of investigation which seeks to understand the scientific and operational principles underlying each step of the translational process” [43] (p. 456). As Fort et al. [44] state succinctly, translational research takes scientific discoveries “from the bench to the bedside and back again”. Or perhaps more appropriately for OSH, from the lab to the field (i.e., the worksite/workplace) and back again. Whereas traditional scientific inquiry is primarily concerned with creating new knowledge, translational science is ultimately focused on the process of applying existing evidence to address health-related problems to generate generalizable knowledge [43,45].Translational science has been conceptualized as crossing all translational or “T” phases of the research continuum, from scientific discovery (T0), to efficacy (T1), effectiveness (T2), D&I (T3), and the outcomes and effectiveness of research in populations (T4) [44,46]. However, in practice, translational science has largely focused on barriers to intervention development at the efficacy and effectiveness stages (T1 and T2), while D&I science has focused on barriers to intervention adoption, use and sustainment (T3 and T4) [45]. More and better integration of D&I science across the translational continuum has therefore been called for [45,47].Increasingly, OSH researchers in the United States are adopting the terminology of mainstream D&I science [19,42]. More work is needed to harmonize terminology so that research in this area can be characterized and synthesized to enhance the impact and understanding of it.The following sections use the Curran [15] tool to expand on important concepts in the D&I field for TWH researchers to consider when planning/conducting D&I studies.Research in D&I is related to, but distinct from, traditional efficacy and effectiveness research. Referring to the Curran definition [15], efficacy research evaluates the initial impact of an intervention (“the thing”) when it is delivered under optimal, controlled conditions. Effectiveness research looks at whether “the thing” works, determining the impact of an intervention with demonstrated efficacy to obtain more externally valid (generalizable) results [37]. As Anger and colleagues note in their systematic review of TWH interventions [3], the first step in identifying programs to disseminate and implement is to establish their effectiveness. As stated previously, while efficacy and effectiveness research are concerned with investigating specific interventions and health or safety outcomes in either ideal or real-world settings, D&I research is particularly concerned with the adoption, successful implementation, and sustainability of the intervention [48]. However, it is important to mention the integration study designs at the intersection of effectiveness and implementation research. These are referred to as hybrid effectiveness–implementation designs [49]. They exist on a spectrum with sub-types depending on the relative emphasis on effectiveness and/or implementation, with a primary focus on effectiveness regarded as type 1, equal attention to effectiveness and implementation referred to as type 2, and primary emphasis on implementation regarded as type 3 [49,50,51]. These hybrid designs relate to the aforementioned “designing for dissemination and sustainability” [34] and better integration of D&I science across the translational continuum [45,47]. This integration is important because if efficacy and effectiveness studies focus only on achieving the maximum effect, this will likely and unintentionally limit their application in real-world settings due to issues of cost, burden, poor fit with local context, lack of buy-in and available expertise for program implementation.Once the “thing” (i.e., the intervention) is developed and tested, the focus of D&I efforts is on “doing the thing” well, which entails the selection of an appropriate theory, model, or framework (TMF) to guide the research. While the terms “theory”, “model”, and “framework” have distinct meanings, they are often used interchangeably [52]. TMFs generally describe tools to plan, evaluate, or understand barriers and facilitators (known as determinants) to D&I processes [16,53,54]. These tools help researchers on the front-end to plan, organize, and understand D&I phenomena, and on the back-end to understand why/how D&I strategies succeed or fail [24]. Table 2 highlights select D&I TMFs such as the Consolidated Framework for Implementation Research (CFIR) [55], the RE-AIM (Reach, Effectiveness, Adoption, Implementation, Maintenance) framework [8,9], the EPIS (Exploration, Planning, Implementation, Sustainment) framework [56], the diffusion of innovations theory [31], and the Theoretical Domains Framework (TDF) [57]. The CDC’s Knowledge to Action Framework [58] is another example of a tool developed for use by public health researchers to explore how evidence-based interventions can be translated into effective programs, policies, and practices. More than 150 D&I TMFs have been identified in the literature [21,22,23], and TWH researchers may wonder how to select an appropriate one. Fortunately, several efforts have focused on collecting and synthesizing the proliferation of TMFs available for D&I research. A key, publicly available resource, the D&I Models in Health (Table 2), provides an interactive webtool for study planning, combining and adapting TMFs, and selecting measurement tools to assess important D&I constructs [24].While there have been a large number of D&I TMFs proposed, leading some scholars to perceive the field as a “Tower of Babel” [65], there are more commonalities than differences across TMFs on factors known to affect the adoption, implementation, and sustainment of evidence-based interventions [66]. Key constructs commonly considered include multilevel contexts (i.e., inner and outer contexts with multiple layers within each, as described above) [8,9,56,61], characteristics of recipients at multiple levels [9,55], intervention and implementation strategy characteristics, and considerations for the various stages or phases of D&I (e.g., exploration, preparation, reach, adoption, implementation, sustainment, and important outcomes) [8,9,56,61].An example of the use of D&I models in OSH is from Tinc et al. [67], who employed the CFIR [55] to evaluate, among key stakeholders, the success of implementing a national program to prevent tractor rollover deaths in the United States with the use of a rollover protection structure (ROPS). The study used stakeholder surveys to assess short- and long-term outcome measures (intakes, funding progress, and tractor retrofits) and identify which CFIR components correlated with these outcomes. Results indicated that eight CFIR survey items reflecting four constructs—access to knowledge and information, leadership engagement, engaging (in fundraising and funding requests), and reflecting and evaluating—were highly correlated with at least one of the outcomes. In the TWH literature, Nobgrea and colleagues [36] used the RE-AIM framework [8,9] to iteratively develop and evaluate a toolkit to enable workplace safety and health practitioners to implement their own participatory TWH programs. An example from Europe [68] involves a process of using a D&I framework to systematically assess adaptations to an occupational health intervention. While these studies describe promising applications of D&I in OSH/TWH, more work is needed to generate generalizable knowledge about “how best to do the thing” in diverse types of workplaces with multilevel stakeholders, such as workers, employers and supervisors, union representatives, and regulators/policy makers.Another critical aspect of “doing the thing” is selecting the correct methods, designs, and measures. D&I science methods are varied, include both qualitative and quantitative techniques [25,26], and increasingly focus on the use of pragmatic, participatory, and mixed-method approaches [69]. Hybrid effectiveness–implementation designs [49,50,51] mentioned previously, promote the examination of both effectiveness and implementation outcomes within the same study to speed-up the research-to-practice process. Brown and colleagues [25] compiled a useful compendium of other D&I study designs. These include within-site designs to evaluate the success of intervention implementation inside a workplace or community, between-site designs that compare implementation processes among sites having different exposure conditions, within- and between-site comparisons with rollout designs where intervention start times are staggered, and factorial designs to examine multiple implementation strategies.Examples of D&I measures that can be used for “doing the thing” that are reliable, have validity data, and are considered to be pragmatic [28] are presented in Table 2, as are resources for identifying commonly used D&I tools, instruments and measures. In the D&I field, using previously developed measures is encouraged so that intervention findings may be compared across studies [64]. While it is not possible to characterize the full extent of available and appropriate D&I methods, designs, and measures, the above-mentioned resources are a starting point for TWH investigators to learn more about what considerations are needed when conducting rigorous D&I studies. It is also important to note more recent calls to integrate a health equity perspective across and within D&I TMFs, methods, and measures, whether health equity is a central focus of the D&I study or not [70].Referring again to the Curran’s [15] plain language explanation, implementation strategies are “the stuff” researchers do to try to help people (e.g., workers and employers)/(work)places “do the thing” as designed/intended. Dissemination and implementation strategies are a collection of methods or techniques to enhance the adoption, implementation, and sustainability of an evidence-based intervention [71,72]. In other words, D&I strategies are activities, approaches, and processes used to spread/deliver interventions to target populations and/or integrate interventions in target settings [73,74,75]. Key strategies have been classified in the Expert Recommendations for Implementing Change (ERIC) compilation, which identified 73 distinct strategies across nine domains: (1) use evaluative and iterative strategies; (2) provide interactive assistance; (3) adapt and tailor context; (4) develop stakeholder interrelationships; (5) train and educate stakeholders; (6) support clinicians; (7) engage consumers; (8) utilize financial strategies; and (9) change infrastructure [76]. The ERIC strategies have been adapted for school settings [77] and may be modified for use in workplaces. Determining which implementation strategies (or bundles of strategies) are the most appropriate is context dependent. Research to identify effective implementation strategies to “do the thing” within workplace settings is nascent [78], as are efforts to capture implementation outcomes, as is described in the next section.Implementation outcomes reflect how much and how well intervention implementers “do the thing” [15]. It is important to note that implementation outcomes, described in Table 3, are distinct from multilevel effectiveness outcomes which are often assessed in TWH studies. These effectiveness outcomes may include well-being, physical and mental health, occupational injuries, illness and fatalities, work-related fatigue, work stress, job performance, job satisfaction, safety climate, work–life balance [3] and occupational health equity [79].As described previously, studies in D&I are typically conducted after program efficacy and effectiveness are demonstrated and an evidence base has been established. Building on the effectiveness evidence base, D&I science outcomes focus on for whom and in what contexts the intervention works [14,80]. A focus on health equity and pragmatic research methods and measures is also important [25,27,28], as discussed previously in this paper. Furthermore, research on mechanisms of change/action that describe the process by which implementation strategies bring about specified implementation outcomes (i.e., mediational analyses) is recommended [81]. Without understanding how implementation strategies work, they will likely fail to achieve a positive impact [82].As stated, although not included in the Curran [15] conceptualization, another key feature of D&I is the importance of context, which has been defined as the unique circumstances within which intervention implementation is embedded [83]. Context is dynamic, multilevel, and cuts across economic, political, social, and temporal domains [17,84,85]. Factors at the system, organizational, worksite, and individual levels can serve as facilitators or barriers to implementation [85]. Characteristics of the intervention itself [31,86], as well as the intervention–context fit, can also have an impact on implementation outcomes, and all of these contextual influences may be present/active at different stages of the implementation process [75].Despite its importance, context is one of the least often reported elements in research [17,85]. A recent scoping review of 17 determinant frameworks in implementation science indicates that most frameworks provide only a limited description and definition of context, and there is inconsistency with regard to which contextual determinants are addressed [16]. As May and colleagues [87] note, context “is a problem” as many efficacy/effectiveness studies and even some D&I research designs try to “control out” contextual confounders, even though these represent the real-world conditions into which interventions must be integrated. Traditional randomized controlled trials do not typically answer the question about why or how intervention impact varies by setting, focusing instead on questions related to internal validity [33]. Despite the challenges and tensions noted, gaining an understanding of context is critical for determining program/policy outcomes [33], including in TWH studies.While researchers have demonstrated the value of considering multilevel contexts and stakeholder engagement in designing TWH interventions [2,11,36,88], a systematic review of TWH studies by Feltner and colleagues [18] identified only a limited number of interventions focusing on multilevel contextual factors, such as work organization and union membership status, health insurance status, access to primary care services, management support, availability of resources, and employee stress or strain related to company downsizing. No studies were identified that systematically assessed possible variation in intervention effectiveness by individual, worksite, organizational, or community factors [10]. Given the variety of contexts in which the TWH model can be implemented and studied—with variations in employers, work environments, and workers—understanding the factors that influence the effectiveness of integrated interventions is important, and more research is needed in this area [10].Moreover, it has been suggested that TWH studies should include a focus on complex systems approaches [89] to gain a deeper understanding of the multilevel influences within the TWH framework acting on intervention outcomes. Understanding context as a “process” rather than a “place” (i.e., the physical environment in which a practice is embedded) acknowledges that the setting in which implementation occurs is the product of “continuous accomplishments” and requires constant negotiation and iteration [87] (p. 4). D&I approaches can help address some of the complex, non-linear systems issues that impact TWH and OSH more generally.To illustrate the D&I process for TWH as described throughout this paper, a logic model is presented in Figure 1. Referring to the model, D&I outcomes (and ultimately work-related health, safety, and well-being outcomes) are conceptualized to be influenced by (a) the intervention (the evidence-based/informed program), (b) the D&I strategies, (c) mechanisms/mediators, or the “how and why” an implementation strategy operates [81], and (d) adaptations to context [9,90], such as modifications that will need to be made iteratively to sustain the initiative over time (e.g., lower cost, different staff/expertise, staff attitudes/buy-in, and different settings). The logic model displays both proximal outcomes, such as adoption and fidelity [37,48], and more distal outcomes and impacts, such as reduced occupational morbidity and mortality, enhanced well-being, and occupational health equity.Despite calls for increasing the emphasis of D&I in TWH studies [3,11,18], the application of these approaches has been limited [10,18]. One reason may be due to a lack of clarity among TWH researchers on what D&I science is—and is not. Recent papers for the broader D&I community elaborate on this issue, addressing misconceptions about specific D&I frameworks [92]. The following section presents a few concepts potentially requring clarification for the TWH context, as garnered from the literature.D&I is not r2p (but there is overlap). At NIOSH, r2p (research to practice) is an approach to communicate and transfer NIOSH “knowledge, interventions, or technologies” to relevant stakeholders for use in workplaces to contribute to reducing and eliminating injuries, illness, and fatalities [93]. In short, the NIOSH r2p program focuses on the transfer of interventions into effective practice. In contrast, D&I—or part of what NIOSH currently refers to as translation research—is the systematic study of these efforts [19,94]. While there are areas of overlap, including the focus on engaging stakeholders, r2p and translation research should be considered as separate, albeit complementary, areas of focus and effort.D&I is not the same as program evaluation (but there is overlap). CDC defines program evaluation as the systematic collection of information on the activities, characteristics, and results of programs in a specific setting to inform local knowledge and practice [95]. Examples of program evaluation may include formative, process, or summative activities [96]. While the boundaries are unclear, an important distinction can be made between the science of improvement (program evaluation) versus the science of dissemination and implementation [96]. Whereas program evaluation might consider intervention outcomes (and even in different settings), the focus is on how to make the intervention itself (i.e., “the thing”) better. In contrast, D&I outcomes relate to how to integrate the intervention into a variety of settings with a focus on methods and measures and the impact of implementation strategies [25,80]. For example, whereas program evaluation may focus on organizational and behavioral outcomes of a worksite injury prevention program, D&I is focused on the outcomes of acceptability, integration, and sustainability of the specific program/policy/practice [9,37,48] in different workplaces and community settings.In summary, D&I science is concerned with dynamic, multilevel contextual factors related to the characteristics of the intervention, implementation strategies, individual, program provider, organizational (workplace), and policy/regulatory levels; pragmatism; and sustainability [47,86]. D&I processes should be considered as a set of complex, non-linear, and iterative accomplishments that are emergent and dynamic [87]. Given this multilevel and systems-level approach, D&I is well-aligned with the OSH field. As noted previously, D&I (also referred to as translation(al) research) has been integrated into strategic NIOSH initiatives [19,20,42]. The case study presented below provides an extended example of identifying an evidence-based, OSH program, adapting it for TWH, and moving it along the translational research continuum [20,97,98,99,100]. Such work sets the stage for future opportunities for D&I science.In the United States, young workers (aged 15–24 years) experience higher rates of job-related injury than adult workers (aged 25–44 years) [101]. During 2012–2018, an estimated 3.2 million nonfatal injuries to young workers were treated in hospital emergency departments [101]. Work-related injuries may be life-altering, and young people hurt at work may experience a “cumulative burden of morbidity” over their lifetimes [102]. Data from the U.S. Bureau of Labor Statistics, Census of Fatal Occupational Injuries, indicate that 2349 young workers died on the job during the 2011 through to 2017 period [103]. Although employers are required by law to provide basic safety training to all workers, the lack of quality safety training has been shown to be a contributor to occupational injury among young workers [104,105].To address this public health problem, NIOSH and its partners developed the classroom-based, “Youth @ Work: Talking Safety” curriculum [106,107,108] to provide youth with a foundation of OSH competencies before they enter the workforce [109]. Results from quasi-experimental and (on-going) randomized trials indicate the effectiveness of the NIOSH curriculum to have a positive impact on adolescent students’ work safety knowledge, attitudes, norms, self-efficacy, and intention to enact safety behaviors in the workplace [110,111].Building on the evidence base for the “Youth @ Work: Talking Safety” curriculum, Promoting U through Safety and Health (PUSH), an online training tool for young workers, was developed by researchers with the Oregon Healthy Workforce Center, a NIOSH Total Worker Health Center of Excellence [112]. PUSH expands “Talking Safety” to include TWH concepts in an online format (versus the classroom format of “Talking Safety”) and for delivery in a different context (city park departments versus middle schools and high schools). The adaptation was based on inputs gathered through a needs assessment conducted with young people aged 14–24 years employed as aquatics workers in a city parks and recreation program in Portland, Oregon [112,113,114]. This work allowed the project team to qualitatively assess the acceptability and fit of the proposed OSH and health promotion content for the target population and establish the feasibility of an online delivery model [112,113]. To assess intervention outcomes, an individual-level randomized controlled trial was conducted with 140 young aquatics workers [114]. Most intervention workers (compared to the control group) reported learning new information (95%), liking the training (59%), and indicated that they had adopted new healthy behaviors (63%). Furthermore, the parks program indicated that the online format was practical and easy to administer.In a subsequent project phase, the PUSH training was scaled out (or disseminated in D&I terms) to young workers in a range of occupations, including cashiers, accountants, service managers, counselors, and lifeguards [115].Referring to the translational research pipeline described earlier in this paper [43,44,45,46,97,98,99,100], the “Youth @ Work: Talking Safety” curriculum was developed and tested in T0 and T1, with further effectiveness testing conducted in T2. The evidence base established for the curriculum was leveraged to develop PUSH with a TWH focus and tested it with other target groups in different contexts, focusing largely on barriers to intervention development at the efficacy and effectiveness stages (T1 and T2). Future D&I studies (T3) involving PUSH could, for example, focus on the testing of theories, models, and frameworks for D&I processes, some of which have been described in this article [21,22,23,24]. Research could be conducted to systematically identify, develop, test, evaluate, and/or refine strategies [73,76] to disseminate and implement PUSH interventions in new community/workplace contexts where young workers are employed. Another example of future opportunities for D&I research would include a systematic investigation of the local adaptations [90,116,117] of PUSH in the context of its implementation within various settings that employ young workers. Studies of influences on the development, packaging, transmission, and reception [118] of PUSH in various settings and contexts where young workers can be reached would also help to move this promising TWH research into (sustained) practice.The aim of D&I is to gain an understanding of the contextual factors—including the needs and priorities of stakeholders at each level [119]—that facilitate and hinder the successful uptake of evidence-based interventions. Although D&I science holds promise for speeding up the translation of TWH interventions to enhance the health and well-being of the global workforce, research in this area remains limited. One barrier to the widespread uptake of D&I may be the need to identify interventions that are ready for translation. As Anger and colleagues [3] note, “Perhaps it is premature to press for dissemination of the TWH programs until their effectiveness is better established.” (p. 243). However, waiting for interventions to meet evidence standards may contribute to the translational lag time while also generating interventions that are not replicable in the real world [84,120]. Types of evidence more typical in OSH than that generated through highly controlled trials include guidelines, recommendations, and observational as well as worker case studies [121]. Communicating which TWH initiatives are promising (i.e., evidence-informed versus evidence-based [37,122]) may speed-up practical and empirical outcomes.To ensure that TWH interventions are replicable in real-world settings, efficacy and effectiveness studies should be designed and conducted with an eye toward feasibility, generalization, dissemination, equity, and sustainability at every phase of the research continuum [9,45]. The concept of designing for dissemination, implementation, and sustainment (D4DIS), mentioned previously, is a process that can be used to ensure that the products of research are developed with the needs, resources, and time frames of the target audience in mind [34]. Practical tools exist to help researchers plan D4DIS and engage stakeholders in these efforts (see, for example, DICEmethods.org). Moreover, D&I science TMFs such as RE-AIM [8,9,92], CFIR [55], and EPIS [56,61] can be used not only for evaluating interventions but can also guide their planning with stakeholders (e.g., workers, employers, and community members), be used iteratively during implementation to adapt the program/practice to better fit the context (e.g., local workplaces), and respond to emerging implementation data. More and better integration of D&I across the translational science continuum has been called for to move research more rapidly into practice to benefit the public [45,47], including workers [42]. Moreover, by integrating D&I early in the research process, activities related to getting scientific innovations into the hands of end users may not be viewed as a burdensome, add-on activity [42], as someone else’s role [34,123,124], or as up to the stakeholders to figure out [125].Consistency in how terminology is applied will also help to facilitate the uptake of D&I science in TWH and to develop generalizable knowledge about D&I outcomes. Rabin and colleagues [37] have assembled extensive glossaries to capture and harmonize the D&I “Tower of Babel” [65]. A project being conducted under the NIOSH Evaluation Capacity-Building initiative [126] is to develop, with D&I science scholars and internal and external stakeholders, a glossary of D&I research terms for OSH researchers. As with any developing field, there are minor differences in terminology and a relative emphasis on different theories, models and frameworks. However, within D&I, there is general agreement on key principles, factors, and methods (see Table 1; also [66]).Another challenge may be limited D&I expertise in the TWH field, as is the case more generally [127]. Furthermore, although the broader D&I science field has expanded and diversified over the years, more investment in resources tailored to meet the needs of OSH and TWH researchers are required to build capacity with D&I science models, theories, frameworks [21,22,23,24], methods and designs [25,26], and pragmatic measures [27,28]. These tools can be used to conduct rigorous studies that bridge the gap between the lab, the field, and public health impact.Another important area of focus for D&I in TWH is how to critically infuse an equity approach into all research endeavors [79,128,129,130]. This includes focusing on equitable reach from the beginning of the intervention planning and development; designing and selecting programs for populations at disproportionate risk for work-related injury, illness and reduced well-being; implementing what works; and developing implementation strategies that can help reduce inequities in OSH [129]. Practical issues—such as the complexity of accessing many workplaces and workers, especially those in smaller and low-resource businesses and who may experience multiple OSH inequities [129,131]—make conducting D&I research challenging [42]. However, D&I research in TWH can help to build the necessary evidence base for employers to adopt a more holistic, feasible, and sustainable approach to promote (current and future) employee health and well-being.Finally, D&I approaches that consider multilevel contextual factors in TWH that influence intervention outcomes [18] are needed to address the dynamic, complex and emergent nature of twenty-first-century challenges in OSH [6,7,90], including global pandemics. TWH research is needed that systematically considers the factors at the system, organizational, worksite, and individual level that serve as facilitators or obstacles to implementation [31,75]. Designing implementation strategies to address contextual barriers [75] may help TWH researchers to consider multilevel influences (such as regulatory changes, or individual worker acceptance of new technologies) as well as other complex phenomena that influence the effectiveness, equity, adoption, implementation and sustainability of TWH programs, both in the United States and internationally [130].Applications of D&I hold promise for addressing the limited movement of intergrated worker protection and health promotion interventions into widespread and sustained practice. Several reasons, including the need to identify promising TWH interventions that are ready to be moved along the research to practice continuum and limited D&I capacity, have been identified that may be potential barriers to the uptake of D&I approaches. However, numerous opportunities also exist. This paper drew upon a synthesis of the D&I science literature to provide TWH researchers and practitioners with an overview of the D&I field and a plain language explanation of key concepts. This analysis also presents D&I examples and resources, and discusses how D&I tools can be used to more rapidly deploy effective TWH programs. The end goal is to improve the current and future safety, health and well-being of working people and enhance occupational health equity in an increasingly dynamic and complex global economy.Conceptualization, R.J.G. and S.M.H.; Methodology, R.J.G., S.M.H., B.A.R. and R.E.G.; Resources, R.J.G., S.M.H., B.A.R., D.S.R., T.R.C., M.R.T., M.P. and R.E.G.; Writing—Original Draft Preparation, R.J.G., S.M.H., B.A.R., D.S.R., T.R.C., M.R.T., M.P. and R.E.G.; Writing—Review and Editing, R.J.G., S.M.H., B.A.R., D.S.R. and R.E.G.; Project Administration, R.J.G. All authors have read and agreed to the published version of the manuscript.This research was primarily supported by internal CDC/NIOSH funding. The project was also funded by: the Oregon Healthy Workforce Center (NIOSH, grant number U19OH010154); the Healthier Workforce Center of the Midwest (NIOSH, grant number U19OH008868); and the Colorado Implementation Science Center (NIH, grant number P50CA244688).Not applicable.Informed consent was obtained from all subjects involved in the case study example.Not applicable.We wish to thank Marcia Ory and Amia Downes for their reviews of and thoughtful feedback on earlier drafts of this manuscript.D.S.R. has a significant financial interest in Northwest Education Training and Assessment, LLC, a company that may have a commercial interest in the results of this research and technology. This potential conflict of interest was reviewed, and a management plan approved by the OHSU and the University of Iowa Conflict of Interest in Research Committees was implemented.The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention.Logic model of a D&I study for TWH. Source: adapted from [91].Considerations for the use of D&I science approaches for rigorous, rapid, and relevant TWH initiatives (adapted from Glasgow and Chambers [12]).Select D&I theories, models, frameworks, tools, and resources.Examples of implementation outcomes adapted for TWH interventions.Sources: Adapted from Glasgow et al. and Proctor et al. [9,48]. For more details and comprehensive definitions see Rabin and Brownson [37].Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11051.txt

@@ -0,0 +1 @@
+The objective of this retrospective study was to identify predictors of angiographic hemostasis among patients with life-threatening traumatic oronasal bleeding (ONB) and determine the threshold for timely referral or intervention. The diagnosis of traumatic, life-threatening ONB was made if the patient suffered from craniofacial trauma presenting at triage with unstable hemodynamics or required a definitive airway due to ONB, without other major bleeding identified. There were 4404 craniofacial trauma patients between January 2015 and December 2019, of which 72 (1.6%) fulfilled the diagnosis of traumatic life-threatening ONB. Of these patients, 39 (54.2%) received trans-arterial embolization (TAE), 11 (15.3%) were treated with other methods, and 22 (30.5%) were excluded. Motor vehicle accidents were the most common cause of life-threatening ONB (52%), and the internal maxillary artery was the most commonly identified hemorrhaging artery requiring embolization (84%). Shock index (SI) was significantly higher in the angiographic hemostasis group (p < 0.001). The AUC-ROC was 0.87 (95% CI, 0.88–1.00) for SI to predict angiographic hemostasis. Early recognition and timely intervention are crucial in post-traumatic, life-threatening ONB management. Patients initially presenting with SI > 0.95 were more likely to receive TAE, with the TAE group having statistically higher SI than the non-TAE group whilst receiving significantly more packed red blood cells. Hence, for patients presenting with life-threatening traumatic ONB and a SI > 0.95, TAE should be considered if preliminary attempts at hemostasis have failed.Life-threatening oronasal bleeding (ONB) after craniofacial trauma is infrequent with an approximate incidence rate of 1%, but mortality rates can be as high as 85.9% if hemostasis is not achieved [1,2,3,4,5]. High rates of mortality in traumatic, life-threatening ONB remain a challenge for clinical physicians [1,3,6,7]. Timely and adequate intervention is critical in reducing hemorrhage-related mortality, while delayed recognition and subsequent management may lead to unfavorable outcomes and dire consequences [1,3,8,9,10]. For patients presenting with traumatic ONB, clinical management follows Advanced Trauma Life Support (ATLS) protocols along with airway protection, vital sign stabilization, bleeder identification, and hemostasis. The current consensus is to perform conservative treatment first (consisting of fluid resuscitation, blood transfusion, and anterior and posterior nasal packing) followed by trans-arterial embolization (TAE). However, the association of TAE with several major complications such as cerebrovascular accident, nephrotoxicity, facial nerve palsy, monocular blindness, and soft tissue necrosis have been previously reported in the literature [11,12,13]. Furthermore, TAE requires adequately trained interventional radiologists with proper facilities and equipment, both of which may not be readily available at all institutions. Based on the availability of trained personnel and facilities, treatment may potentially be delayed until the proper staff and facilities are secured. As a delay in intervention increases the complication rate, prompt recognition of patients requiring TAE is crucial. This study aims to identify possible predictors for patients requiring angiographic hemostasis, allowing first-line physicians to respond accordingly.This study was conducted retrospectively and approved by the Institutional Review Board (#104-2789B). Patients with craniofacial trauma were recruited in a tertiary referral center from January 2015 to December 2019 (Figure 1). All patients were managed according to ATLS protocol. Diagnosis of life-threatening ONB was made if the patient was bleeding from the oronasal area with accompanying unstable hemodynamics, had systolic blood pressure (SBP) < 90 mmHg or heart rate (HR) > 100 beats per minute, or required a definitive airway establishment due to excessive hemorrhage. Patients were excluded if they died before TAE, underwent angiography without embolization, received embolization without evidence of contrast extravasation or pseudoaneurysm during angiography, or had incomplete medical records. Patients were included if they were managed by conservative treatment, consisting of fluid resuscitation, blood transfusion, anterior nasal packing with a gauze, and/or posterior nasal packing with 10–14 Fr Foley catheters with 10–30 mL balloons. TAE was indicated if there was persistent uncontrolled bleeding or unstable hemodynamics despite conservative management as assessed by the physician on site. Evidence of active contrast extravasation or pseudoaneurysm during angiography was also considered a positive finding and mandated embolization. All patients included in this study must have reached transient stability after initial management at the emergency department and did not experience persistent hemorrhage.We compared two group of patients. The first group consisted of patients who did not require TAE for hemostasis (non-TAE group) and the second group consisted of patients requiring TAE for hemostasis (TAE group). The medical records were independently collected by two physicians. The radiological images, findings, and procedures were reviewed by an experienced radiologist. Patient characteristics, vital signs, laboratory data, mechanism of injury, associated injury, patterns of facial fractures, amount of blood transfusion, TAE procedures and findings (if applicable), and clinical outcome were recorded. The Shock Index (SI) for each patient was also calculated from the vital signs measured after completion of anterior and posterior nasal packing (SI = HR/SBP). Data was compiled and analyzed using the SPSS package Version 20.0 for Windows. The data was summarized with nominal variables expressed with counts and percentages, while continuous variables were presented as mean ± SD. Variables were analyzed using Fisher’s exact test and Wilcoxon signed-rank test for nominal and continuous variables, respectively. The level of significance was set at p < 0.05. Receiver operating characteristic (ROC) curve was used to evaluate the performance of variables and to determine the cut-off value.There were 4404 craniofacial trauma patients with 72 fulfilling the criteria of life-threatening traumatic ONB (1.6%). Twenty-two patients were excluded due to incomplete data (n = 6), having undergone TAE without indication for angioembolization (n = 11), expiring before TAE could be performed (n = 1), and negative findings on angiography (n = 4) (Figure 1). The remaining fifty patients were included in this study. Motor vehicle accidents were the most common cause of life-threatening ONB (52%) with motorcycle accidents accounting for 68% of them. Patients were predominantly male (76%), in their 30s (age 35.1 ± 18.8), and suffered from blunt trauma to the facial region (98%). Eleven patients achieved hemostasis after nasal packing, while the other 39 patients only achieved hemostasis after TAE. All patients who required TAE for hemostasis were confirmed to have contrast extravasation or pseudoaneurysm formation of the external carotid artery origin, with the internal maxillary artery being the most commonly identified hemorrhaging artery (84%) (Figure 2 and Figure 3).Fracture patterns, mechanism of injury, associated injuries, GCS, and hematocrit did not reach statistical difference when comparing the two groups. SBP at triage in the TAE group was significantly lower than in the non-TAE group (p = 0.01). HR at triage in the TAE group was significantly higher than in the non-TAE group (p = 0.004) (Table 1). Hence, the average SI at presentation for the TAE group (1.31 ± 0.49) was much higher than that calculated for the non-TAE group (0.73 ± 0.17) (p < 0.05). After nasal packing, SI for the TAE group improved (1.14 ± 0.37) compared to baseline values, but the change was not statistically significant. SI for the non-TAE group did not change (0.74 ± 0.23). SI for the TAE group was assessed after TAE (0.9 ± 0.31) and showed a significant improvement from baseline values (p < 0.05). The amount of blood transfused with packed red blood cells (PRBC) at the emergency department was 6.61 ± 5.0 units in the TAE group and 2.9 ± 3.2 units in the non-TAE group. The AUC-ROC curve was 0.87 (95%, CI = 0.88–1.00) for SI > 0.95 to predict successful angiographic hemostasis (Figure 4).One patient had a palatal wound infection and required a local flap for reconstruction. Two patients had nasal-oral fistula formation requiring surgical intervention. Eight patients died of traumatic brain injury (TBI). Two patients died of delayed hemorrhage of non-oronasal origin, with one incidence of retroperitoneal hemorrhage and one incidence of pulmonary hemorrhage. No mortality was directly linked to ONB.Seven patients experienced rebleeding events after TAE, and three of these required secondary interventions to reach hemostasis (Table 2). Six out of the seven rebleeding events occurred within 3 days after initial TAE. There were no sequelae or major complications associated with angiography.The incidence of life-threatening ONB after craniofacial trauma was reported to be about 1% [1,2,3,4,5]. In our study, the incidence of life-threatening ONB after craniofacial trauma was 1.6%. In these patients with severe, life-threatening ONB, those who required intervention with TAE had significantly higher SI at presentation (1.14 ± 0.37) compared to those who did not require further intervention (0.74 ± 0.23). Although ONB is diagnosed clinically, delayed management could occur from underestimating the severity and amount of blood loss, which may lead to unfavorable outcomes. ONB should be regarded as a consequential source of bleeding in the presence of hypovolemic shock in major trauma patients despite its relatively low incidence, as mortality rate and development of serious complications positively correlate to delay in management [1,3,9]. Early recognition, timely referral, and prompt intervention are imperative in approaching ONB patients. To that end, according to our findings, shock index may be a promising early predictor of severe disease in oronasal trauma patients, as the parameters required for calculating shock index are readily available at triage. In terms of management, TAE has been proven to be an effective method in hemostasis and has replaced surgery as the first line invasive procedure in ONB [1,14,15,16,17,18]. In this study, TAE played a role in hemodynamical stabilization in ONB with SI having significantly improved after TAE. TAE is safe, more selective, less invasive, and allows access to bleeding sites otherwise inaccessible to surgery. Despite its obvious advantages over more conservative treatment, TAE requires both specially trained radiologists and appropriate equipment. Such resources may not be readily available in every hospital, especially small-scale primary medical institutions. This limitation accentuates the importance of recognizing the need of TAE in the context of ONB. To the best of our knowledge, no parameter has been identified to be an effective predictor for the necessity of TAE in ONB. Hence, in order to help primary institutions, where resources for TAE may not be available, provide the patient with timely referral and arrange appropriate intervention as needed, this retrospective study was designed, which identifies possible predictors for the necessity of TAE in life-threatening traumatic ONB.Motor vehicle accidents are the main cause of life-threatening ONB worldwide [3,14]. The concept of “golden time” emphasizes the importance of urgent care in the management of trauma patients. Effective field triage is essential in reducing transport time. If the injury exceeds the capability of the hospital providing first-aid, patients should be transferred to a higher-level facility promptly, as the availability of care at a trauma center is shown to directly affect patient mortality [19]. In the context of life-threatening ONB, prudent decision-making for referral to another institution may be a matter of life or death because interventional radiologists may not be readily available at every medical facility. Vital signs, heart rate, respiratory rate, blood pressure, and body temperature are fundamental measurements and are easily accessible by emergency medical responders (EMS) or medical facilities with minimal equipment. SI was suggested as an accurate tool for identifying early hypovolemic shock, severity of illness, and prognosis, especially in vascular injuries, hemorrhagic events, or trauma patients [20,21,22,23,24,25]. SI was also found to be an independent risk factor and a fast guide for the need of massive transfusion after trauma [25,26,27], and increased blood transfusion is regarded as an independent risk factor correlating with poor prognosis in trauma patients because of the alteration of cytokine levels and inflammatory processes caused by the transfused blood products [20,28]. Jonas et al. concluded that an SI of greater than 0.9 predicts the necessity of intervention for hemostasis with high specificity (93.6%), and by lowering the threshold to ≥0.8, sensitivity is increased to 76.1% [22]. Nakasone et al. demonstrated that SI correlates with extravasation during gastrointestinal hemorrhage via multivariate logistic regression analysis [29]. Kuo et al. reported similar results in patients with pelvic fracture, whereupon relative hypotension is an indicator for TAE regardless of negative contrast extravasation [30]. In our study, there was a significant difference between the TAE group and non-TAE group in the initial SI at presentation, with a mean value of 1.14 and 0.74, respectively (p < 0.05). According to the ROC curve, the cut-off value of SI in predicting the necessity of TAE was 0.95 with AUC of 0.8 (Figure 4). Despite modest sensitivity (74%), the high positive predictive value (100%) and specificity (100%) warrant further evaluation or intervention. With a predictor for the necessity of TAE, physicians could provide timely intervention to prevent possible mortality and morbidity in the event of ONB. In our study, 15 patients who had negative findings for angiography had a SI of less than 0.95. Therefore, we would recommend that if a patient is diagnosed with post-traumatic life-threatening ONB with an initial SI of greater than 0.95 following unsuccessful nasal packing, TAE should be initiated as soon as possible to achieve hemostasis, effectively decreasing the amount of blood transfusion required and associated morbidity. We have devised a possible algorithm for the management of life-threatening oronasal bleeding after craniofacial trauma accordingly (Figure 5).The introduction of damage control resuscitation focuses on hemorrhage control, maintaining hemodynamic stability, and correcting physiological derangements such as coagulopathy, acidosis, and hypothermia to improve patient survival [31]. The development of coagulopathy after trauma has a profound impact on survival and has been identified as an independent predictor of mortality [32,33]. With advancements in the understanding of trauma-induced coagulopathy (TIC) and evolution in resuscitation protocols, there is potential for improvement in prognosis of trauma patients [34]. In trauma patients, post-traumatic hypoperfusion and related tissue injury activate the response cascade of the endothelium, platelets, and the immune system, consequently disrupting the balance of coagulation [35]. Devastating results could occur if such a response is not promptly managed. TIC could be recognized by standard coagulation panels, thromboelastography (TEG), or in combination [33,36]. In this study, the TAE group was found to have significant prolonged INR and thrombocytopenia compared to the non-TAE group. This finding infers that there were more severe tissue injuries in the TAE group, and in turn, TIC. Unfortunately, an effective scoring system for predicting TIC is lacking, and the diverse phenotypes of TIC pose a challenge in making the diagnosis [35,36]. Hence, hemostasis should be the fundamental principal in management, and resuscitation should be tailored individually [37].In this study, no patients reported a history of congenital coagulopathy, while only one patient was receiving anti-platelet medication due to a history of cerebrovascular accident. During this study, TEG was not widely adopted clinically, so conventional coagulation panels were the only available tools for evaluating coagulopathy. Of the fifty patients included in this study, fifteen patients had prolonged prothrombin time (PT) and thrombocytopenia in various severities. Yet, among these patients, only two patients were clinically recognized as having significant coagulopathy and requiring blood products for correction. As for the patients who met the inclusion criteria but had died from their injuries, TBI was determined to be the cause of death as opposed to coagulopathy. Nevertheless, the association between TIC and TBI has been reported in the literature and the relationship between the two is still evolving [38,39,40]. To this day, TIC remains a challenge in life-threatening hemorrhage, and further studies are warranted for better clinical practice in the context of oronasal bleeding.In our series, LeFort fractures are associated with greater severity compared to other types of orofacial fracture and account for 42% of all reported fractures. Previous studies have demonstrated that there is a significant correlation between life-threatening ONB and LeFort II or III fractures [1,4]. While Shimoyama et al. reported five patients with LeFort II or III fractures experiencing ONB who had undergone TAE [5], in our study, there was no statistical difference between fracture patterns to recommend assessing the necessity of TAE based on fracture type. The most common bleeder identified in life-threatening ONB is the internal maxillary artery and its branches at the Woodruff plexus. Local compression is rarely effective in this area due to difficulty of access and failure in hematoma formation. Anterior and posterior nasal packing is required to provide adequate tamponade effect. Shimoyama et al. described five patients who achieved hemostasis and temporary reduction by nasal packing after a mid-facial fracture with massive oral bleeding [5]. Soyka et al. reported that successful hemostasis in epistatic patients with a posterior source of nasal bleeding could be achieved by nasal packing in 62% of cases [41], whereas Cogbill et al. found that primary hemostasis with nasal packing proved effective in only 29% of cases [3]. In our study, nasal packing provided successful hemostasis in 22% of patients. Limited effectiveness in nasal packing may be due to the compromise in the integrity of the nasopharyngeal wall secondary to comminuted fractures or multiple bleeders [10]. Though the rate of efficacious hemostasis is suboptimal, nasal packing could be performed with readily available equipment and relatively low risks. This would allow makeshift stabilization of the patient during transferal, which is valuable in trauma management.Rebleeding after seemingly successful hemostasis following TAE in life-threatening ONB has not been thoroughly discussed. Lee et al. reported that more severe injury, multiple contrast media extravasations during angiography, incomplete TAE, and elevated HR were associated with rebleeding after TAE in blunt liver injury [42]. Chen et al. found that rich collaterals to the bleeding site and a distance of greater than 5 cm between the catheter tip and the bleeding point were significantly related to rebleeding after TAE treatment in the setting of gastrointestinal bleeding [43]. There were seven patients who experienced rebleeding events in this study (Table 2). Six of the seven patients experienced rebleeding within 3 days, while rebleeding was found in the last patient 12 days after initial TAE. Three patients underwent secondary intervention to stop the bleeding. No major consequences or mortality was noted. Application of gelatin foam for embolization tends to have fewer rebleeding events compared to other materials (p = 0.05). Gelatin foam has been extensively used in TAE for temporary obstruction, allowing distal embolization by small-size particles. Thus, it is effective in bleeders with rich collaterals. Despite infection or ischemia due to gaseous emboli or erroneous embolization of unintended vessels [44,45], there were no TAE-related complications caused by gelatin foam application in this study.Dubel et al. reviewed the complication rate for treating epistaxis with TAE, ranging from 0% to 11% with an average major complication rate of 2.5% [46]. With adequate and timely hemostasis, ONB is unlikely to instigate undesirable consequences. In this study, there were no ONB-related mortalities or TAE-related complications. Caution should still be taken for possible complications, including possible iatrogenic cerebrovascular accidents, soft tissue necrosis, ischemic pain, and facial nerve injury.There are some limitations to the current study. First, the study was conducted in a retrospective manner with a relatively small sample size, resulting in recall bias and a decrease of the statistical power. Second, there were limited cases with penetrating injuries included in this study, suggesting that the result might not apply to patients with similar trauma mechanisms. There was only one patient with penetrating injuries included in this study. Further studies may be warranted to better justify the application of SI in the assessment of patients presenting with penetrating injuries and the management of life-threatening traumatic ONB in such instances. Finally, the lack of consensus in the literature on the definition of post-traumatic life-threatening ONB results in difficulties during the application of our findings in comparative circumstances.Early recognition and well-timed intervention are crucial in the management of post-traumatic life-threatening ONB. We demonstrated that TAE and both anterior and posterior nasal packing can be effective treatments for life-threatening ONB among patients with severe craniofacial trauma. Patients initially presenting with SI > 0.95 were more likely to receive TAE and required a significantly greater number of packed red blood cells transfused. Hence, for patients presenting with life-threatening traumatic ONB and a SI > 0.95, TAE should be considered after failed preliminary attempts at nasal packing. When an appropriate, timely intervention is applied, the outcomes for patients who receive angiographic hemostasis are not inferior to those of less severe patients who achieve hemostasis without embolization. Still, further research is needed before a more definitive relationship can be established.Conceptualization, C.-H.C. and Y.-C.W.; methodology, C.-H.C.; validation, F.-Y.H., S.-H.M. and A.D.-C.C.; formal analysis, F.-Y.H. and S.-H.M.; investigation, F.-Y.H.; data curation, F.-Y.H.; writing—original draft preparation, F.-Y.H.; writing—review and editing, C.-H.C.; supervision, C.-H.C. and Y.-C.W. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of Chang Gung Memorial Hospital (IRB No.: 104-2789B).Written informed consent has been obtained from the patient to publish this paper.The data presented in this study are available on request from the corresponding author. The data are not publicly available due to patient privacy.The authors declare no conflict of interest.The flowchart of patients who met inclusion/exclusion criteria of this study.A 55-year-old female who presented with massive oronasal bleeding after facial trauma. (a) Patient was intubated for airway protection. Arterial and posterior nasal packing was performed after intubation. (b) CT scan shows bilateral maxillary fractures.Angiogram of the patient shown in Figure 2. (a) Lateral angiogram of common carotid artery shows contrast extravasation from internal maxillary artery (arrow). (b) Post gelatin foam embolization images showing hemostasis (arrow).ROC curve of shock index and prediction of angioembolization. Cutoff value of SI, 0.95; sensitivity, 74%; specificity, 100%; PPV, 100%.An algorithm for the management of life-threatening oronasal bleeding after craniofacial trauma.Demographics of patients with post-traumatic life-threatening oronasal bleeding between angiographic hemostasis and non-angiographic hemostasis groups.MVA = motor vehicle accident; CNS = central nerve system; NOE = nasoorbitoethmoidal; ZMC = zygomaticomaxillary complex; SBP = systolic blood pressure; HR = heart rate; SI = shock index; GCS = Glasgow coma scale; pRBC = packet red blood cell; WB = whole blood; FFP = fresh frozen plasma; ISS = Injury Severity Score; LOS = Length of hospital stay. * = significant difference.Demographics of patients with traumatic life-threatening oronasal bleeding (ONB) requiring TAE with rebleeding and without rebleeding.NBCA = N-butyl cyanoacrylate; PVA = polyvinyl alcohol; IMA = internal mammary artery; ECA = external carotid artery; TAE = transarterial embolization; SI = shock index; ISS = injury severity score; LOS = length of hospital stay.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11052.txt

@@ -0,0 +1,2 @@
+The use of game-like elements is become increasingly popular in the context of fitness and health apps. While such “gamified” apps hold great potential in motivating people to improve their health, they also come with a “darker side”. Recent work suggests that these gamified health apps raise a number of ethical challenges that, if left unaddressed, are not only morally problematic but also have adverse effects on user health and engagement with the apps. However, studies highlighting the ethical challenges of gamification have also met with criticism, indicating that they fall short of providing guidance to practitioners. In avoiding this mistake, this paper seeks to advance the goal of facilitating a practice-relevant guide for designers of gamified health apps to address ethical issues raised by use of such apps. More specifically, the paper seeks to achieve two major aims: (a) to propose a revised practice-relevant theoretical framework that outlines the responsibilities of the designers of gamified health apps, and (b) to provide a landscape of the various ethical issues related to gamified health apps based on a systematic literature review of the empirical literature investigating adverse effects of such apps.Gamification can be generally defined as the use of techniques and elements of video game design in non-game contexts [1,2]. In the context of health tracking and wearable health devices, gamification can be and is being used to encourage health and wellness activity. In particular, wearable activity trackers, in conjunction with gamified smartphone apps, have been promoted as promising tools for increasing physical activity among their users [3]. Some examples of game-like elements used in gamified health apps include points and rewards for health activity as well as social elements like competitions and challenges with other people [1,4]. Gamification is often distinguished from more immersive, full-fledged, or “serious games”, and the intention in gamification is to mimic experiences reminiscent of games to affect the behavior and motivation of users [5]. In the context of health, gamification generally seeks to alter user behavior by increasing their physical activity and/or adopting a healthier lifestyle through game-like experiences.The use of such game-like elements, and gamification in general, is not unique to the case of health and fitness. Gamification techniques have found their application in a diverse range of areas including business organizations attempting to enhance customer engagement as well as employee performance, public policy initiatives, as well as in classrooms and other learning environments [6]. The increase in popularity of gamification in the past decade has been concurrent with the rise in accessibility of digital technologies, particularly smart phones as well as digital infrastructure that has created a networked world. Social networks, and other similar networked platforms, have also contributed to an increase in prevalence of gamification, as designers have leveraged such networks to improve interaction and engagement with users [5]. Despite its potential, gamification has also found its critics, including those who have questioned the moral and ethical legitimacy of gamification [2,7,8]. Kim & Werbach [2] have argued that such criticism suffers from painting with too broad a brush in denouncing almost all forms of gamification as vicious and/or exploitative. They also state that existing normative accounts of problems with gamification fall short of providing guidance to practitioners, particularly designers of gamified apps and platforms. Kim and Werbach have, instead, proposed a practice-relevant, context-sensitive and situated approach to the exploration of ethical issues associated with gamification. They have significantly enhanced the normative discussions about gamification, and to our knowledge, present the most comprehensive conceptual framework of ethical issues associated with gamification thus far available.Yet, in their methodology, Kim and Werbach [2] were primarily concerned with business practices, and this may reflect the shortcomings of the study when applied to other contexts, such as health and fitness tracking. Further, while some recent studies have pointed out the “darker side” of health gamification, there is currently a lack of systematic reflection and compilation of such issues [9]. Recent studies also suggest that the negative effects of gamified health apps can have adverse effects on users motivations and lead to discontinuance of app use [9,10]. A thorough landscape of the ethical issues of gamified health apps could not only help designers carry out their potential moral duties towards the users of apps, but also lead to better long-term user engagement with their products. This paper seeks to advance the goal of facilitating a practice-relevant guide for designers of gamified health apps to address ethical issues raised by the use of such apps. More specifically, the paper seeks to achieve two major aims: (a) to propose a revised practice-relevant theoretical framework that outlines the responsibilities of the designers of gamified health apps, and (b) to provide a landscape of the various ethical issues related to gamified health apps as found in the empirical literature about such apps.To achieve these objectives, we first conduct a theoretical analysis of the conceptual framework of ethical issues in gamification provided by Kim and Werbach [2]. The aim of this analysis is to propose amendments and refine the theoretical framework, in order to be useful particularly for the designers of gamified health apps. To this end, we created a tripartite framework based on the types of responsibilities designers of such apps may have. This tripartite theoretical framework can facilitate the taxonomizing of various ethical issues related to gamified health apps, based on how designers may address such issues. We then conducted a systematic literature review to investigate empirically supported ethical issues related to gamification in health and fitness tracking. The results of this review serve as a guiding list of ethical issues likely to be encountered in the design and use of a gamified health app. Such a review also allows us to explore the strength of our revised framework by investigating whether the kinds of responsibilities identified by our framework can address such issues. Finally, based on our analysis, we posit some limitations of this framework and offer suggestions for future studies that aim to locate further ethical issues related to gamified health apps or to test how such ethical issues actualize in specific circumstances or with particular gamified health apps. Our analysis also offers ways for users and, in particular, designers of such apps to navigate through, anticipate and avoid potential ethical issues related to gamified health apps.The field of ethics is a branch within philosophy that deals with systemizing, defending and recommending concepts of right and wrong conduct [11]. As a response to social and technological developments, the twentieth century saw the development of an ethics of technology as a subdiscipline within ethics, which involved investigations specifically concerning the role of technology [12]. Gamification ethics, or the ethics of gamification, can be intellectually located within this tradition of the ethics of technology as part of a further sub-discipline—applied ethics—which seeks to apply theories, normative standards, concepts, and methods developed within ethics (or moral philosophy) to, for example, inquiries concerning specific technologies. Seeking to explore ethical issues raised by gamification, gamification ethics is a recent, yet rapidly developing topic [13].Kim and Werbach [2] contend that prior to their work, gamification ethics displayed a tendency to over-generalize from particular examples and under-theorize, partly owing to the speed with which technologies associated with gamification advanced. To cover these gaps, they propose a “conceptual map of the terrain” that can offer normative guidance to gamification scholars as well as practitioners in identifying underlying structures that tie together what may seem like disjointed and disparate phenomena related to gamification. They share an aim with this paper: developing a framework that can be useful to the designers (and practitioners) of gamification. To this end, they propose four broad categories of ethical difficulties with gamification which encapsulate a cluster of concerns. In this section, we discuss these four categories as well as their underlying theoretical framework which allows these categories to be mapped onto a two-dimensional map. We then discuss the limitations of their framework and this conceptual map and propose a new framework that may help designers of gamified health apps locate the kind of responsibilities they may have to address ethical issues related to such apps.Kim and Werbach [2] Framework for Gamification Ethics. Kim and Werbach propose that the “ethical status of a practice of gamification, primarily, but not exhaustively, is determined by the extent to which the practice”:Is exploitative;Is manipulative;Is intentionally or unintentionally harmful to the parties involved;Has a socially unacceptable level of negative effect on the character of the parties involved.Is exploitative;Is manipulative;Is intentionally or unintentionally harmful to the parties involved;Has a socially unacceptable level of negative effect on the character of the parties involved.Exploitation—Kim and Werbach argue that gamification is exploitative in situations where it is unfair to one party. For example, if, in the workplace, gamification techniques may benefit the employer by increasing employee efficiency, but these benefits may not be translated or trickle down to employees, or may be unfair to them in other ways (such as employees not being able to say no to such techniques), then gamification can be exploitative.Manipulation—Kim and Werbach propose that since gamification essentially targets behavior change, it is prima facie open to the charge of being manipulative. In their discussion, they explore multiple accounts of manipulation, and offer two main ways in which gamification can be manipulative:When the gamification elements and mechanisms are hidden from those it is applied on (deception);When gamification techniques inhibit rational self-reflection and undermine autonomy in unjustifiable ways.When the gamification elements and mechanisms are hidden from those it is applied on (deception);When gamification techniques inhibit rational self-reflection and undermine autonomy in unjustifiable ways.They state that it is largely an empirical question whether particular instances of lack of transparency of gamification techniques or the undermining of autonomy are manipulative. One may need more information, for example, to ascertain whether lack of transparency about game elements in a particular gamified health app is intended to deceive the user or not. As examples, they argue that addiction and distraction are two ways in which gamification can undermine autonomy.Harm—Kim and Werbach [2] write that gamification can lead to both physical and psychological harm. Further, they state that, “the risks of physical harm due to gamification primarily involve injury to others outside the gamified system, while the risks of psychological harms generally involve the players themselves.”Detrimental effects on character—One threat involved with gamification is that it can rely on rewards or incentives that are detrimental to one’s character. A standard example of the negative effects of an incentive to good behavior is a parent using candy to change or nudge their child’s behavior [2]. There are two related but distinct worries about gamification in relation to effects on character: (a) individuals relying on the wrong kinds of incentives, and (b) individuals excessively or obsessively relying on an incentive.The analysis offered by Kim and Werbach relies on there being two primary reasons for these prima facie ethical issues related to gamification:Overlay of virtual and real norms;Conflict between the interests of individuals subjected to gamification and those who provide or design gamification elements.Overlay of virtual and real norms;Conflict between the interests of individuals subjected to gamification and those who provide or design gamification elements.Overlay of virtual and real norms—According to Kim and Werbach, gamification ethical issues such as manipulation or exploitation arise because gamification brings in conflict the different set of norms in play in “the real world” and the “game world”. For example, within a game, it may be acceptable to manipulate or deceive someone (such as in poker, for example). Yet, such a norm is hardly acceptable in the real world and if one were to apply a gamification technique that transposes a game-world norm to the real world, ethical issues may arise.Individual vs gamification provider—The second source of ethical tensions, according to Kim and Werbach, is dissonance between motivations and interests of those subjected to gamification, and those who provide or deploy them. For example, in a gamified workplace, an employer may want to excessively track and reward employee productivity, but employees may consider this as an infringement of privacy.This two-dimensional framework leads to the following conceptual map proposed by Kim and Werbach:Here, they deem that exploitation and manipulation are “relational” concerns since they can only be evaluated in the context of the relation between individuals subjected to gamification and those providing/designing it. For example, as stated earlier, gamification, under this framework, is exploitative when there is an asymmetry or imbalance in the consequences of gamification such that the user either does not reap symmetrical rewards, relative to the designer, or even accrues harm. On the other hand, harms and detrimental effects to character can be evaluated “purely with reference to the players as individuals” [2]. Similarly, the dimensions of real-world and game lead to different relational and individual issues. Exploitation, according to this conceptual framework, is an issue where the gamification designer exploits a real-world vulnerability of the user, while manipulation is an issue that arises because the game elements are such that they inhibit a user’s autonomy. While this framework is helpful in understanding the four prima facie ethical issues related to gamification, there are also reasons to be skeptical about whether this framework is comprehensive or appropriate enough for locating ethical issues related to health gamification.The two dimensions underlying the conceptual framework offered by Kim and Werbach—relational vs. individual and real-world vs. game-world—offer a good way to map several different ethical issues related to gamification. Yet, there are reasons to believe that the four categories defined by them—exploitation, manipulation, harm, and detrimental effects to character—only capture a narrow range of issues their conceptual framework has to offer. In what follows, we discuss the four categories further and, when applicable, present some limitations of applying these categories to the specific case of gamified health apps.Using the conceptual map (Table 1) offered by Kim and Werbach, the first category is one that should map the issues that relate to the relation between the designer and the users of a gamified system. Further, according to Kim and Werbach, these issues arise due to designers “exploiting a real-world” imbalance between designers and users [2]. There are at least two problems with labeling this category of issues as “exploitation”. First, it is not necessary that an imbalance between the designer and the user is a case of exploitation. A mere asymmetry in the distribution of rewards from the implementation of a gamified system does not constitute the wrong of exploitation. Second, there may be other kinds of wrongs that may arise out of the asymmetrical relationship between users and designers of a gamified system. Consider, for example, a gamified health app designed to motivate users to exercising more. In order to motivate its users, say that the app tracks a user’s activity, shares it with their friends on a social network, and gives them digital rewards if they outperform their friends. The social network, then, decides to allow third parties (such as other data brokers) to scrape this data off their network, which in turn, may lead to privacy harms to the user. It is hard to argue here that the designer is exploiting the user. Yet, the harm to the user is because of an imbalance between the designer and the user—namely, the choice of how the gamified app is designed and its data sharing policy rests with the designer and not the user.This is not to argue that designers of a gamified health app cannot commit the wrong of exploitation. In the previous example, if the app itself was designed to scrape and store user data for sale to a third party, it may be deemed exploitative. The argument here is that exploitation is only one of the wrongs that may be involved in the category covered by the conceptual map (Table 1) offered by Kim and Werbach.Kim and Werbach term the second category of the ethical issues of gamification “manipulation”. This category, according to their conceptual map, tracks issues that can only be evaluated in the context of how a user interacts with the game elements. Kim and Werbach term them under “manipulation” as they arise because “providers have created an environment such that, in the game, the players cannot make autonomous choices, and instead make choices that serve the providers” [2]. Again, one problem in this phrasing is that it seems to exclude cases where users’ autonomy is undermined even without designers intending that to be the case or intending it for their own purposes. One could, for example, imagine a user of a gamified health app who is obsessively addicted to the game elements (such as in-game rewards like points or badges) even without the designer intending that to be the case. Kim and Werbach also cite addiction as an example of the kind of problem they have in mind here, and it is not sufficiently clear whether they want to restrict the category to cases of such addiction being a result of users “serving their (designers’) purpose”. Further, given the practice-relevant aim of the framework, the omission of such cases may not cover the full scope of the potential duties and responsibilities of designers and providers of a gamified system. One could argue that designers and providers of gamified systems are not merely responsible for consequences of intended actions, but also, at least some of the unintended actions. Many philosophers and ethicists, for example, believe that people should not only be held morally responsible for wrongdoings they are aware of, but also in cases where they should have known better [14]. Ascriptions of such moral responsibility (for should-have-known cases) may be even more justified in cases where the professional role of a person may morally require them to have known certain things. A doctor, for example, cannot claim ignorance for misdiagnosing a disease they were not, but should have been, aware of. In cases of gamified health apps, we may find similar cases where the designers and providers of such apps may be morally required to inquire into at least some of the ways in which the app undermines user autonomy.Another limitation of discussions offered regarding both category 1 (user–designer in the real world) and category 2 (game–user relation) is that it does not distinguish between different roles providers and designers of a gamified system play. Kim and Werbach use both terms in their paper, but do not elaborate on how each could affect the system in their role. Distinguishing between the morally relevant actions available to each could be significant for the practice-relevant aims of the framework.The third category is not relational, in the sense that to evaluate ethical issues within this category, one need not look at the actions of the designer or the game elements. This category tracks a consequentialist approach to gamification ethics. A consequentialist approach to ethics, as the name suggests, is roughly the idea that whether an act is ethical or not depends on the consequences of that act [15]. Applied to the case of health gamification, this approach dictates that one only needs to look at the consequences of the game/gamified system, in the real world, to determine whether an individual has been harmed. In their introduction to this category, they state that it primarily involves physical harms to other individuals and psychological harms to the user of the gamified system. Yet, they do give some examples where the user may also be physically harmed, so the category should indeed include harm of a physical and psychological nature to both users of the game as well as others affected by it.The fourth category deals with ethical issues that are also not relational and arise in the game. Kim and Werbach define this category as one which has issues that arise if “there is an ethical lapse in the game, such that players act to satisfy the game’s objectives and are indifferent to fundamental human values”. Yet, stating the problem this way also narrows down the potential problems involved here. Specifically, defined this way, the category leaves out issues where a user of a gamified system acquires character flaws that are not simply “lapses in the game” but also carry outside it.Besides the limitations already discussed, there may be additional problems that the framework does not address.First, health, as a category, has an important social and structural dimension that may not be covered by a focus on individuals and their motivations. One’s health status, as well as one’s possibilities to engage in a healthy lifestyle, are conditioned by social factors such as one’s relative economic or social status. This may mean that the conflict between the motivations of individual players and gamification designers may not capture the entire breadth of ethical issues associated with gamification in healthcare. This may be further exacerbated by the fact that many gamified health apps also deliberately include social dimensions, such as leaderboards, competitions, badges, etc., and there is also evidence that users of such apps actively seek social validation in their gameplay [16].Second, the dimensional contrast between the real world versus the game world may also be elusive. In their discussion of how to identify the relevant ethical concern for an individual, Kim and Werbach write:“If the gamification activity produces an injury manifested in the real world, whether physically or psychically, the issue is one of harm. If instead there is an ethical lapse in the game, such that players act to satisfy the game’s objectives and are indifferent to fundamental human values, the issue is character.”Yet, gamified health apps are not perfectly closed environments and there may be instances in health gamification where the game world may reinforce or affect the norms in the real world, and it may not be easily determinable whether the ethical concern arises in the game world or the real world. A gamified health app, for example, may not only push a player to satisfy the goals in the game, it may also change or influence what the player deems to be healthy in the real world too.That there are limitations to the application of Kim and Werbach’s framework and conceptual map to the specific case of gamified health apps is not surprising. Kim and Werbach also anticipate this possibility, as they state that (a) their framework is conceptualized with the case of gamification in the workplace at the forefront, and (b) their attempt was not to provide a comprehensive mapping, leaving open the possibility of issues that may not be covered by their framework. Further, the discussion offered by Kim and Werbach does make important strides towards their practice-relevant aim of outlining ethical issues with gamification that could be useful for designers and providers of gamification. They also make the important observation that analyzing and identifying ethical problems with gamification requires more than just a consequentialist perspective in that not all the wrongs associated with such practices are related to the outcomes of the gamified system. Some of the wrongs, for example, are better analyzed from a virtue ethics approach, to figure out how a gamified system or app affects a user’s character. (In contrast to consequentialism, virtue ethics emphasizes moral character.) [17]. A virtue ethicist, for example, would recommend helping someone not for its consequences but because it is benevolent to do so. Similarly, from the point of view of the designers, which is the focus of this paper, a deontological perspective can give us crucial insights. Kim and Werbach, for example, state that in outlining the problems of exploitation and manipulation, they appeal to the deontological values of autonomy, fairness, and reason-responsiveness. In this section, we build on such insights offered by Kim and Werbach, and propose a new framework, geared towards outlining the types of responsibilities designers of gamified health apps have.Before we outline our proposed framework, however, some important observations need to be stated. First, as discussed in Section 2.2, the designers of gamified apps may have the responsibility of preventing not just intentional wrongdoings, such as exploitation of vulnerabilities or undermining user autonomy, but also unintentional wrongdoings, especially cases where they can be expected to have known better. The latter may require designers, for example, to actively inquire into the outcomes of their designed apps, and as Kim and Werbach’s discussion illustrates, such inquiry should not be limited to a consequentialist perspective. Designers may also need to reflect on whether their design has negative effects on the character development of the users. Further, following a deontological (or Kantian) approach, designers may need to reflect on the possible wrongs that arise out of a designer’s lack of respect for the user or treating them as a mere means. On Kant’s view, we must always have respect for persons and there is something intrinsically wrong in treating them as mere means [18]. To treat them as mere means implies treating them only for our own ends and advantages, without regard for their interests [18]. For a designer to have respect for the users, and not treat them as mere means, implies being responsive to the needs and values of the users. Given that gamified health apps operate in the context of health, where special duties of care and beneficence are often emphasized, designers of such apps may even have special duties to actively consider the needs of the users [19]. In this sense, designers may be said to have design-related duties that are negative, in the sense that they require them to not harm the users, as well as duties that are positive, that require them to actively consider the good of the users.Second, while designers have active responsibilities related to the design of the apps, other stakeholders may also share responsibilities for outcomes related to the use of an app. This includes users, but also providers, or other stakeholders who may force, push, or incentivize users to use such apps. One example could be a physician or a doctor who prescribes the use of a gamified health app to her patients. In such cases, these stakeholders may be more aware of the contexts within which a user is using the app, which may dictate that they also have moral responsibilities for outcomes for the users. Even in such cases, however, designers may also have responsibilities that are not limited to design. Such a model, where designers not only have responsibilities to make a “safe design” but also to actively and responsibly share responsibilities, has been argued by van de Poel and Robaey [20], amongst others. Under such a model, designers may, for example, be required to engage with such physicians and doctors in not only understanding the best design features, but also to communicate how to best integrate the app with other kinds of interventions physicians or doctors may be planning. Such communication-based duties may also be stated in terms of designers’ relation with, and as part of, the general society they inhabit. As stated, health as a category has an important social and structural dimension, and gamified health apps exist within such social and structural conditions. As far as possible, designers may need to engage with such social and structural systems to ensure that their apps are responsive to such conditions and that others also understand their role and utility as best as possible.Another important stakeholder with whom designers may have to share responsibility is, of course, the user. From an ethical perspective, the need for such sharing or even transfer of responsibility to the user arises from the possibility that users may deviate significantly from what the designers intend or foresee as a way to use or engage with the gamified app. This includes misuse of the app in ways that are harmful to the user. One way in which designers can share or transfer responsibility to the users is through a “use plan” [21,22]. Summarizing the work of Houkes & Vermaas  [22], Poel and Robaey [20] write that “the design of artifacts always includes the design of use plans, [where] a use plan is a sequence of actions with an artifact that will lead to the realization of a goal”. Such use plans may be communicated by the designer to the user through written manuals but also through other ways such as instructional videos, advertisements, etc., [20,22]. Further, use plans may even allow users to deviate from plans designers had in mind. Robaey [23] has argued that successful use plans should even consider such deviations, and encourage users to adapt to the use of artefacts in particular contexts to avoid hazards. To this end, Robae argues for epistemic access to the design of the artefacts, such that the artefact is not a black box for the user. The point here is not to argue that gamified health apps should necessarily have such use plans, but that designers of gamified health apps may have duties and responsibilities related to the successful transferring or sharing of responsibilities to/with the users of the apps.With these observations in mind, we can now propose our framework, based on the three types of responsibility designers of gamified health apps may have:Responsibilities for proper design—As the name suggests, this includes the responsibilities of the designers directly related to the design of their gamified health apps. This involves, for example, negative duties which require designing the game elements such that the users are not harmed or wronged, as well as potentially positive duties which help or facilitate the achievement of the user’s good. As stated, such duties may also involve designers actively inquiring into the consequences of their design activities.Responsibilities to facilitate proper use—While design features are an essential part of facilitating an ethically good user experience, design and designers cannot account for all possible outcomes from the use of a gamified health app. There are various uncertainties and indeterminacies related to how users will, in practice, use the app. Avoiding wrongdoings because of, for example, misuse of the app, requires that designers share and transfer some of the responsibility to current as well as prospective users. As mentioned, one way to achieve this would be through use plans that designers can share with the users. There may also be other ways in which designers may encourage morally desirable behavior in users as well as foster the virtue of taking responsibility amongst users. One example may be through designer-organized forums and meetings that facilitate interaction amongst current and prospective users, such that they are able to share and create new beneficial ways of engaging with the apps that even designers may not have anticipated. There is evidence, for example, that such forums and meetings have helped members of the Quantified Self (QS) movement, which includes users of apps such as Fitbit, that measure and promote physical activity [24].Responsibilities related to ensuring proper embedding of the apps within the larger social context—Besides users, designers may also need to share responsibilities with other stakeholders associated with gamified health apps. This may include the general public, but may especially include actors whose actions are directly related to gamified health apps. This includes, for example, and as stated earlier, doctors and physicians who may want to use such gamified health apps in planned interventions for their patient groups. It may also include insurance companies who may want to include data from gamified health apps and offer users monetary incentives to be more physically active in demonstrable ways. As informed stakeholders who may understand the nuanced ways in which the actions of actors such as the aforementioned insurance companies and physicians may affect users of gamified health apps, designers may have the responsibility to engage in interactions with other actors to facilitate the use of such apps in ways that promote better outcomes. The designers’ duties may also involve pushing forward and facilitating an active and democratic societal discourse on how such apps may be used and integrated within a given society’s health system. This may especially include engaging with other designers of such gamified health apps. More generally, there is a need for designers to reflect more broadly on the wider social and economic implications of their apps.Responsibilities for proper design—As the name suggests, this includes the responsibilities of the designers directly related to the design of their gamified health apps. This involves, for example, negative duties which require designing the game elements such that the users are not harmed or wronged, as well as potentially positive duties which help or facilitate the achievement of the user’s good. As stated, such duties may also involve designers actively inquiring into the consequences of their design activities.Responsibilities to facilitate proper use—While design features are an essential part of facilitating an ethically good user experience, design and designers cannot account for all possible outcomes from the use of a gamified health app. There are various uncertainties and indeterminacies related to how users will, in practice, use the app. Avoiding wrongdoings because of, for example, misuse of the app, requires that designers share and transfer some of the responsibility to current as well as prospective users. As mentioned, one way to achieve this would be through use plans that designers can share with the users. There may also be other ways in which designers may encourage morally desirable behavior in users as well as foster the virtue of taking responsibility amongst users. One example may be through designer-organized forums and meetings that facilitate interaction amongst current and prospective users, such that they are able to share and create new beneficial ways of engaging with the apps that even designers may not have anticipated. There is evidence, for example, that such forums and meetings have helped members of the Quantified Self (QS) movement, which includes users of apps such as Fitbit, that measure and promote physical activity [24].Responsibilities related to ensuring proper embedding of the apps within the larger social context—Besides users, designers may also need to share responsibilities with other stakeholders associated with gamified health apps. This may include the general public, but may especially include actors whose actions are directly related to gamified health apps. This includes, for example, and as stated earlier, doctors and physicians who may want to use such gamified health apps in planned interventions for their patient groups. It may also include insurance companies who may want to include data from gamified health apps and offer users monetary incentives to be more physically active in demonstrable ways. As informed stakeholders who may understand the nuanced ways in which the actions of actors such as the aforementioned insurance companies and physicians may affect users of gamified health apps, designers may have the responsibility to engage in interactions with other actors to facilitate the use of such apps in ways that promote better outcomes. The designers’ duties may also involve pushing forward and facilitating an active and democratic societal discourse on how such apps may be used and integrated within a given society’s health system. This may especially include engaging with other designers of such gamified health apps. More generally, there is a need for designers to reflect more broadly on the wider social and economic implications of their apps.As stated, this paper has the dual aim of providing a practice-relevant theoretical framework to address ethical issues with gamified health apps as well as to provide a landscape of various ethical issues related to gamified health apps as found in the empirical literature about such apps. The first aim—the proposed theoretical framework—facilitates the taxonomizing of ethical issues related to gamified health apps, based on the type of designer action they may be addressed by. The second aim—of providing a landscape of empirically identified ethical issues related to gamified health apps—serves as a guiding list for designers of such apps and facilitates the addressal of such ethical issues. The attempt here is also to see how such empirically identified ethical issues may be addressed by the three types of designer responsibilities we have outlined here. Mapping the identified ethical issues on our practice-oriented framework would be an aid to the designers of gamified health apps who may seek to avoid harm to the users of such apps. In the next section, we discuss our methodology to answer the main question about what such effects on users of gamified health apps are: What ethical issues can be identified in the existing empirical work on the effects of gamification in health tracking?To answer our question, we conducted a systematic review of the literature on the effect of health gamification. We reviewed those publications that discuss the effects of gamified apps based on health and fitness tracking. The main aim of the systematic literature review, as stated earlier, was to achieve our second objective in this paper: facilitating a landscape of empirically identified ethical issues encountered in use of gamified health apps. While extracting these ethical issues from the empirical literature, we also noted recommendations for designers of gamified health apps given in the literature to address such ethical issues. We then mapped these recommendations onto our tripartite framework as proof of its utility in taxonomizing various ethical issues related to gamified health apps and corresponding designer responsibilities.The study protocol consisted of the following steps:Search for papers published after 2010 that discuss the effects of gamification in health and fitness apps (see Section 3.2 for details of search string and criteria).Remove duplicates from the retrieved articles.Apply the inclusion and exclusion criteria described in Section 3.2.Apply backward snowballing method to systematic reviews within our reference list to find additional studies.Check for sampling bias by searching for strings related to “ethics of health gamification”.Extract data from the selected papers to answer our research question.Search for papers published after 2010 that discuss the effects of gamification in health and fitness apps (see Section 3.2 for details of search string and criteria).Remove duplicates from the retrieved articles.Apply the inclusion and exclusion criteria described in Section 3.2.Apply backward snowballing method to systematic reviews within our reference list to find additional studies.Check for sampling bias by searching for strings related to “ethics of health gamification”.Extract data from the selected papers to answer our research question.To select search databases and design our search string, we analyzed methodologies described in other systematic reviews on gamification in health (these are [5,25,26,27,28]. Since these reviews had different research questions than ours, we modified our search string and database list accordingly. Based on these reviews and the needs of our study, the electronic databases used included those identified as relevant to information technology, social science, ethics, psychology, and health: ACM digital library, Scopus, Web of Science, PubMed, PhilPapers, and IEEE explore. Following the account of the timeline of the popularity of gamification in health in [5,28], only papers after 2010 were included. While our main purpose was to identify potential ethical issues related to gamification in health and fitness, based on results and methodology used by [28], we were aware that such issues may be referred to in the literature as “negative”, “unintended” effects, “risks”, or similar terms, and designed our search string accordingly. Prior to applying the search protocol, we had also already identified that papers by [3,29,30] were relevant for our study. We, therefore, used these papers to use as a control, to make sure our search string did not skip relevant results. The following is our final search query (used for ACM database):“query”: { Abstract:(gamif*) AND AllField:(health* OR medic* OR life* OR fitness OR well-being) AND AllField:(risk* OR danger* OR peril* OR effect* OR negative* OR unintended OR ethics OR ethical) }“filter”: { Article Type: Research Article, Publication Date: (01/01/2010 TO 12/31/2020), ACM Content: DL, NOT VirtualContent: true }This search strategy resulted in 621 results of which 459 were unique.We then applied the following inclusion and exclusion criteria to narrow our search:Inclusion criteria:Peer-reviewed (incl. peer-reviewed conference papers);Full papers (incl. full conference papers);Clearly focused on gamification and described gamification elements (type of game design elements);Addresses gamification in health and fitness tracking through use of devices and/or mobile apps;Discusses empirical evidence related to the effects of such apps. The empirical evidence here denotes a reported effect of a gamified health app. The effect could be in terms of impact (affect, behavior, social, cognitive) or in terms of user experience when using the gamified health app.Peer-reviewed (incl. peer-reviewed conference papers);Full papers (incl. full conference papers);Clearly focused on gamification and described gamification elements (type of game design elements);Addresses gamification in health and fitness tracking through use of devices and/or mobile apps;Discusses empirical evidence related to the effects of such apps. The empirical evidence here denotes a reported effect of a gamified health app. The effect could be in terms of impact (affect, behavior, social, cognitive) or in terms of user experience when using the gamified health app.The first two criteria were developed to maintain the quality of the articles. The second and last two were developed to make sure the literature clearly focuses on gamification within health and fitness tracking. We screened the articles initially based on their titles. We then consulted the abstract or the text of the article when it was necessary to reach a confident judgement. Based on these criteria, 80 relevant papers were found.The exclusion criteria focus on excluding literature which only superficially mentions our terms of interest but does not contain sufficient detail for analysis:Mentions health and fitness tracking but do not explicitly focus on gamification in such devices;Addresses gamification in health tracking but does not give relevant empirical information on the effects of such gamification.Mentions health and fitness tracking but do not explicitly focus on gamification in such devices;Addresses gamification in health tracking but does not give relevant empirical information on the effects of such gamification.Here, relevant empirical evidence is deemed limited to:Evidence about the effect of gamified health app on the user through qualitative user feedback (surveys, questionnaires, user reviews);Evidence about potential negative effects of gamified health app through content analysis of the app.Evidence about the effect of gamified health app on the user through qualitative user feedback (surveys, questionnaires, user reviews);Evidence about potential negative effects of gamified health app through content analysis of the app.In applying these exclusion criteria, the initially identified papers were carefully analyzed. Following this screening, we did backwards snowballing to two relevant systematic reviews, which also discussed the empirical effects of gamification in health, included in our list to retrieve additional papers. This gave us a list of 23 final papers. We also searched for multiple permutations of the strings “ethics of health gamification”, “negative effects of health gamification”, etc. in Google Scholar to check for papers published after 2010 that may have been missed. We manually screened through the first 50 results and did not find any relevant studies that had not already been included.To facilitate objectivity, we piloted the process of inclusion and exclusion using 10 papers that were independently assessed by three different researchers, including the authors of this paper. The rest of the articles were screened for inclusion and exclusion after it was established that the three assessors agreed on the inclusion and exclusion assessment of the 10 articles.All selected papers were read in their entirety, looking for relevant phrases, arguments, or discussion points that address some ethical issue or negative effects related to gamification in health and fitness tracking. For studies that were based on empirical evidence regarding subjective user experience of using a gamified health app, we only counted it as a reported effect and/or a related ethical issue when the study reported it as a significant effect, for example, because it was applicable for a significant number of users (and not, for example, when researchers expected to find it or mentioned it as a possible issue but which was not studied). Besides effects reported from such studies of user experience, we also included a couple of studies which were based on discourse analysis of gamified health apps. These studies looked at game elements and linguistic components (words used to describe health status or prospective users, for example) of the app and applied sociological theories to articulate the ethical issues at play with the apps. Through our analysis, we collected a list of all reported negative effects and/or related ethical issues of gamified health apps.Of the selected papers, roughly 50% (12 out of 23) were based on qualitative studies and employed methods such as semi-structured interviews of a selected group of users of gamified health apps. These studies monitored the users over multiple days, ranging from one week to eight weeks. Nine (39%) were based on surveys or questionnaires conducted over a large number of existing users of gamified health apps. A third of the studies (33%) focused on a range of gamified health apps, while the rest were focused on a particular gamified health app. Among the latter, 33% (5 out of 15) used an app (or a prototype) not yet available in the public domain, while the rest employed use of an existing, often popular, app.Our review of the literature yielded various potential ethical issues with gamified health apps. Table 2 gives an overview of these issues along with sources citing such issues. While describing these issues, we also noted recommendations within the identified literature to designers of gamified health apps of ways in which they may potentially address these issues. We then mapped these recommendations onto our tripartite framework based on the type of designer responsibility a given ethical issue may be addressed by.In Table 3, we encapsulate how the recommendations in the literature for designers of gamified health apps to address these ethical issues can be mapped onto our proposed categories. We indicate which type of designer responsibility may potentially address that particular ethical issue. It should be noted that while the table only includes recommendations in the literature, these are by no means an exhaustive set of recommendations to designers of gamified health apps related to addressing potential ethical issues. In the text below, as examples of further possible steps for designers, we also note some additional observations of our own. For example, although Maturo & Setiffi [30] analyze and introduce the issues of biosociality, amorality, and the neoliberal objection, they do not offer an explicit recommendation to address these issues. Their analysis, however, can be used to deduce some possible steps, and we note them below, in the text. Further, our recommendations are also not meant to be exhaustive and there are, of course, other steps designers could take to address some of these ethical issues. For example, we indicate that privacy-related issues may be addressed by proper design and proper use, as noted in the analysis presented above. This should not be taken to mean that there aren’t ways to address such privacy issues through means that may be characterized as belonging to the third category, of proper embedding in the social system. Table 3, and our analysis in general, indicate the scope for future research—particularly the need to investigate other ways in which designers of gamified health apps may address potential ethical issues by assuming one of the types of responsibility in our framework.
+        Privacy-related issues—Privacy was a chief concern among many users of gamified health apps. We found multiple studies that reported users being concerned about lack of privacy when using a gamified health app. This concern was either a result of users not comfortable with their data being tracked or shared, or because they were unsure how their data may be used by the app. Users also expressed concern with certain features of the app, intentionally designed, to lure them into using the app more or reminding them to use it. There was also evidence that some apps were intentionally designed to lure users into sharing more personal data [35]. There was also evidence of the privacy concerns of users translating into psychological concerns, such as feelings of being surveilled and corresponding anxiety. This clearly points to the need for designers to assume the responsibility of protecting user privacy. Orji et al. [31], for example, suggest that app designers should allow users to hide their identity and other personal information from other users of the app. They also suggest other “personalization” features to allow users to choose what information is shared and collected about them. Trang and Weiger [35] suggest that app providers should explicitly ask users’ permission before processing private information as well as inform users as much as possible about ways in which their information is used.Cognitive manipulation—In their review of multiple gamified health apps, Maturo & Setiffi [30] write of apps exploiting concepts from cognitive psychology to manipulate users into using apps or oversharing information on them. Such design features are also partly responsible for the addicting nature of such apps, and Attig and Franke [3] have done an important study demonstrating the dependence of users on gamified health apps. Attig and Franke [3] write that such features rarely lead users into adopting an active lifestyle (or exercise) in the long run, and that designers should instead focus on facilitating the internal motivation of the users.Dependence and addiction—Besides Attig and Franke [3], Barratt [29], in his qualitative study on the use of gamified apps by cyclists, also found evidence of such dependence and addiction to apps. Barratt also reported that some users also found their autonomy constrained, as they did not expect they would be so easily lured into the game rewards and incentives, such that they would complete the game challenges sometimes at the expense of other important personal and social commitments. At least some of these effects, at least to some extent, may be unforeseeable or unintended by the designers. It is hard to say from the available evidence the extent to which issues such as addiction or extreme dependence on an app are always solely a result of design features and not unhealthy ways of engaging with an app on the part of the user. As mentioned earlier, Attig and Franke write that app dependence rarely translates into user’s adopting a healthy lifestyle in the long run, and designers are better off aiming for the internal motivation of users. They suggest apps that allow for self-determination and are self-rewarding for users. Some of this may also be rectified by designers sharing or transferring responsibility (for proper use) to users. Yet, in so far as these issues are foreseeable, some, or significant responsibility also lies with the designers of apps, depending on the circumstances and game elements of the app.Psychological harm—A similar case exists for design features that potentially lead to psychological harm to the users other than dependence or obsession with game rewards. These include, as stated earlier, feelings of being surveilled, and not feeling under control (lack of perceived autonomy). Some users also experienced extreme psychological states (such as anger or anxiety) because of gamified health app. This could be sometimes caused by lagging in the competition (or not having enough game rewards) or also when users suspected others of cheating [41]. Some design features seem to be responsible for incentivizing users to cheat, although part of the responsibility, again, lies with the users as well. A more concerning psychological aspect of gamified health apps seems to be their detrimental effects on existing internal motivations, as well as on the confidence of users [40], for example, point out that some users can be left with a strong sense of defeat, and it is therefore very important that game elements are designed to avoid such scenarios, particularly in serious contexts such as gamified systems for improving heart activity. There is definitely a case to be made for designers to review such cases and ensure that design features minimize the occurrence of such negative effects as much as possible. Besides the moral implications of such negative effects on users, evidence also suggests that it has adverse effects on user engagement with apps and leads to discontinuance [9]. Recommendations within the literature include: giving users more autonomy and personalization of app features [31], allowing cheating to a limited extent (for example, by allowing users more autonomy over how their results are displayed and building an app community that is tolerant of individual users making such choices in order to save “face”) [41], avoiding giving users a sense of defeat in serious apps [40]. Physical harms—Gamified health apps use game elements to motivate users to increase physical activity in their lives. However, for some users, this may result in side effects such that they may overexert themselves or engage with the app in ways that are harmful to them. The most obvious evidence of physical harm was through reports of users overtraining or overstressing themselves in search of game rewards [29]. At least some of these harms may be reduced through the use plans and other strategies designers may employ to transfer responsibility for proper use to the users. As discussed, there may be other ways of fostering virtuous use of apps amongst users by facilitating forums and other places where users may learn from each other how they can best engage with an app.Hermeneutic problems—Designers and design features also seem to be directly responsible for various “hermeneutic” problems posed by gamified health apps. This problem relates to the use of terms within the app that may reinforce stereotypes. Lupton & Thomas [46], for example, write of gamified pregnancy apps which represent pregnant women in stereotypical ways, such as a Barbie doll.A related concern comes from Maturo and Setiffi [30] who argue that gamified health apps “atomistically insulate” individuals from other individuals even though, simultaneously, the individuals are “widely socially connected through a potential network of app users”. This insulation of users brackets out the social determinants/dimension of health in a sort of hermeneutic reductionism [30]. This hermeneutic reduction can lead to a phenomenon that Cheng [48] describes, where users feel pressured or compelled to look for and log only particular types of data, possibly at the cost of what they may have found meaningful or motivational. For example, Cheng [48] notes that “by only providing functionality to record performance metrics (i.e., distance, duration and location of a run), and rewarding based on these metrics, the Nike+ system implicitly communicates the other enjoyable aspects of running, such as the runner’s high, or the mindful interaction between human and environment, are less important”. Additionally, the proxies used in gamification elements can come to represent definite truths about what they are gamifying, as well as become privileged over other ways of knowing. This points to the problem of gamified health apps not properly embedded within a larger structural context.Biosociality—The problem entails that certain gamified apps may reinforce physical stereotypes and also force the formation of groups based on such physical attributes [30]. Designers’ efforts of fostering and encourage virtuous behavior for proper use among users may partly address this problem.The Neoliberal objection—As previously stated, factors such as education, income, and living condition have a huge influence on one’s health status [30]. Designers of gamified health apps should also be aware of the social dimensions and contexts within which their apps are used. It has been argued that the individualistic view underlying gamified health apps can lead to problems such as the depoliticization of the role of the state, which reduces the responsibilities of the state for the health of its citizens and shifts the burden to individuals. This objection states that such apps “foster a neoliberal ideology that implicitly stigmatizes people who are not capable of meeting the standard definition of ‘healthy’” [30]. Through a discourse analysis of major gamified health apps, Maturo and Setiffi [30] point out how the design and linguistic features of such apps may lead to such stigmatization. While designers can “fix” some of these linguistic issues, more a holistic solution to such problems perhaps lies in more active engagement of the designers with other actors and stakeholders in society. This could enable a more successful integration and embedding of gamified health apps within a larger structural quest to promote healthy lifestyle and outcomes for citizens.Amorality—Another detrimental effect of gamified apps is that they may lead to/incentivize users to choose goals that are potentially harmful without caveats. This issue may be characterized as one of individual users choosing the wrong kind of incentive within a game, and it may be partly addressed by both better design features and virtuous user engagement with the app. Yet, as Maturo and Setiffi [30] point out, this can be more than an individual issue, and one where social norms may play a part. For example, dieting apps may lead a user to choose goals that other users are accomplishing or people around them find healthy, rather than what may actually be healthy for the individual.Issues related to providers and facilitators in specific contexts—Finally, there is also evidence of there being merit in designers engaging with providers or facilitators of gamified health apps in particular contexts such as doctors and physicians. Writing about the use of gamified health app in the context of therapy (for mental well-being), [47], for example, write how therapists could benefit from having more control (and hence, responsibility) over the features of apps, and that this could be done through direct interactions between the designers of such apps and the therapist planning an intervention that uses the app.Privacy-related issues—Privacy was a chief concern among many users of gamified health apps. We found multiple studies that reported users being concerned about lack of privacy when using a gamified health app. This concern was either a result of users not comfortable with their data being tracked or shared, or because they were unsure how their data may be used by the app. Users also expressed concern with certain features of the app, intentionally designed, to lure them into using the app more or reminding them to use it. There was also evidence that some apps were intentionally designed to lure users into sharing more personal data [35]. There was also evidence of the privacy concerns of users translating into psychological concerns, such as feelings of being surveilled and corresponding anxiety. This clearly points to the need for designers to assume the responsibility of protecting user privacy. Orji et al. [31], for example, suggest that app designers should allow users to hide their identity and other personal information from other users of the app. They also suggest other “personalization” features to allow users to choose what information is shared and collected about them. Trang and Weiger [35] suggest that app providers should explicitly ask users’ permission before processing private information as well as inform users as much as possible about ways in which their information is used.Cognitive manipulation—In their review of multiple gamified health apps, Maturo & Setiffi [30] write of apps exploiting concepts from cognitive psychology to manipulate users into using apps or oversharing information on them. Such design features are also partly responsible for the addicting nature of such apps, and Attig and Franke [3] have done an important study demonstrating the dependence of users on gamified health apps. Attig and Franke [3] write that such features rarely lead users into adopting an active lifestyle (or exercise) in the long run, and that designers should instead focus on facilitating the internal motivation of the users.Dependence and addiction—Besides Attig and Franke [3], Barratt [29], in his qualitative study on the use of gamified apps by cyclists, also found evidence of such dependence and addiction to apps. Barratt also reported that some users also found their autonomy constrained, as they did not expect they would be so easily lured into the game rewards and incentives, such that they would complete the game challenges sometimes at the expense of other important personal and social commitments. At least some of these effects, at least to some extent, may be unforeseeable or unintended by the designers. It is hard to say from the available evidence the extent to which issues such as addiction or extreme dependence on an app are always solely a result of design features and not unhealthy ways of engaging with an app on the part of the user. As mentioned earlier, Attig and Franke write that app dependence rarely translates into user’s adopting a healthy lifestyle in the long run, and designers are better off aiming for the internal motivation of users. They suggest apps that allow for self-determination and are self-rewarding for users. Some of this may also be rectified by designers sharing or transferring responsibility (for proper use) to users. Yet, in so far as these issues are foreseeable, some, or significant responsibility also lies with the designers of apps, depending on the circumstances and game elements of the app.Psychological harm—A similar case exists for design features that potentially lead to psychological harm to the users other than dependence or obsession with game rewards. These include, as stated earlier, feelings of being surveilled, and not feeling under control (lack of perceived autonomy). Some users also experienced extreme psychological states (such as anger or anxiety) because of gamified health app. This could be sometimes caused by lagging in the competition (or not having enough game rewards) or also when users suspected others of cheating [41]. Some design features seem to be responsible for incentivizing users to cheat, although part of the responsibility, again, lies with the users as well. A more concerning psychological aspect of gamified health apps seems to be their detrimental effects on existing internal motivations, as well as on the confidence of users [40], for example, point out that some users can be left with a strong sense of defeat, and it is therefore very important that game elements are designed to avoid such scenarios, particularly in serious contexts such as gamified systems for improving heart activity. There is definitely a case to be made for designers to review such cases and ensure that design features minimize the occurrence of such negative effects as much as possible. Besides the moral implications of such negative effects on users, evidence also suggests that it has adverse effects on user engagement with apps and leads to discontinuance [9]. Recommendations within the literature include: giving users more autonomy and personalization of app features [31], allowing cheating to a limited extent (for example, by allowing users more autonomy over how their results are displayed and building an app community that is tolerant of individual users making such choices in order to save “face”) [41], avoiding giving users a sense of defeat in serious apps [40]. Physical harms—Gamified health apps use game elements to motivate users to increase physical activity in their lives. However, for some users, this may result in side effects such that they may overexert themselves or engage with the app in ways that are harmful to them. The most obvious evidence of physical harm was through reports of users overtraining or overstressing themselves in search of game rewards [29]. At least some of these harms may be reduced through the use plans and other strategies designers may employ to transfer responsibility for proper use to the users. As discussed, there may be other ways of fostering virtuous use of apps amongst users by facilitating forums and other places where users may learn from each other how they can best engage with an app.Hermeneutic problems—Designers and design features also seem to be directly responsible for various “hermeneutic” problems posed by gamified health apps. This problem relates to the use of terms within the app that may reinforce stereotypes. Lupton & Thomas [46], for example, write of gamified pregnancy apps which represent pregnant women in stereotypical ways, such as a Barbie doll.A related concern comes from Maturo and Setiffi [30] who argue that gamified health apps “atomistically insulate” individuals from other individuals even though, simultaneously, the individuals are “widely socially connected through a potential network of app users”. This insulation of users brackets out the social determinants/dimension of health in a sort of hermeneutic reductionism [30]. This hermeneutic reduction can lead to a phenomenon that Cheng [48] describes, where users feel pressured or compelled to look for and log only particular types of data, possibly at the cost of what they may have found meaningful or motivational. For example, Cheng [48] notes that “by only providing functionality to record performance metrics (i.e., distance, duration and location of a run), and rewarding based on these metrics, the Nike+ system implicitly communicates the other enjoyable aspects of running, such as the runner’s high, or the mindful interaction between human and environment, are less important”. Additionally, the proxies used in gamification elements can come to represent definite truths about what they are gamifying, as well as become privileged over other ways of knowing. This points to the problem of gamified health apps not properly embedded within a larger structural context.Biosociality—The problem entails that certain gamified apps may reinforce physical stereotypes and also force the formation of groups based on such physical attributes [30]. Designers’ efforts of fostering and encourage virtuous behavior for proper use among users may partly address this problem.The Neoliberal objection—As previously stated, factors such as education, income, and living condition have a huge influence on one’s health status [30]. Designers of gamified health apps should also be aware of the social dimensions and contexts within which their apps are used. It has been argued that the individualistic view underlying gamified health apps can lead to problems such as the depoliticization of the role of the state, which reduces the responsibilities of the state for the health of its citizens and shifts the burden to individuals. This objection states that such apps “foster a neoliberal ideology that implicitly stigmatizes people who are not capable of meeting the standard definition of ‘healthy’” [30]. Through a discourse analysis of major gamified health apps, Maturo and Setiffi [30] point out how the design and linguistic features of such apps may lead to such stigmatization. While designers can “fix” some of these linguistic issues, more a holistic solution to such problems perhaps lies in more active engagement of the designers with other actors and stakeholders in society. This could enable a more successful integration and embedding of gamified health apps within a larger structural quest to promote healthy lifestyle and outcomes for citizens.Amorality—Another detrimental effect of gamified apps is that they may lead to/incentivize users to choose goals that are potentially harmful without caveats. This issue may be characterized as one of individual users choosing the wrong kind of incentive within a game, and it may be partly addressed by both better design features and virtuous user engagement with the app. Yet, as Maturo and Setiffi [30] point out, this can be more than an individual issue, and one where social norms may play a part. For example, dieting apps may lead a user to choose goals that other users are accomplishing or people around them find healthy, rather than what may actually be healthy for the individual.Issues related to providers and facilitators in specific contexts—Finally, there is also evidence of there being merit in designers engaging with providers or facilitators of gamified health apps in particular contexts such as doctors and physicians. Writing about the use of gamified health app in the context of therapy (for mental well-being), [47], for example, write how therapists could benefit from having more control (and hence, responsibility) over the features of apps, and that this could be done through direct interactions between the designers of such apps and the therapist planning an intervention that uses the app.Recommendations in the literature to app designers to address ethical issues.Personalization to allow users to choose what information they want to shareExplicit notification and seeking user permission before processing private informationInform users explicitly about how their private data will be usedAvoiding manipulative featuresProviding warning and safety restrictions against harmful cognitive effectsAllowing users to be more self-determined and self-rewarding in the usage of the app rather than offering extrinsic pre-determined rewardsGiving users more control over reward featuresAvoiding psychological harms such as a sense of defeat in serious appsGiving users more autonomy in choosing their goalsTolerating some level of “cheating” from users if that translates to better health choicesAn empathetic approach to design that allows users to be autonomous and some self-determination over their goals as well as how they might be displayedFacilitating a community of users who are empathetic to other users of the appAllowing users to choose their own goalsImproving user autonomy as well as giving warnings about dangers of overuse and exertionAvoiding game elements/rewards/terms that may reinforce harmful stereotypesAvoiding reductionism in rewarding systems/game elements, for example, in ways that may incentivizes users to interpret their health and lifestyle in potentially harmful terms and/or though narrowly conceived metricsEngage with providers and facilitators of gamified health apps such that the apps cater to appropriate contextual information.Our first task in this paper was to analyze and revise the framework offered by Kim and Werbach to identify ethical issues in gamification. Based on a theoretical analysis of this framework and arguments from moral theory, we argued for a revised practice-relevant theoretical framework that suggests three broad categories of responsibilities designers have in addressing ethical issues in gamified health apps. We have argued that the categorization, and the framework encapsulating this categorization we propose are better-equipped to help practitioners, such as designers, in the specific context of gamified health and fitness apps. This categorization also served as a guideline to identify and map ethical issues that have been discussed in the empirical literature on gamified health apps. We presented these in Table 3 in Section 3. We want to emphasize that, theoretically, there are more issues as well as steps designers could take to address those issues, within the 3 categories in Table 3, that we did not find in the empirical literature on gamified health apps. While this may partly be because of the limitations of design and methodology of our study, it also points to a space for further research that looks for evidential proof of other issues (and corresponding steps to address them) that one can anticipate based on other literature on games and gamification. For example, there is a possibility that gamified health apps may lead to a trivialization of health as an unintended effect of game elements that try to simplify complex health variables for users of an app. Nguyen ([49] has theorized a similar possibility, arguing that one possible effect of gamified health apps might be a simplification of users’ health goals. For example, a user may get so obsessed with numbers or rewards on their gamified health app that they may lose track of their original goal of being healthy. Similarly, one may possibly observe ethically problematic effects, other than those we found in the empirical literature, because of the use of gamified apps. Zuboff [50], for example, has argued that many digital environments, including health-related apps, commodify user behavior in the interests of the designers of these environments. Our review indicates the possibility for empirical investigation of such a hypothesis in the context of gamified health apps in future work.It should be noted that our attempt in this paper was to locate as many ethical issues related to gamified health apps as we could find in the empirical literature. One limitation to note here is that there are related areas of research, such as research focusing on behavior change technologies in general, which also deal with some similar ethical issues. Future research may seek to broaden the scope of our research to include ethical issues identified from those domains as well. Another related limitation is that some of the ethical issues may need more or stronger evidence, particularly about the extent to which they are universally or even widely operational across gamified health apps. For example, we stated that Maturo and Setifi’s [30] work on gamified health apps, which is based on analysis of the design features of the apps, as well as a discourse analysis of reviews of the apps on various internet forums, notes that such apps may lead to stigmatization of people who may not be able to meet standard definitions of “healthy”. Yet, it is also a possibility that, in practice, user communities (facilitated by forums and groups, for example) may subvert the affordances of such apps, and negate such tendencies of stigmatization. The evidence for such optimism comes from other works on self-tracking (not necessarily gamified) health apps. Sharon & Zandbergen [51], for example, through their study of self-tracking communities, elucidate how theoretically postulated ideas about self-trackers, such as their engagement in a form of “data-fetishism”, are limiting. They assert that instead of being obsessed with narrow notions of objectivity (an idea encapsulated within the data-fetishism charge against self-trackers), self-tracking communities actually attribute meaning to their quantified data in ways that resist such objectivity. Further, self-trackers also use their practice to resist social norms (instead of reinforcing them) as well as invent imaginative ways of using self-tracking as a narrative aid [51]. Their study points to the need for further ethnographic and anthropological studies of self-trackers, as well as users of gamified health apps, to understand how such users and communities of users may resist theoretically anticipated problems with such apps. This is not to say that the theoretically postulated and anticipated sources of ethical issues with gamified health apps are not of value. Even if they are eventually resisted by users of such apps, theoretical critiques may offer themselves as a source of critical reflection on behalf of the users as well as designers of such apps. Such users and designers may then use the analysis offered in these theoretical critiques to find ways of resisting and escaping the anticipated problems. This reiterates the need for the sharing of responsibilities between the various stakeholders related with gamified health apps.This brings us to a final point about the utility of the analysis offered in this paper. Our aim has been to provide a practice-relevant framework to identify different types of designer responsibilities that can address ethical issues in gamification. Further, we also aimed at providing a landscape of various ethical issues related to gamified health apps as found in the empirical literature about such apps. This framework, based on designer responsibility as well as the list of various issues, can be useful for designers, users of gamified health apps, as well as other stakeholders, to anticipate as well as avoid ethical issues in their interaction with such apps. Given the recent evidence suggesting that ethical issues, such as potential psychological harm to app users [9], can lead to app discontinuance, addressing such issues may also serve to improve the long-term user engagement with such products. The designers of such apps, in particular, can use our framework to foresee possible issues as well as plan validation studies to ensure that the apps they design are able to avoid foreseeable ethical problems as well as problems that may arise as unintended and unforeseeable effects of their design features. The designer duties prescribed by our framework also emphasize the need for designers to reflect more broadly on the socio-economic implications of the technologies they seek to introduce and point to the potential utility of seeking more democratized approaches towards technological design.Conceptualization, C.A. and M.R.; methodology, C.A. and M.R.; analysis, C.A. and M.R.; investigation, C.A.; writing—original draft preparation, C.A.; writing—review and editing, C.A. and M.R.; supervision, M.R. All authors have read and agreed to the published version of the manuscript.The funding for the APC came from Information Systems Group, TU Eindhoven.Not Applicable.Not Applicable.The authors thank Elizabeth O’ Neill and Anthonie Meijers for their valuable comments on drafts of this paper. Elizabeth O’ Neill also helped by independently validating the inclusion and exclusion criteria as part of the review protocol.The authors declare no conflict of interest.Conceptual Mapping of Gamification Ethics.Reported Ethical Issues in Gamified Health and Fitness Apps.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11053.txt

@@ -0,0 +1 @@
+Healthy aging is a new challenge for the world. Therefore, health literacy education is a key issue in the current health care field. This research has developed a robot-assisted learning system to explore the possibility of significantly improving health literacy and learning perception through interaction with robots. In particular, this study adopted an experimental design, in which the experiment lasted for 90 min. A total of 60 participants over the age of 50 were randomly assigned to different learning modes. The RobotLS group learned by interacting with robots, while the VideoLS group watched health education videos on a tablet computer. The content dealt with hypertension related issues. This study used the European Health Literacy Survey Questionnaire (HLS-EU-Q16), Health Knowledge Questionnaire, Reduced Instructional Materials Motivation Survey (RIMMS), and Flow Scale as evaluation tools. The result shows no significant difference in the pre-test scores between the two groups. Compared with the video-assisted learning system, the robot-assisted learning system can significantly improve health knowledge, health literacy, learning motivation, and flow perception. According to the findings of this study, a robot-assisted learning system can be introduced in the future into homes and care institutions to enhance the health literacy of the elderly.With the increase of the elderly population and the average life expectancy, healthy aging is a new challenge facing the international community. People gaining better health knowledge and behaviors will help achieve the goal of healthy aging [1]. Health literacy education is one of the feasible solutions [2] and refers to the ability of individuals to obtain, process, understand and apply health information to make the right health decisions and take appropriate health actions [3,4]. Individuals’ inadequate health literacy may result in poor health decisions, behaviors, and outcomes [5,6]. Therefore, many countries regard health literacy education as an important national strategy [7].When people encounter health problems, their most typical impulse would be to look for information through Internet search engines, ask relatives and friends, or watch health channels on TV. However, in general, they may not be able to judge the correctness of the information. Hence, if information and communication technology can play the role of providing correct health information, it will help fill the gap in health education and enable people to obtain adequate and correct health information and services. Furthermore, scholars consider that robots have great potential as a learning technology and can be applied to various types of educational purpose [8,9,10,11]. An educational robot has five basic educational applications, i.e., language education, robotics education, teaching assistance, social skill development, and special education, as well as guided learning through feedback [9]. Robots can serve as peers or mentors of learners and can effectively improve learners’ cognition and affection [12]. In addition, it is worth noting that traditional learning technology lacks the robot’s characteristic of physical presence [13]. The functions and interaction methods of robots are different from those of computers, tablets, and mobile phones. Most robots use facial expressions, body movements, and natural language to interact with users. Therefore, personalized learning content and methods can be designed in the robot to provide learners with a customized learning experience.As elderly people spend longer at home, they may lack comprehensive and friendly health education channels. Meanwhile, it is possible to receive false health food advertisements on TV programs, causing great damage to their bodies. If there is a mechanism that can effectively improve the health literacy of the elderly, it will help maintain their physical and mental health, thereby reducing the medical burden on the family, society, and the country. Therefore, this study attempts to explore the feasibility and substantial utility of applying robotic technology to health education for the elderly. In particular, a robot-assisted learning system (RobotLS) based on humanoid robots was implemented to provide users with health information. The system interacts with users in a voice-based manner to reduce operational complexity. The findings of this study can contribute to home care and care institutions in improving people’s health literacy and well-being.The health information received by the elderly mainly comes from newspaper advertising, radio, and television. Among these media, video is more effective in delivering knowledge. Therefore, this study uses videos as a benchmark for comparing the effectiveness of health education for the elderly. The ARCS (Attention, Relevance, Confidence, Satisfaction) motivation model proposed by Keller [14] aims to provide a systematic teaching evaluation model. If teachers design content based on the ARCS model, it will help learners engage in learning activities, thereby improving their motivation and effectiveness. Moreover, if the elderly can interact with the robot to generate a flow state, the RobotLS can help them focus and enjoy the interactive learning process. They can immerse themselves in the present and forget the passage of time. Therefore, this study uses robots for health literacy education, which is expected to help the elderly improve their learning motivation and flow state.According to the above research background and purpose, this study proposes the following research question. Compared with the video-assisted learning system (VideoLS), can the robot-assisted learning system (RobotLS) significantly improve the health knowledge, health literacy, learning motivation, and flow perception of the elderly? In order to answer the research question, an experimentation method is used to evaluate the effectiveness and feasibility of robots in health education for the elderly. Chi-square and independent-sample t-tests are used to test whether there are significant differences in the demographic variables of the two groups. The paired sample t test is used to test whether the post-test scores of each group are significantly better than the pre-test scores in terms of health knowledge. Analysis of covariance (ANCOVA) is used to test whether there is a significant difference in the effectiveness of health knowledge learning between the two groups. Regression analysis is used to explore the effects of pre-test scores, learning modes, and demographic variables on post-test scores. Independent sample t-test is also used to analyze the differences in health literacy, learning motivation, and flow perception between the two groups. All analysis procedures are completed using IBM SPSS Statistics 26.Health literacy is a multidimensional concept that involves a series of skills that people need to apply effectively and efficiently in a healthcare environment. Many studies have confirmed that there is a significant direct relationship between health literacy and health behaviors or health outcomes [15]. Insufficient health literacy can lead to poor health behaviors, disease prevention and treatment capabilities [16]. Therefore, health literacy is usually used as an indicator to measure the effectiveness of health education. Health literacy can be distinguished into three dimensions: functional, interactive and critical. Functional health literacy concerns the basic reading and writing skills for understanding and using health information. Interactive health literacy includes the advanced cognitive and literacy skills to be able to interact with healthcare providers, and the ability to interpret and apply information in a constantly changing environment. Critical health literacy is an advanced cognitive skill that can judge and analyze a variety of health information, and effectively apply judgment to improve one’s life [17,18].Sørensen et al. [19] used a systematic review to analyze articles related to health literacy and proposed an integrated conceptual model. The process of people acquiring health literacy can be divided into four steps: access, understand, appraise, and apply [19]. If people can perform the following four procedures, it will help promote their personal health.Access. The ability to find, discover, and obtain health information.Understand. The ability to comprehend the health information obtained.Appraise. The ability to interpret, filter, judge, and evaluate the obtained health information.Apply. The ability to communicate with professional medical staff and use health information correctly to make decisions to promote health.Health literacy is a lifelong learning process. The knowledge acquired at all stages of life will be forgotten over time, and health knowledge may need to be updated. Therefore, people must have good health literacy to learn appropriate knowledge relevant to themselves. Sørensen et al. [19] divided the health issues people face in life into three domains: healthcare, disease prevention, and health promotion. Although the procedures for acquiring health knowledge in each domain are the same, the scope of application is different. Nevertheless, good health literacy helps people maintain and improve their quality of life.Healthcare refers to patients who have fallen ill or are being cared for in a medical institution. It emphasizes the acquisition, understanding, and evaluation of medical or clinical information and the ability to make wise treatment decisions and comply with medical advice.Disease prevention refers to the state of being at risk of disease, emphasizing the retrieval of information on health risk factors, understanding, interpreting and evaluating relevant information, and making smart decisions that help maintain health.Health promotion refers to health education for the general public, including the efforts of communities, workplaces, education systems, political organizations, and the health industry to strengthen people’s health. It emphasizes regular updates of health information regarding the social and physical environment, leading to wise decisions.Education is one of the important ways to improve people’s health knowledge and literacy. This study uses four steps—health information acquisition, understanding, evaluation, and application—as the basic framework for developing the RobotLS to promote learning among the elderly regarding health knowledge and literacy.In response to the trend towards an aging population and the shortage of nursing manpower, robots have gradually been used in home care or medical institutions. Since the elderly may easily feel lonely when living alone at home, manufacturers have also developed various types of companion or service robots. For example, the Robear developed by Riken of Japan can assist in carrying patients [20], and the RoNA developed by Hstar Technologies in the United States is a typical care-based robot that can assist medical practitioners in taking care of patients [20]. Furthermore, the Paro robot has the appearance of a seal and is usually used for life companions. When the user touches Paro, it will give appropriate feedback. The scholars apply Paro to accompany elderly people with dementia [21]. Another robot, Nao, is mainly based on dialogue interaction and body movements, and can be used as objects for users to talk and listen to [22].At present, the application of robots in health education is mostly in special education. Scholars, such as van den Heuvel et al. [23] used the ZORA robot to intervene in the rehabilitation and special education of children with severe physical disabilities. A total of 17 disabled children, aged between 2 to 8 years old and seven professionals participated in the exploratory study over a period of 2.5 months. All participants participated in six activities related to the ZORA robot, including sports exercises, dance exercises, robot control, and cognitive exercises. The study results found that ZORA has greatly contributed to the rehabilitation treatment and cognitive development of children with disabilities, while children experienced a great deal of fun. Scholars consider that the robot is most suitable in three fields—motor, communication, and cognitive skills [19,20,21]. In addition, ZORA also helps stimulate learning motivation, concentration, an active attitude and an improvement in children’s attention. Conti et al. [24] used the Unified Theory of Acceptance and Use of Technology (UTAUT) to explore the important factors in using robots as special teaching tools. The overall results show that subjects have a positive attitude towards the use of robots. Since humanoid robots are often expensive, some people think that applying robots is impractical for educational purposes. However, research suggests that, as long as the benefits exceed the costs, robots still have considerable potential in the field of special education.Although most studies support the benefits of robots in education, some scholars have given warnings about their application in special situations. Alcorn et al. [25] invited 31 autism educators (teachers, teaching assistants, and speaking and speech therapists) to discuss the role of social robots in the education of autistic learners through semi-structured interviews and focus groups. Although almost all interviewees are happy to use humanoid robots in the classroom (e.g., NAO, KASPAR, Milo), the paradox is that these educators worry that time will bring counter-effects. For example, robots can stimulate students’ learning motivation and classroom participation, but they may also hinder students from participating in group activities. Whether students with autism can transfer the knowledge or skills learned from the robot to the real environment, what are the substantial benefits and negative effects, and which functions of the robot can meet the specific needs of these students, all need more evidence and research.From the above discussions, the nursing staff provided positive feedback on the robot used for assisted care. Robot-assisted education is often used in children’s rehabilitation and special health education and is found to improve the learning effectiveness of children and special needs students. However, there is no research on health education using robots for the elderly. Thus, it is unknown whether the application of robots to health education for the elderly can also produce positive effects. This study focuses on the elderly and robot health education to explore the effects of robot-assisted learning for this group.The framework of the health literacy education proposed in this study includes robots, cloud databases, and interactive algorithms. The user’s personal data and physiological information will be recorded in the cloud database, and the teaching materials placed in the cloud will be updated in a timely fashion. The system will provide appropriate health information and health improvement suggestions according to users’ needs. As a robot is mainly used as an interface for human–computer interaction, this study terms this system a robot-assisted learning system (RobotLS).This study has designed two ways of presenting learning materials: a robot-assisted learning system (RobotLS) and a video-assisted learning system (VideoLS). The RobotLS uses a humanoid robot as a learning device to assist the elderly in learning health knowledge through voice and touch interactions, while the VideoLS uses animation videos to deliver learning contents. This study aims to explore the effects of the two learning modes on the elderly’s health knowledge, health literacy, learning motivation, and flow perception.In order to avoid the influence of exogenous variables on the research results, this study uses an experimental research method to explore the impact of different learning modes on the learning outcomes of participants. The experimental research method refers to the researcher’s investigation as to whether there is a causal relationship between the independent variable and the dependent variable under the control of irrelevant variables [26]. Therefore, during the experiment, subjects will be randomly assigned to the experimental group or the control group. Different experimental treatments will be then applied to different groups. The researcher can observe the influence of independent variables on dependent variables and explore its causality. The independent variables of this study are different learning methods, while the dependent variables are health knowledge, health literacy, learning motivation, and flow perception.In this study, a humanoid robot named Robelf, as shown in Figure 1, was selected for system development. This robot has a pleasing appearance and uses a tablet to present facial features. When the robot interacts with the user, it can express happiness, anger, sadness, joy, and boredom. In addition, the robot also has basic functions, such as face recognition, remote control, information provision, real-time reminders, facilities for children’s education, home safety intelligent monitoring, and so on.This study uses Android Studio and Robelf SDK to design a RobotLS. Since the subjects are elderly, the fonts are enlarged, and the color contrast is enhanced when designing the user interface. The learning content is mainly presented with keywords and images, and is explained by voice. The interactive design will slow down the speech speed, amplify the volume, and repeat reminders of key content [27]. Scholars [28,29] have pointed out that the behavior of the robot can make the user produce corresponding emotions and improve human–computer interaction. Body movements and hand postures can increase the robot’s popularity, anthropomorphism, and social participation. Facial expressions can improve users’ understanding of the interaction context and trust in the robot. Therefore, the RobotLS designed by this study can display facial expressions and body movements based on content and interaction status to enhance the sense of participation and identity from the elderly.Many elderly people suffer from at least one chronic disease. The main types of common chronic diseases are cardiovascular diseases, such as hypertension, diabetes, hyperlipidemia, and heart disease. Among them, high blood pressure is an invisible killer. The control of high blood pressure is closely related to daily habits, exercise status, and diet. Therefore, this study uses hypertension prevention as the content of the experimental course. Consequently, the elderly, regardless of whether they have been diagnosed with hypertension or not, can learn about hypertension.The experimental contents of this study are designed based on the hypertension prevention and treatment manuals published by government health agencies. It has two topics: knowledge of hypertension prevention and treatment, and diet control for hypertension. The related knowledge of hypertension prevention and treatment includes five units: introduction of blood pressure, introduction of hypertension, blood pressure monitor, medication, and lifestyle improvement. The diet control for hypertension includes three units: dietary style, diet seasoning, and diet choice. All learning materials have been confirmed by an attending physician in the family medicine department and two nurses to confirm its correctness. Experts in these fields have more than 10 years of clinical qualifications. Before the experiment, participants’ blood pressures were measured, and the experimental system gives the participants corresponding content based on their respective blood pressure statuses.The RobotLS will follow the process of acquiring health literacy for interactive teaching in each unit. In the access stage, the robot actively provides users with health information based on personal blood pressure. In the understand stage, the robot will explain and use pictures, tables, animations, videos, and multimedia presentations to help users further understand the content. In the appraise stage, the robot will use the interactive method of question-and-answer to evaluate the user’s understanding of information and judgment of health events. Meanwhile, in the apply stage, the robot will ask questions on real-life situations to determine whether participants can apply health knowledge about their daily lives.The content of the VideoLS is the same as that of the RobotLS to avoid the influence of content differences on the research results. Therefore, the production method of the VideoLS is to remove the robot’s question-and-answer and physical interaction capabilities. The content remains unchanged and retains the function of voice output. When learning, users can control the speed and progress of the presentation of the content to meet their learning situation. The VideoLS is shown in Figure 2.There are 20 questions in the health knowledge test, with a full score of 100 points. Among them, 10 questions are on hypertension-related knowledge, and the other 10 questions are on precautions for hypertension diet. The questions are all true-false items, and the participants will need to mark “correct”, “wrong”, or “don’t know” based on the description. This study asked three clinical medical experts to confirm the appropriateness and correctness of these questions.In order to accurately measure people’s comprehensive health literacy, Sørensen et al. [30] designed and developed the European Health Literacy Survey Questionnaire (HLS-EU-Q). Since 47 items are excessively time-consuming, scholars have simplified the HLS-EU-Q47 into 16, 12, and 6 item versions, which are convenient for testing under the condition that the reliability and validity can be ensured [31]. This study uses the HLS-EU-Q16 version to measure the health literacy of the participants. HLS-EU-Q16 has a total of 16 questions, evaluating health literacy in health care (seven items), disease prevention (five items), and health promotion (four items). The scale uses a four-point Likert scale, ranging from “1 = very difficult” to “4 = quite easy”. According to the scoring formula proposed by Sørensen et al. [31], the health literacy score ranges from 0 to 50 points. Limited health literacy is 0–33 points; sufficient health literacy is 33–42 points; and excellent health literacy is 42–50 points.Compared with the traditional teacher-oriented mode, the appropriate use of robots for teaching aids in the teaching process can help learners improve their learning interest, attitude, motivation, and effectiveness [32,33]. The four steps of the ARCS motivation model [14] include:Attention. The course design can arouse and maintain students’ curiosity and interest in the learning content.Relevance. The course design must meet students’ personal learning goals to promote positive learning attitudes.Confidence. The course design can help students construct self-confidence and complete learning tasks.Satisfaction. When students successfully achieve the preceding three goals, they can obtain internal and external encouragement and feedback, motivating them to continue learning.Attention. The course design can arouse and maintain students’ curiosity and interest in the learning content.Relevance. The course design must meet students’ personal learning goals to promote positive learning attitudes.Confidence. The course design can help students construct self-confidence and complete learning tasks.Satisfaction. When students successfully achieve the preceding three goals, they can obtain internal and external encouragement and feedback, motivating them to continue learning.The assessment of learning motivation uses the Reduced Instructional Materials Motivation Survey (RIMMS) developed by Loorbach et al. [34]. This scale is a condensed version of the motivation scale developed by Keller [35] from 36 to 12 questions. The test result reliability test of the overall questionnaire is better than the original scale. The purpose of using the RIMMS is to assess the influence of the learning mode on the learning motivation of the elderly. The scale is measured using a five-point Likert scale, from “1 = strongly disagree” to “5 = strongly agree”.Csikszentmihalyi [36] proposed the concept of flow, which means that when people are engaged in an activity, they are fully immersed and attentive, and enjoy their current state of mind. Flow state can be interpreted as the best type of activity experience. In the flow state, people will be completely immersed in the current situation and ignore the perception of the outside world. Challenges and skills are two important factors that affect the perception of flow [37]. When challenges and skills reach a balance, a state of flow will occur. However, if people lack skills in facing great challenges, they will feel anxious about the current situation. In addition, when people have excellent skills but not enough challenges, they may get bored.Webster et al. [38] pointed out that the higher the level of flow, the more autonomously they can participate in current activities. For different learners, learning tasks of appropriate difficulty must be designed to help learners immerse themselves in the learning process. The flow state can be evaluated using four dimensions, namely control, attention focus, curiosity, and intrinsic interest [36,38,39,40].Control. Individuals have the ability to control the current situation or properly handle the operation of equipment.Attention Focus. The individual’s attention is focused on the current situation, showing a state of non-distraction. The focus of the flow state emphasizes the concentration of the interaction between the user and the device.Curiosity. The individual feels that the task being performed is novel and therefore has the intention to explore.Intrinsic interest. The individual feels fun in the current state and enjoys the process.This study uses the flow scale developed by Webster et al. [38] to evaluate the flow perception of the elderly when using RobotLS or VideoLS. This scale has a total of 12 items, using a five-point Likert scale, ranging from “1 = strongly disagree” to “5 = strongly agree”.The participants of this study are healthy or sub-healthy elderly people who are over 50 years old and have retired. They have the ability to live independently and do not need the care of others. They can also listen, speak, read, and write Chinese. Participants were randomly assigned to the RobotLS and VideoLS groups. To avoid being overloaded, this study takes the critical issue of hypertension for the elderly as the health education content.71 seniors were recruited to participate in the experiment—36 in the RobotLS group and 35 in the VideoLS group. For the RobotLS group, four elders thought that the interaction time with the robot was too long and interrupted the experiment, while two elders thought there were too many questions in the questionnaire and were unwilling to answer all of them. Among the participants in the VideoLS group, two elders interrupted the experiment, without the patience to watch the full video, two elders did not complete the questionnaire, and one elder had poor Chinese communication skills and could not understand the content of the video. Finally, there were 30 people in each group to complete the experiment.Before the experiment, participants signed informed consent for the study and their blood pressure was measured and recorded. Then, they supplied personal information and finished a pre-test on health knowledge. After the experiment, a post-test of health knowledge was carried out, and the health literacy, learning motivation, and flow perception scales were filled in. In order to help participants to use the learning systems smoothly, this study instructed them in detail on how to use robots and tablets before conducting the experiment, including volume adjustment, progress control, functional operations, and interactive methods.The intervention of the RobotLS group is designed to allow participants to interact with Robelf humanoid robots. The robot guides participants to learn about hypertension prevention and proper diet. Before the formal experiment, there were 10 minutes of robot operation instructions and interactive exercises so that participants could master the timing and method of answering the robot’s questions. Participants can decide whether to take a break at the end of each topic to avoid being overworked. The intervention of the VideoLS group was conducted by watching videos on the same health content through a 10-inch tablet computer. The system has play, pause, playback, and speed adjustment functions, allowing participants to control their progress in watching the video independently. The total experiment time was about 90 min.The participant’s personal information collected in this study includes gender, age, education level, presence or absence of hypertension, exercise frequency (hours/week), and health class attendance frequency (times/week), as shown in Table 1. A chi-square test analysis showed no significant differences in categories, such as gender, education level, and hypertension. In addition, the independent sample t-rest showed no significant difference in continuous variables, such as age, exercise frequency, and health class attendance frequency. The result shows that this study has indeed achieved the purpose of random assignment.As shown in Table 2, the mean of the pre-test on health knowledge of the RobotLS group was 55.50 (SD = 10.53), while that of the VideoLS group was 55.83 (SD = 10.51). After participating in experiments with different learning modes, the mean of the post-test of the RobotLS group was 85.33 (SD = 9.91), while that of the VideoLS group was 75.50 (SD = 11.32). The paired sample t-test results showed that the health knowledge of the two groups was significantly improved. In other words, regardless of which learning mode was adopted, participants can effectively learn from well-planned and designed learning materials.Since the participants’ health knowledge was inconsistent before the experiment, this study used an analysis of covariance (ANCOVA) to evaluate the impact of the learning mode on post-test health knowledge in order to exclude the influence of prior knowledge on learning effectiveness. In this study, the pre-test health knowledge was used as a covariate. Although the pre-test scores of the participants had a significant impact on the learning, there was still a significant difference in the post-test health knowledge between the two groups (F = 19.423, p < 0.001) as shown in Table 3. This study verifies that the learning effectiveness of the RobotLS group was significantly better than that of the VideoLS group.In order to understand whether demographic variables will also have a significant impact on learning effectiveness, regression analysis was used to explore the effects of pre-test scores, learning modes, and demographic variables on post-test scores. The analysis results showed that the variables that have a significant impact on post-test scores are learning modes and pre-test scores, as shown in Table 4. This finding is consistent with the result of ANCOVA analysis. That is, the participants’ prior knowledge of hypertension and the way they interact with the learning materials have significant influence on the participants’ learning effectiveness.In this study, the HLS-EU-Q16 scale [41] was used to evaluate participants’ health literacy. Health literacy is composed of three constructs: health care, disease prevention, and health promotion. The Cronbach’s α are 0.884, 0.872, and 0.887, respectively, which shows that the health literacy scale used in this study has a high degree of internal consistency.Table 5 shows the health literacy scores of the participants measured after the end of the experiment. The mean of the health literacy in the RobotLS group was at a good level, while that in the VideoLS group was at a sufficient level. In all the health literacy constructs, the RobotLS group is significantly better than the VideoLS group. The results show that the participants’ health literacy can reach an ideal level through an appropriate learning process. Moreover, the learning performance will be more significantly improved with the support of human–robot interaction design.This study used the RIMMS [34] to measure participants’ learning motivation. The scale comprises four dimensions: attention, relevance, confidence, and satisfaction. The Cronbach’s α are 0.924, 0.905, 0.912, and 0.919, respectively, indicating the questionnaire’s good internal consistency reliability.According to the statistical analysis in Table 6, the means of the RobotLS group in the four dimensions of learning motivation are significantly better than those of the VideoLS group. This shows that the involvement of robots in health education helps arouse users’ motivation for learning.The flow scale is composed of four dimensions: control, attention focus, curiosity, and intrinsic interest [38]. The Cronbach’s α are 0.874, 0.858, 0.935, and 0.893, respectively, indicating the questionnaire’s good internal consistency reliability.Table 7 shows the mean, standard deviation, and test results of the two groups in terms of flow perception. The mean of each dimension of the RobotLS group is significantly better than that of the VideoLS group. This shows that when users use robots to learn, they can immerse themselves in the learning situation and arouse their curiosity and intrinsic interest in participating in the course.The results show that the health knowledge of the RobotLS group is significantly better than that of the VideoLS group. In the process of human aging, memory tends to gradually decline with age. The elderly may also be unable to concentrate for a long time, which may reduce the effectiveness of learning new content. Their memory can be improved and maintained through repetitive exercises and the use of appropriate strategies. Liu et al. [42] pointed out that asking the elderly to answer relevant questions after learning and then providing appropriate materials based on the correctness of the answers will help improve their learning effectiveness. The RobotLS developed in this study has a question-and-answer function, which can immediately assist in reviewing correct knowledge when the elderly make incorrect answers. The system will increase the interest and pleasure of learning for the elderly through appropriate interaction, rather than just the transfer of knowledge.Humanoid robots can make users feel their physical and social presence [13,43]. The robot used in this study has a human-shaped structure and is capable of simple dialogue and physical interaction with the user. The robot is given an identity symbol. It appears as a cute boy at the age of 8, and his name is Xiaobei. When conducting the experiment, Xiaobei would introduce himself and greet the elders, just like a young grandson naturally sharing health knowledge with his grandparents. The research finding confirms that a good human–robot interaction process is indeed helpful for the acquiring of health knowledge. Some studies have found that when people get along with a specific doll or robot for a certain period, they will regard it as a listener or companion [44,45]. Additionally, they will also have emotional engagement. This phenomenon is especially obvious for preschool children and elderly people because they lack the stimulation of new events. Therefore, compared with traditional learning methods, the RobotLS can improve the learning performance of the elderly more effectively.The result shows that the health literacy of the RobotLS group was significantly higher than that of the VideoLS group. The average scores of the RobotLS group fall within the standard of good health literacy (42–50 points), and those of the VideoLS group falls within the interval of sufficient health literacy (33–42 points). A high health literacy score means that the user has a more precise concept of access, understanding, evaluation, and application of health information. The most common health problems faced by the elderly are whether or not to believe in drug or health food advertisements, dealing the health experience of relatives or friends, and online health information. The robot in this study is only used as a front-end device for interacting with users and presenting health materials. The complete RobotLS also has a back-end cloud database that records personal health status, medication usage, medical needs and other information. The system can provide appropriate and correct health information according to the user’s physical condition to prevent the elderly from misbelieving the wrong health information. The RobotLS in this study considers the characteristics of the elderly’s health information needs and is designed and developed based on four important procedures for health literacy. This is also one of the reasons why the health literacy of the RobotLS group is better than that of the VideoLS group.The results show that the RobotLS group’s learning motivation in the four dimensions—attention, relevance, confidence, and satisfaction—is significantly better than that of the VideoLS group. The anthropomorphic characteristics and interactive design of robots are the main factors affecting learning motivation. Notably, similar findings have been found in previous research. Liew et al. [46] pointed out that the enthusiastic teaching model of virtual teachers can significantly enhance learning motivation and students’ willingness to use multimedia materials. Hsieh [33] considers that, compared with the traditional teacher-centered model, adding robot-assisted teaching can effectively increase the motivation of students to continue learning.Previous studies have pointed out that when learners feel a high degree of flow in the learning process, it will help improve learning performance [40]. The results show that the RobotLS group’s control, attention focus, curiosity, and inner interest in the flow perception are also significantly better than the VideoLS group. Skadberg and Kimmel [47] pointed out that attractiveness is an important factor in the flow state, while robots also have the characteristics of attracting the attention of the elderly. The RobotLS has anthropomorphic characteristics and interacts with participants with voice, facial expressions, lights, and body movements. Webster et al. [38] indicate that people tend to engage in activities that make them happy. Therefore, the higher the degree of flow perception, the more one can encourage learners to participate in the learning process. This is also one of the key factors in improving learning performance and health literacy.In constructs of learning motivation and flow perception, the significant level of confidence and control is smaller than that of other variables. Confidence is a measure of the participants’ confidence in operating the device to complete the learning task, while control is a measure of the participant’s ability to actually operate the learning device. For participants, RobotLS is definitely a relatively novel learning system. Although participants were asked to practice RobotLS before the experiment, they sometimes still needed the help of researchers in actual operation. Therefore, the means of confidence and control in the RobotLS group are the lowest scores in the construct of learning motivation and flow perception, respectively. Since participants may use mobile phones or computers to access information and engage in social activities, they may not find it difficult to use a tablet for learning. For the participants in the VideoLS group, the mean of the confidence variable is higher than the other three variables in the learning motivation construct, while the mean of the control variable is higher than the attention focus and curiosity variables in the flow perception construct. Due to the differences in the participants’ familiarity with learning equipment, the two groups are less significant in the two variables of confidence and control. Based on the findings of this research, we suggest that future research should give older people longer practice time to adapt to the use of new learning technologies.In the past, the application of robots in education mainly focused on formal school education and special education. While people need more health knowledge as they grow older, such demands are often ignored. This study integrates the four core procedures of health literacy into the design of the RobotLS to deliver health knowledge. This study verifies the feasibility and effectiveness of this learning method with experimental research. The result shows that, compared with traditional video-based health education, the RobotLS helps improve the health knowledge, health literacy, learning motivation, and flow perception of the elderly.There are some research limitations that must be considered when applying the results. Participants in this study are members of senior-age institutions. Because the members of these institutions are mostly women, female participants are obviously higher than male participants, thus showing that older women are more willing to participate in community group activities and leisure courses. Future research can invite male subjects extensively to understand how male elderly people feel about using the RobotLS. Moreover, it is difficult recruit a large number of elderly participants. Although the total sample is 60, each test has reached a significant level, indicating the feasibility of using robots for health literacy education. In the future, researchers can expand the number of samples on the basis of our research to increase the power of the test.At present, the robot can only speak Chinese and recognize Chinese pronunciation, and there is no corpus built for other national languages and local dialects. The elderly who can only communicate in local dialects are more disadvantaged in obtaining correct health information. It is hoped that there will be a complete language package in the future so that the RobotLS can reach more users and exert greater utility.This study is limited by the amount of experimental equipment and the difficulty encountered in long-term experiments. It also takes into account the physical strength, endurance, and concentration of the elderly. Therefore, the experiment time was set to 90 min. Although it is hard to ask participant to repeat experimental interventions, the research results still show that the RobotLS can achieve significant learning performance. If it can be implemented for a longer period, it is expected to continuously improve health knowledge and literacy.Presently, the RobotLS takes the four procedures in obtaining health literacy as the main design. In the future, we can further discuss the design methods and application effects of robots in the three major areas, namely health care, disease prevention, and health promotion.All of the authors (C.-W.W., H.-Y.K., W.-H.W., C.-Y.C. and H.-P.F.) contributed to conceptualization, methodology, system design, experiment, data collection and analysis, writing, and editing. All authors have read and agreed to the published version of the manuscript.This research was funded by Ministry of Science and Technology of Taiwan, grant number MOST 109-2511-H-037-008.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Kaohsiung Medical University Chung-Ho Memorial Hospital (protocol code: KMUHIRB-E(I)-20200390 and date of approval: 5 January 2021).Informed consent was obtained from all subjects involved in the study.Data are available upon reasonable request by contacting cwwei@kmu.edu.tw. The data are not publicly available due to privacy concerns.The authors would like to thank all of the study participants.The authors declare no conflict of interest.The robot-assisted learning system.The video-assisted learning system.Results of demographic variables.Paired sample t-test of pre- and post-test regarding health knowledge.The result of ANCOVA on health knowledge. Dependent variable: post-test on health knowledge.a R Squared = 0.448 (Adjusted R Squared = 0.429).The result of regression analysis on health knowledge. Dependent variable: post-test on health knowledge.a R Squared = 0.570 (Adjusted R Squared = 0.503).Independent samples t-test on health literacy.Independent samples t-test on learning motivation.Independent samples t-test on flow perception.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11054.txt

@@ -0,0 +1,10 @@
+Soil erosion is a serious ecological problem in the fragile ecological environment of the Tibetan Plateau (TP). Rainfall erosivity is one of the most important factors controlling soil erosion and is associated with the El Niño southern oscillation (ENSO). However, there is a lack of studies related to the spatial distribution and temporal trends of rainfall erosivity on the TP as a whole. Additionally, the understanding of the general influence of ENSO on rainfall erosivity across the TP remains to be developed. In this study, long-term (1971–2020) daily precipitation data from 91 meteorological stations were selected to calculate rainfall erosivity. The analysis combines co-kriging interpolation, Sen’s slope estimator, and the Mann–Kendall trend test to investigate the spatiotemporal patten of rainfall erosivity across the TP. The Oceanic Niño Index (ONI) and multivariate ENSO Index (MEI) were chosen as ENSO phenomenon characterization indices, and the relationship between ENSO and rainfall erosivity was explored by employing a continuous wavelet transform. The results showed that an increasing trend in annual rainfall erosivity was detected on the TP from 1971 to 2020. The seasonal and monthly rainfall erosivity was highly uneven, with the summer erosivity accounting for 60.36%. The heterogeneous spatial distribution of rainfall erosivity was observed with an increasing trend from southeast to northwest. At the regional level, rainfall erosivity in the southeastern TP was mainly featured by a slow increase, while in the northwest was more destabilizing and mostly showed no significant trend. The rainfall erosivity on the whole TP was relatively high during non-ENSO periods and relatively low during El Niño/La Niña periods. It is worth noting that rainfall erosivity in the northwest TP appears to be more serious during the La Niña event. Furthermore, there were obvious resonance cycles between the rainfall erosivity and ENSO in different regions of the plateau, but the cycles had pronounced discrepancies in the occurrence time, direction of action and intensity. These findings contribute to providing references for soil erosion control on the TP and the formulation of future soil conservation strategies.Soil erosion has already emerged as one of the most serious ecological and environmental problems globally, which not only threatens terrestrial ecosystems, but also severely restricts the security of human existence and the sustainable development of economy and society [1,2]. Soil erosion not only contributes to land degradation, but even interferes with the ability of the soil carbon cycle to mitigate the greenhouse effect [3,4]. Soil erosion by water is considered to be one of the most detrimental types of soil erosion, causing a loss of soil nutrients, which reduces crop yields, pollutes water quality, contributes to the sedimentation of rivers, and raises flooding [5,6,7,8,9]. Therefore, the accurate prediction of water erosion is of great significance for the comprehensive management of soil erosion and effective soil protection.The causes of water erosion are related to a series of natural factors involving rainfall, soil, topography, vegetation, and other human factors such as land use and crop cultivation management [10,11]. In particular, rainfall is the principal climatic factor responsible for water erosion, which influences water erosion through the duration, amount, and intensity of rainfall events [12]. The principal predictive tools for water erosion are the Universal Soil Loss Equation (USLE) and the Revised Universal Soil Loss Equation (RUSLE), which have been applied worldwide, and rainfall erosivity (R-factor), one of the key input parameters in the model, is the potential capacity of rainfall to induce water erosion [12,13]. The R-factor is defined as the product of the rainfall energy and the maximum rainfall intensity in a 30 min period (EI30), and the calculation requires the use of consecutive rainfall data series with a temporal resolution of at least 15 min, which is however, hardly available in many countries and regions. Even if an adequate rain gauge data can be accessed, the complicated calculation process is time-consuming and laborious, which dramatically restricts model promotion and implementation [14].In this context, as alternative algorithms based on the relationship between R-factor and available rainfall data were developed, including the calculation of rainfall erosivity based on annual [15,16], monthly [17,18], and daily [19,20] rainfall data from meteorological stations or satellite radar [21,22]. Among these, daily rainfall data are widely used due to their relative accessibility, which provides more characteristic information of rainfall and facilitates the precision and reliability of R-factor estimation [23]. In the daily rainfall data model, the rainfall erosivity algorithm was divided into linear exponential, logarithmic, and power functions to fit the relationship between rainfall and rainfall erosivity, and the models are mostly combined experimental and empirical based [24,25,26]. In general, these models perform an important role in the quantitative evaluation of rainfall erosivity, and provide scientific reference for the forming mechanism of water erosion, evolutionary process and even the mechanics of climate change.It has become an indisputable fact that the global climate is changing remarkably, with extreme weather events growing stronger, more frequent, and lasting longer [27]. El Niño southern oscillation (ENSO) is the most intense sea-air interaction event affecting the global climate, and although it usually occurs in the eastern equatorial Pacific region, it can be responsible for rainfall anomalies spreading globally [28]. The ENSO cycle has a pronounced periodic character as a result of the interaction between the ocean and the atmosphere, with El Niño (warm phase) and La Niña (cold phase) as the two extreme phases of the ENSO cycle. Considerable work has been conducted on the relationship between ENSO and precipitation events, anomalous temperature, wet and dry variability, and atmospheric circulation [29,30,31,32]. The studies also pointed out that El Niño and La Niña showed diverse rainfall patterns, for example, compared with the La Niña period, northern China is more arid during El Niño in the northern hemisphere, while rainfall in the southeast of China appears to increase substantially, while the contrary phenomenon is present in the southern hemisphere [33,34]. Although these studies have enhanced our comprehension of atmospheric tele-correlation model (ENSO) effects on rainfall, currently the effect of ENSO on rainfall erosivity is still only shown in a few studies [35,36,37,38,39]. A significant dependence between rainfall erosivity and the ENSO indices has been observed in eastern China [35,37], northeastern Spain [40], and the southwestern United States [41], while studies on how ENSO affects rainfall erosivity on the TP are still unknown. The Tibetan Plateau (TP) is the largest and highest geographical unit in the world, with an average altitude of over 4000 m, and is called the Earth’s “third pole”. It is of extreme importance to regional economic development and ecological security, as well as global climate, water resources, and ecosystem functioning [42]. Since the 21st century, however, drastic environmental changes have been remarkably observed on the TP [43]. These changes have become key drivers of increased soil erosion risk. Studies have demonstrated that grassland ecosystems on the TP are suffering from severe degradation due to the combined effects of climate change and human activities. This in turn has triggered a decline in biomass, biodiversity, and landscape complexity, fragmentation or complete loss of services such as soil and water conservation, and an increase in rainfall erosivity and sandstorms [44,45,46,47]. Permafrost degradation can reduce the stability of soil aggregates and the water content in the surface soil is abnormally high during the thawing stage, thus shortening the time of runoff generation and exacerbating erosion caused by rainfall [48,49]. The glaciers’ retreat and the rise of the snow line in cold areas at high altitudes have changed the surface albedo and atmospheric heat circulation and thus have affected the local rainfall intensity, and the form of erosion caused by glacial meltwater and snowmelt runoff generated is one of the main reasons for increased erosion [50,51]. Additionally, according to observations and climatological models, the TP has suffered a faster rate of warming since the 1960s, which is three times the global average [52]. Notable changes in the plateau climate system, such as short periods of intense rainfall triggered by extreme precipitation events, may have led to an increasing trend in the rainfall erosivity on the TP [51]. Some studies have analyzed the variation in rainfall erosivity in the catchment and local scales of the TP, indicating an increasing trend of rainfall erosivity [53,54,55]. These studies provided useful information on the variation in rainfall erosivity, but a further analysis is necessary for the TP as a whole.Soil erosion is serious on the TP, with 70% of the area suffering from varying degrees of soil erosion [56]. In both sides of the Yarlung Tsangpo River and the South Qiangtang area, gully erosion is widely distributed, while in the interior of the plateau scale erosion becomes the main type of erosion in grasslands [57]. Soil erosion on the TP has caused irreparable soil degradation and land area reduction, and is leading to the sedimentation of downstream rivers, landslides, mudslides and other disasters, posing a threat to transportation, agriculture, and animal husbandry. Moreover, soil conservation is particularly important in the TP due to its harsh physical environment, widespread permafrost and fragile alpine ecosystems making it the most sensitive and fragile region [58]. Once erosion happens, its rehabilitation process is prolonged and difficult.Detection of long-term trends in rainfall erosivity can provide information regarding the potential impact of rainfall changes on soil erosion. It is particularly useful for the TP region, which is more sensitive to water erosion and climate change because of the fragile biophysical conditions [59]. However, these unique geographical features and complicated terrain have restricted soil erosion studies due to the scarce observational data on precipitation and soil erosion. Previous studies have focused on local watersheds or small areas of the TP, while the spatial and temporal characteristics of how rainfall erosivity vary over the entire TP have not been adequately studied [60,61]. Moreover, periodic factors lead to ‘poverty years’ and ‘abundant years’ of precipitation in the highlands in different years. The interannual variation in precipitation erosivity on the TP may be the result of ENSO action, but the general effect of ENSO on rainfall erosivity in the TP is not clear at present, and it is necessary to expand the related understanding.In view of this, the TP as a whole was chosen as the study area, and daily rainfall data from 91 meteorological stations were selected to calculate the rainfall erosivity and to explore its relationship with ENSO. The objectives of the study are as follows: (1) to characterize the temporal trends of rainfall erosivity during 1971–2020 across the entire TP; (2) to present the spatial distribution of rainfall erosivity on the TP; (3) and to investigate the impacts of ENSO on rainfall erosivity in different regions of the TP.This study was based on a single case study of the TP. In this section, the basic information about the study area, the required data handling process and the methods related to rainfall erosivity were described in detail.The Tibetan Plateau is located in the southwestern part of China, with an area of 2.74 × 106 km2 and an average altitude of over 4000 m. It is known as the “roof of the world” [62]. It is included in the Tibet Autonomous Region and Qinghai Province, and the southern part of the Xinjiang Uygur Autonomous Region, the western part of Gansu Province, the western part of Sichuan Province and the northern part of Yunnan Province. The main mountain ranges are the Kunlun Mountains, Qilian Mountains, Karakorum Mountains, Himalayas, and Hengduan Mountains. The climate ranges from a humid monsoon climate in the southeast to an alpine arid plateau climate in the northwest, controlled by the Pacific monsoon, Indian monsoon, and prevailing westerly winds, and is influenced by the mountain terrain [63]. Diverse climate types form subtropical rainforests, shrubs, alpine meadows, alpine grasslands, and alpine desert vegetation types are present. TP precipitation exhibits a distinct gradient, gradually decreasing from more than 1000 mm in the southeast to less than 50 mm in the northwest [64]. The region has experienced soil erosion, desertification and landslide hazard [51,65]. Referring to [66], the criteria for the physical geographic zoning of the TP divided the plateau into Region I (arid zone) and Region II (humid zone) (Figure 1).The observed daily precipitation data used in this study was obtained from the Climatic Data Center, National Meteorological Information Center of the China Meteorological Administration (CMA) (http://data.cma.cn (accessed on 10 October 2021)). The data included a total of 91 meteorological stations (Figure 1), with complete data series, covering the time period 1971 to 2020 (Table S1). Moreover, considering the continuity of spatial interpolation and the stations spreading over the entire TP as much as possible, 27 meteorological stations around the study area were selected with the criterion that the shortest linear distance from a meteorological station to the TP boundary is not greater than 100 km. The observation records of all surrounding stations were recorded at the same time as the study period. In order to ensure data reliability and continuity, each meteorological data record was evaluated by the National Meteorological Center [67].ENSO is a phenomenon of irregular periodic changes in sea surface temperature and wind occurring in the equatorial eastern Pacific Ocean, one of the strongest natural signals of interannual climate change worldwide. The typical characteristics of ENSO events are commonly known to be anomalous SSTs (±0.5 °C) in the eastern Pacific Ocean for more than 5 months, where warm episodes are El Niño events and cold episodes are La Niña events [68,69]. The multivariate ENSO Index (MEI) was obtained as the first non-rotating principal component (PC) of the six variables (sea-level pressure, zonal and meridional components of the surface wind, sea surface temperature, surface air temperature, total cloudiness fraction of the sky) over the tropical Pacific [70,71]. it is considered as a better index for detecting the ENSO phenomena with respect to other indices because it takes into account more information and fewer data failures [37]. Therefore, in this study, the occurrence and duration of the El Niño event and La Niña event were determined based on the Oceanic Niño Index (ONI), and MEI was selected as the ENSO proxy to probe the relationship between rainfall erosivity and ENSO during the time period of 1971–2020. These indexes are obtained from the National Oceanic and Atmospheric Administration (NOAA). Specifically, ONI was acquired from NOAA Climate Prediction Center [72], and MEI was acquired from NOAA Earth System Research Laboratory.The study was divided into four steps (Figure 2):Step 1: Estimation of rainfall erosivity at different time scales. Based on the daily rainfall data of 91 stations, the annual, seasonal, and monthly average rainfall erosivity of TP from 1971 to 2020 were calculated using the daily rainfall erosivity model.Step 2: Trend analysis and spatial distribution of rainfall erosivity in TP. Firstly, the temporal changes trend of rainfall erosivity was evaluated by using Sen’s slope estimation and the MK trend test, and then the co-kriging method was used for the spatial mapping of rainfall erosivity for the period of 1971–2020. Step 3: Pattern identification and analysis of rainfall erosivity change at each meteorological station. Firstly, the change trend of rainfall erosivity at each station was classified by integrating multiple indicators. Then, the coefficient of variation (CV) and the seasonal spatial distribution of rainfall erosivity of each site was analyzed.Step 4: Relationship between rainfall erosivity and ENSO. Based on ONI, the variation in monthly mean rainfall erosivity during El Niño and La Niña events from 1971 to 2020 was analyzed; based on the MEI index, a continuous wavelet transform analysis method was used to examine the influence of ENSO on rainfall erosivity, and to clarify the response of the resonance period in different regions of the TP.The half-monthly rainfall erosivity was estimated for each of the 91 meteorological stations from 1971 to 2020 using the daily rainfall erosivity model. The agent model was originally built from Richardson’s equation [73], and later improved by Zhang [20]. Previous studies have demonstrated that this method is reliable and has been widely used on the national and regional scales in China [36,74,75,76]. This method is based on daily rainfall data to obtain monthly, seasonal, and annual rainfall erosivity. The calculation procedures are as follows:(1)Ri=α∑j=1kPjβ
+          (2)α=21.586β−7.1981
+          (3)β=0.8363+18.144Pd12+24.455Py12
+          where Ri is the rainfall erosivity in the i-th half-month period (MJ mm ha−1 h−1), k is the number of days in the half-month period, and Pj is the daily erosive rainfall amount (mm) on the j-th day during the half-month period. The half-month interval method is as follows: with the fifteenth day of each month as the dividing point, the whole year is divided into 24 half months. The half-month period as a basic statistical unit is used to calculate the corresponding half-month rainfall erosivity. According to the national rainfall and runoff analysis, ≥12 mm is defined as erosive rainfall [77]. Therefore, the daily rainfall ≥12 mm is applied to Formula (1), otherwise, it is regarded as value of 0 in the calculations.The terms α and β are two parameters to be determined in the model. Pd12 is the average daily erosive rainfall amount (mm) and Py12 is the average annual erosive rainfall amount (mm). In this study, Formulas (1)–(3) are used to calculate the half-month rainfall erosivity of each meteorological station. The annual and seasonal rainfall erosivity of is the cumulative value of rainfall erosivity in every half-month period.In this study, the trends magnitude of annual rainfall erosivity was estimated with the non-parametric Sen’s method. The trends and significance of annual and seasonal (monthly) rainfall erosivity were detected with the non-parametric Mann–Kendall test. Sen’s slope estimation is a non-parametric method of slope calculation, which is commonly used in the trend analysis due to its high robustness and computational efficiency [78]. The determination for the slope of annual rainfall erosivity is as follows: first, the values of Qi calculated by the Formula (4) are ranked in order of magnitude, and then determines the overall estimator (SLOPEmed) as the median of these Qi by Formula (5).The slope in the N pairs of samples is calculated as follows:(4)Qi=xj−xkj−k i=1,…,N
+          where xj and xk are values of the rainfall erosivity corresponding to periods j and k, respectively (j > k). SLOPEmed is calculated according to the following formula:(5)SLOPEmed=QN+12<0if N is odd QN+12+QN+122if N is even
+          where SLOPEmed > 0 indicates an upward trend, and vice versa. Its value indicates the magnitude of the trend change.The non-parametric Mann–Kendall test is a widely used technique for to assess the significance of trends in long time series [79,80]. It is distribution free and not affected by missing values and outliers, and is highly recommended by the World Meteorological Organization [81]. This method is primarily based on two parameters, S and Z, to determine whether a time series has a significant trend. The intermediate variable S is computed as:(6)S=∑k=1n−1∑j=k+1n sgnxj−xk
+          where n is the length of the time series, xj and xk are values of the rainfall erosivity corresponding to periods j and k, respectively (j > k). S is the summation of sgnxj−xk, which takes the value of −1, 0, or 1 when xj−xk is less than, equal to, or greater than 0, respectively. The variance of S can be acquired as follows:(7)varS=nn−12n+518Then the normalized statistical value Z is denoted as follows:(8)Z=S−1varSif S>00if S=0S+1varSif S<0
+          where a positive (negative) value of Z indicates an upward (downward) trend. In bilateral trend detection, a time series with a significant trend is indicated if |Z|≥ Z1−α/2 at a certain significance level α, where Z1−α/2 is obtained from the standard normal cumulative distribution tables. The trend is statistically significant at the 0.1, 0.05, and 0.01 significance level when |Z| > 1.645, 1.96 and 2.576, respectively. Besides, the Mann–Kendall test can also be used to detect the abrupt points. The abrupt points and the approximate time of occurrence can be located according to the intersection of the progressive and retrograde sequences within the sequence. More details of the abrupt points calculation on the Mann–Kendall Test are available from the network resources. The mean annual rainfall erosivity at each of the 91 stations was calculated by a long-term (1971–2020) average value of annual rainfall erosivity. Based on these station’s values, the co-kriging interpolation method was used to interpolate the spatial distribution of the average annual erosivity of the TP, using the geostatistical analysis tool ArcGIS 10.4.Different from the inverse distance weighting (IDW) method which only considered one assumption: nearby points should be closer to the value of the interpolation position than distant points, the co-kriging interpolation method allowed the addition of covariates to improve the accuracy of estimation or prediction [74,82]. Considering the complex terrain of the TP, the elevation factor was defined as a co-variable in the co-kriging interpolation method [36]. The elevation data of each meteorological station was provided by the China Meteorological Administration (CMA). Based on the values of rainfall erosivity for 91 meteorological stations on the TP, the co-kriging interpolation method was performed and generated the spatial distribution map of rainfall erosivity for the period of 1971–2020.In this study, the rainfall erosivity anomalies of the TP is expressed as the difference between the annual erosivity value of the observation year and the 50-year average value. The 5-year moving average anomaly can smooth fluctuations and reduce potential errors, and was used to analyze the temporal changes of the rainfall erosivity across the TP. The seasonal (monthly) average rainfall erosivity of the TP was calculated by averaging the seasonal (monthly) erosivity of 91 meteorological stations during the same time span, from 1971 to 2020. The annual variation from the meteorological site is represented by the coefficient of variation (CV), which is expressed as a percentage of the standard deviation of the annual erosivity to the average of the observation year. The map of seasonal spatial distribution of rainfall erosivity is expressed as a percentage of the seasonal rainfall erosivity in the annual total erosivity at each weather station. With reference to the time series trend identification method of Ray [83], rainfall erosivity change patterns were identified for each meteorological station from 1971 to 2020. Multiple indicators were integrated: the Z values calculated by Mann–Kendall, SLOPE% and RST.The SLOPE% value represents as a percentage of SLOPE for the average rainfall erosivity for each meteorological station during 1971–2020. The calculation formula is as follows:(9)SLOPE%=SLOPE∑i=1n xi/n×100
+          where SLOPE is the Sen’s slope value of rainfall erosivity changes, xi is the value of the rainfall erosivity corresponding to period i, and n is the length of the time series. The RST is defined as the ratio of the average rainfall erosivity for the last 3 years to the maximum 3 year moving average. This is used to identify whether the increasing trend of annual rainfall erosivity is interrupted, shifting to a decline at later stages. The equation is expressed below:(10)RST=avexn−2,xn−1,xnmaxAVEx1,x2,x3,AVEx2,x3,x4,…AVExn−2,xn−1,xn
+          where xi is the value of the rainfall erosivity corresponding to period i, and n is the length of the time series.The trend of rainfall erosivity was classified by the above-mentioned three indicators into four patterns of decreasing, stagnant, increasing-stagnant and increasing (Table 1), abbreviated as DE, ST, IN-ST and IN, respectively. The Z value indicates whether there is a significant trend of rainfall erosivity (|Z| > 1.96 at the 0.05 significance level), a non-significant change trend (0.675 < |Z| ≤ 1.96 at the 0.05–0.5 significance level), and no change trend (|Z| ≤ 0.675 at the below 0.5 significance level); SLOPE% indicates whether the magnitude of the rainfall erosivity trend change is significant. References [84,85] used 0.25% as a criterion, i.e., a change greater than 0.25% is assumed to be significant. RST is used to determine whether the annual rainfall erosivity growth trend is interrupted, or turns down and mitigates at a later stage.The continuous wavelet transform (CWT) is a method to decompose a time series into a two-dimensional phase plane of the time-frequency simultaneously. It is commonly applied to the analysis of various hydrological and meteorological processes with high variability to detect non-stationary trends, periodicities, and durations as it better characterizes oscillatory behaviors of signals than discrete wavelet transforms [86,87,88]. Specifically, two CWTs, cross wavelet transform (XWT), and wavelet transform coherence (WTC), were constructed to investigate whether there is any periodicities or correlations between rainfall erosivity and ENSO. The XWT reveals regions of high common power in the time-frequency spectrum, and calculates the phase relationships between signals. The WTC identifies two time series variation correlations in both time and frequency space, even in the absence of high-power regions. In this study, the wavelet power spectrum of CWT was employed to analyze the relationship and the possible periodicity between rainfall erosivity in different regions and changing patterns of MEI. XWT revealed high common power regions and phase relationships between the two variables, and WTC was used to determine the correlation position of the two variables at local scales. The CWT toolbox package for MATLAB was used to perform all wavelet analyses. For further details about CWT, refer to [89].In the following section, the results are represented according to the technical approach mentioned in Section 2. This section analyzed the variability characteristics of rainfall erosivity at different time scales and the spatial distribution pattern of rainfall erosivity at each station, while identifying the relationship between rainfall erosivity and ENSO on the TP.Sen’s slope estimation analysis showed an increasing trend of rainfall erosivity on the Tibetan Plateau from 1971 to 2020, with a Sen’s slope value of 2.69 for annual rainfall erosivity (Figure 3). The average annual erosivity range from 713.50 to 1495.41 MJ·mm·ha−1·h−1, with a multi-year average erosivity of 1071.42 MJ·mm·ha−1·h−1. An anomaly analysis indicated obvious inter-annual fluctuation in rainfall erosivity. The magnitude of rainfall erosivity undulation was relatively small until 1996 and increased significantly after 1996, with longer fluctuation periods. For the entire study period, the annual rainfall erosivity was above the mean for the same duration as the periods below the mean, with the highest value of 1495.41 MJ·mm·ha−1·h−1 in 2020; the lowest value of 713.50 MJ·mm·ha−1·h−1 in 2009; the extreme value ratio was 2.1 (Figure S1). Meanwhile, the Mann–Kendall trend analysis had a Z value of 1.67, indicating that this trend passed the significance test at the 90% confidence level, with the mutation point occurring in approximately 2017 (Figure S2).The seasonal mean rainfall erosivity showed significant discrepancies at 91 meteorological stations on the TP. The rainfall erosivity in order was summer > autumn > spring > winter, with a range of 89.33–662.58 MJ·mm·ha−1·h−1. In particular, the average rainfall erosivity in summer was the highest, accounting for 60.36%, while winter was the lowest, accounting for only 8.14% of the total annual erosivity (Figure 4). This phenomenon was mainly influenced by the heterogeneity of the seasonal distribution of precipitation. With the transport of water vapor from the North Indian and Western Pacific monsoons, the summer monsoon brought 58.5% of the year’s rainfall, while late spring and early autumn accounted for 90% of the year’s rainfall [90]. As shown in Figure S3, the summer rainfall erosivity showed a non-significant increasing trend, with the MK statistical value of 1.54. In contrast, there was a decreasing trend in spring, autumn, and winter rainfall erosion; the spring and autumn MK statistic passed the significance test (p = 0.05), which were −2.19 and −2.09, respectively.Although the monthly average rainfall erosivity was highly variable, there was a clear temporal consistency with the seasonal rainfall erosivity. June, July, and August, corresponding to summer, were the three months with the highest percentage of rainfall erosivity for the year. The monthly rainfall erosivity was 168.73 MJ·mm·ha−1·h−1, 264.75, and 229.10 MJ·mm·ha−1·h−1, respectively. Rainfall erosivity was highest on the Tibetan Plateau in July, the proportion of erosivity reached 24.19% for the year. November was the lowest with 1.12%, but a higher variability was found in this period, with a high extreme ratio of 18.71 (Figure 4). The trends in the monthly average rainfall erosivity from 1971 to 2020 were further examined, showing an increasing trend in eight months and a decreasing trend in four months. Specifically, the greatest increasing trend in mean rainfall erosivity was in August, but the increasing trend was insignificant in all months. Three months showed a significant decreasing trend, including March and April at 90% and 95% confidence levels, respectively, and the most significant decreasing trend was in November, which passed 99% confidence level (Figure S3).In order to reduce the boundary effect on the annual rainfall erosivity spatial pattern, 91 meteorological stations were selected for the interpolation better to reveal the spatial variation in rainfall erosivity. In general, the rainfall erosivity on the TP from 1971 to 2020 had obvious spatial differences, roughly exhibiting a spatial pattern of decreasing distribution from southeast to northwest (Figure 5). The high value zone of rainfall erosivity was approximately in the southeastern part of the TP, mainly distributed in the lower altitude regions such as the Hengduan Mountains and the Yarlung Tsangpo Valley. There were three stations with an average annual rainfall erosivity greater than 6000 MJ·mm·ha−1·h−1, of which the highest value occurs at Dujiangyan station in Sichuan Province, with an average annual rainfall erosivity of 6605.21 MJ·mm·ha−1·h−1. The other two stations are Gongshan station and Huaping station in Yunnan Province, with an average annual rainfall erosivity of 6152.44 MJ·mm·ha−1·h−1 and 6193.75 MJ·mm·ha−1·h−1, respectively. The zones with low average annual rainfall erosivity were mainly found in the northern and western parts of the TP, including concentrations in the Qiangtang Plateau and the Qaidam Basin. For example, the lowest value was at Shiquanhe station in the Tibet Autonomous Region, where the annual rainfall erosivity was only 103.46 MJ·mm·ha−1·h−1. In summary, the average annual rainfall erosivity was less than 500 MJ·mm·ha−1·h−1 which accounted for 48.35% of all stations, 500–1000 MJ·mm·ha−1·h−1 for 24.16% and that of more than 1000 MJ·mm·ha−1·h−1 accounted for 27.47%.The long time series trend analysis of each meteorological station on the TP from 1971 to 2020 showed that the annual rainfall erosivity exhibited an increasing trend at 49 meteorological stations, accounting for 54% of all stations. Thereby this demonstrated the reason for the increasing rainfall erosivity across the entire TP since 1971 (Figure S4). Of these, 13 meteorological stations increased consistently, mainly in the eastern Hengduan Mountains of the TP, accounting for 27% of the total increases. The meteorological stations that showed a pattern of rainfall erosivity increased first and then gradually stabilized were mainly located in the Hehuang valley in the northeastern part of the Tibetan Plateau and the southern Tibetan valley in the southeast, accounting for 27% of the total increase in stations. The annual rainfall erosivity showed a long-term stable (no significant trend) pattern at 37 meteorological stations, accounting for 41% of all meteorological stations. Only five stations showed a gradual decrease in annual rainfall erosivity, accounting for 5% of all meteorological stations, but both patterns of change were not significant. In addition, as shown in Table S2, further analysis based on the statistics results of the multiple indicators at each station revealed that the types of trends at each site differ significantly in terms of significance levels and magnitude of change. Of all the increasing pattern stations, the significant increases were seen at Min station (p = 0.05) and Derong station (p = 0.01). The largest and smallest increases were at Min station and Pishan station, respectively. Of all the increasing-stagnating pattern stations, there are five meteorological stations at more than 95% confidence level, namely Zekog station (p = 0.01), Gerze station (p = 0.01), Dulan station (p = 0.01), Wuwei station (p = 0.01), and Guinan station (p = 0.05). Of all the decreasing pattern stations, the largest and smallest increases were at Gongshan station and Artux station, respectively.As shown in Figure 6, the mean coefficient of variation (CV, the ratio of the standard deviation to the mean) in the interannual rainfall erosivity for 91 stations since 1971 was 0.61, indicating moderately high rainfall erosivity variability across the plateau. The spatial distribution pattern of CV had high consistency with annual rainfall erosivity. In other words, the CV increased from south-east to north-west. Specifically, 11 meteorological stations were in regions of intense variation (CV > 1), mostly in the north-western flank of the Kunlun Mountains on the TP, the Ali Mountains in the Tibet Autonomous Region, and the northern Qilian Mountains in eastern Qinghai Province. while 54 meteorological stations were in regions of lesser variation (CV < 0.5), accounting for 59% of the total meteorological stations, mainly in the south-eastern part of the TP. Furthermore, the meteorological station with the smallest CV in the interannual rainfall erosivity was Jiulong Station, located in Sichuan Province in the southeastern part of the Tibetan Plateau, while the largest CV was Pishan Station, located in Xinjiang Autonomous Region in the northwestern part of the Tibetan Plateau. In summary, over the past half century, rainfall erosivity exhibited clear spatial disparities on the TP, specifically, annual rainfall erosivity in the southeast were mainly characterized by a slowly and steadily increase, while annual rainfall erosivity in the northwestern part of the plateau showed greater fluctuations and instability, with no significant trends.The seasonal spatial distribution pattern of rainfall erosivity varied widely across the TP (Figure 7). Five meteorological stations (5.5% of the total) with the highest percentage of spring rainfall erosivity were concentrated in the Kunlun Mountains on the Tibetan Plateau near the Pamir Plateau and in the Nu River basin in the Eastern Himalaya. In particular, the erosivity of spring rainfall accounted for more than 50% of Pishan station, Kashgar station, and Zayu station, and Pishan station was as high as 79%. (Figure 7a). Nearly 92% of the meteorological stations (total 84) had the highest percentage of summer rainfall erosivity. The largest was Shiquanhe station, which surprisingly had 94.28% of the annual rainfall erosivity (Figure 7b). There was only one meteorological station with the highest percentage of fall rainfall erosivity, with three stations accounting for more than 30%, namely Nyalam, Burang, and Keriya station (Figure 7c). Winter was the season with the lowest percentage of rainfall erosivity, all stations had less than 30% of rainfall erosivity (Figure 7d). Summer and autumn were the most erosive seasons. It is worth noting that rainfall erosivity was generally higher across the plateau in summer, particularly in the southeastern part of the Tibetan Plateau, whereas rainfall erosivity in autumn and winter was still higher proportion in the south-western part of the plateau near the Himalayas. Thus, extra caution will be needed to prevent aggravation of soil erosion in this region.The El Niño and La Niña events are the ENSO cycles of the warm and cold periods, respectively. Statistics on the rainfall erosivity in Region I (arid zone) and Region II (humid zone) of the TP and whole plateau during the El Niño (ENSO warm event) and La Niña (ENSO cold event) periods are presented in Table 2.The occurrence and duration of El Niño event and La Niña event were determined based on ONI.In terms of the degree of influence of cold and warm events on the rainfall erosivity, the average monthly rainfall erosivity for the El Niño event was slightly lower than the La Niña event across the TP, but the average of both events was less than the monthly rainfall erosivity for the period 1971–2020. During the El Niño event, the maximum monthly average rainfall erosivity was 225.18 MJ·mm·ha−1·h−1 and the minimum value was 120.58 MJ·mm·ha−1·h−1, with an extreme value ratio of 1.87; during the La Niña event, the maximum monthly average rainfall erosivity was 486.03 MJ·mm·ha−1·h−1 and the minimum value was 101.67 MJ·mm·ha−1·h−1, with an extreme value ratio of 4.78; thus, higher variability occurred during the La Niña event. In terms of the presence or absence of ENSO events, the average rainfall erosivity during the ENSO and Non-ENSO periods were 192.67 MJ·mm·ha−1·h−1 and 213.77 MJ·mm·ha−1·h−1, respectively. It was evident that the average monthly rainfall erosivity during the non-ENSO period was not only greater than that during the ENSO period, but also greater than the total average monthly rainfall erosivity for the whole study period.The impact of ENSO events on the monthly mean rainfall erosivity in different regions was notably dissimilar. For Region I, the average monthly rainfall erosivity for the El Niño and La Niña events were 147.17 MJ·mm·ha−1·h−1 and 173.85 MJ·mm·ha−1·h−1, respectively. This was higher than the average monthly rainfall erosivity for this region for 1971–2020. Moreover, we found that when El Niño events or La Niña events occurred, there was a significant increase in rainfall erosivity in Region I relative to the Non-ENSO period, but La Niña events had a greater impact on the monthly average rainfall erosivity, compared with El Niño events; For Region II, the average monthly rainfall erosivity for the El Niño and La Niña events were 208.93 MJ·mm·ha−1·h−1 and 210.13 MJ·mm·ha−1·h−1, respectively, with the El Niño event slightly lower than the La Niña event. There was a modest gap between the two events. However, compared with Region I, the direction of influence of ENSO in Region II was in the opposite direction. In other words, when ENSO occurs, the average monthly rainfall erosivity in this region decreased more significantly than the average for the study period. Due to the difference in the magnitude of rainfall erosivity on the Tibetan Plateau during the ENSO period and the non-ENSO period, under the premise that other contributing factors was fixed, rainfall erosivity was stronger during the non-ENSO period and soil erosion concerns and soil conservation measures should be strengthened during this period. Considering the obvious spatial heterogeneity of the impact of ENSO on the Tibetan Plateau, the emphasis should be on erosion in the north-west during El Niño or La Niña events, especially during the La Niña event when control measures should be enhanced.To examine the extent and impact of ENSO on rainfall erosivity, an XWT and WTC analysis were conducted on the time series of rainfall erosivity and MEI index in different regions of the Tibetan Plateau from 1971 to 2020, revealing the periodicity characteristics of both. As shown in Figure 8, in the time-frequency space domain of Region I, it is obvious that there was 3–5 years of high-energy resonance cycle between rainfall erosivity and the MEI index for the period of 1981–1988, during which there was a negative correlation between both time series. In the Region II power spectrum, there were two significant high-energy domains, specifically a 3–5 years resonance cycle from 1981 to 1988 was similar to that of Region I, indicating a consistent ENSO effect across the plateau during this period, but the intensity of the Region II resonance cycle was higher. The other was that there was a 2–5 years resonance cycle of rainfall erosivity and the MEI index from 1995 to 1999, and the mean phase angle was nearly 90° vertically upwards, indicating that the rainfall erosivity change was later than the MEI index. In other words, rainfall erosivity had a lag compared with ENSO over the same period. As shown in Figure 9, in the Region I WTC power spectrum, there were negative phase cycles of 3–5 years and 2–3 years in 1985–1992 and 2006–2009, respectively, indicating a negative correlation between rainfall erosivity and the MEI index during this period. Regarding Region II, there were negative phase cycles of 3–7 years from 1977 to 1988, 1–3 years from 1994 to 1998 and 2–4 years from 2007 to 2013, indicating a negative correlation between rainfall erosivity and the MEI index during these periods. While positive phase cycles of 1–2 years from 1987 to 1989 indicate a positive correlation between rainfall erosivity and the MEI index during this period.This study revealed the spatial and temporal characteristics of rainfall erosivity on the TP from 1971 to 2020 and their relationship with the ENSO Index. The results showed that the average annual rainfall erosivity on the TP since 1971 was 1071.42 MJ·mm·ha−1·h−1. According to previous studies, this value is higher than in northwestern China but lower than in southeastern China [74], and the overall degree of erosion is light, approximately 0.5 times the global average [91]. Upward trends are shown for rainfall erosivity during 1971 to 2020. Gu et al. also found an increasing trend in rainfall erosivity from 1981 to 2015 in the Tibet Autonomous Region (TAR) [60], and Wang et al. found a same uptrend in rainfall-runoff erosivity from 1961 to 2012 in Sanjiangyuan region, Qinghai Province [53], which is consistent with this study. Fan et al. [61] used TRMM 3B42 data to assess the spatial and temporal variability of rainfall erosivity in the TAR from 2000 to 2010 and found that the average rainfall erosivity was 768 MJ·mm·ha−1·h−1, which is lower than the results of this study, probably due to the different extent of the study area and the accuracy of the data. Moreover, the reasons for the trend of increasing annual rainfall erosivity but significant decreasing rainfall erosivity in spring and autumn may be related to the variation in rainfall on the TP [92]. Previous studies have shown that since 1961, the Tibetan Plateau is gradually warming and humidifying [93], which may contribute to an increase in rainfall erosivity. Meanwhile, while changes in the westerly circulation lead to a reduction in rainfall in spring and autumn which in turn affects rainfall erosivity [94].In the previous section, this work indicated rainfall erosivity on the TP varied greatly not only seasonally but also monthly. This may be caused by its complicated geography and dominant atmospheric circulation conditions [95]. The plateau spans a wide range of latitudes and longitudes and has a variety of climate types. It is also at the crossroads of monsoon and non-monsoon zones, and is influenced by the prevailing westerly winds, the South Asian monsoon and the East Asian monsoon circulation, resulting in an uneven spatial and temporal distribution of rainfall. Besides, according to previous reports, the amount and intensity of rainfall are the main factors affecting the rainfall erosivity [96]. The spatial and temporal variability of rainfall at various magnitudes will certainly contribute to soil erosion by water at different times and regions to different degrees [97,98]. The average heavy rainfall and average heavy rainfall erosivity for each station from 1971 to 2020 are presented in Table S3. It can be seen from Table S3 that the average heavy rainfall (≥25 mm) of 91 meteorological stations in the plateau was 33.05 mm, and the average rainfall erosivity was 243.35 MJ·mm·ha−1·h−1. The absolute amount of erosivity caused by heavy rainfall on the TP is low compared with the eastern coastal areas of China, mainly because the total rainfall erosivity in the plateau is much lower than those in the east [35,99,100]. Furthermore, the comparison of the proportion of heavy rain and rainfall erosivity found that the average heavy rainfall on the TP accounted for only 27.02% of the total rainfall, but the rainfall erosivity caused by heavy rainfall occupied by 43.3% of the total rainfall erosivity (Figure 10). In addition, the change in the percentage of heavy rainfall and heavy rainfall erosivity has a high consistency on the TP. In case of heavy rainfall, it can be easily seen that rain erosivity becomes more intense. The maximum values of heavy rainfall and heavy rainfall erosivity were 71.26% and 81.46%, respectively, both of which occurred at Dujiangyan station, and the minimum values of 6.77% and 8.15%, respectively, which occurred at Zhidoi station. Studies have revealed that rainfall is the most important climatic factor contributing to soil erosion, and in particular, heavy rainfall (≥25 mm) is one of the main factors affecting the rainfall erosivity [101]. In general, the higher the intensity of the heavy rainfall, the greater the amount of soil erosion. The frequency of extreme rainfall events is growing as a result of global climate change [102,103], Future research should pay more attention to high intensity rainfall and soil erosion.The spatial distribution of rainfall erosivity on the TP decreased roughly from south-east to north-west, with significant spatial heterogeneity. Previous studies have shown that mountain tectonics and topographic gradients are essential factors influencing precipitation [104]. The high rainfall erosivity in the south-east is largely attributed to the roughly north-south alignment of the Hengduan Mountains and the gradual rise in elevation from south to north, which is a natural water vapor corridor and facilitates the deeper uplift of monsoon air masses and the formation of rainfall. The north-west of the plateau, on the other hand, is located in an inland region, with high altitude and low temperatures. It is extremely hard for the monsoon to reach it, and rainfall is minimal throughout the year, resulting in lower rainfall erosivity. Besides, in the southern part of the plateau, although it is on the leeward slopes of the Himalayas, where water vapor is not easily accessible, the Yarlung Tsangpo valley is at a relatively low altitude and has locally better hydrothermal conditions, resulting in a higher rainfall erosivity in the region. In addition, the role of anthropogenic activities should not be neglected. Nearly 38.8% of the grasslands on the TP have been degraded [105], and the degradation of meadows caused by overgrazing is a serious environmental and ecological problem. Consequently, soil erosion caused by the contradiction between people and land may exacerbate the rainfall erosivity. Clarifying the temporal and spatial patterns of rainfall erosivity on the TP over the past 50 years is of great significance for soil conservation and future land use planning. This paper indicates that the tendency of increasing rainfall erosivity was identified in the southeastern part of the TP. Published studies have shown that the Hengduan Mountains region has the highest soil erosion modulus of the TP [106]. Furthermore, the alternation of steep slopes and deep ravines, the superimposed effect of complex topography and intensified rainfall erosivity may further magnify soil erosion in this region. The Yarlung Tsangpo Valley and the Hehuang Valley are areas of intensive human activity on the TP, and are also major wheat and barley cultivation areas, with crops mostly grown on the slopes of the valleys [107]. Although rainfall erosivity has not shown a sharp rise over the past 50 years in these origins, the region’s originally high rainfall erosivity may still result in frequent natural hazards and elevated ecological risks. The deployment of measures against landslides and debris flows should be considered as a priority. In the western and northern parts of the Tibetan Plateau, there is higher variability, although with lower rainfall erosivity. On the one hand, in the context of increased rainfall erosivity across the plateau, it is still possible to damage the low-cover turf. In particular, the interactive effects of human activities such as grazing and soil erosion processes can also exacerbate the ‘black beach’ degradation of plateau grasslands. On the other hand, the north-west has a higher altitude and fragile natural environment where seasonal differences in rainfall erosivity is more likely to cause damage to alpine ecosystems. Therefore, soil erosion management strategies in this region should not be neglected either.The impact of global-scale climate oscillation regimes on climate change has received widespread attention. This paper used MEI to characterize the global-scale climate oscillation model ENSO in an attempt to explore the relationship between ENSO and rainfall erosivity on the Tibetan Plateau, with a view to providing guidance for collaborative work on climate change and soil conservation. This study found that rainfall erosivity was lower during ENSO than during non-ENSO, and other studies have found the same results [35,36]. ENSO is a major factor influencing temperature and precipitation in China. Some studies have shown that precipitation anomalies can reach up to 30% of the average precipitation during ENSO periods [108]. It is worth noting that rainfall erosivity was higher during the La Niña period than during the El Niño period on the TP. This is not in agreement with previous studies, with results in places such as Fujian in southeastern China [37], but is consistent with studies in Guizhou in southwestern China [36]. The other research indicated that El Niño occurs with a delayed arrival of the southwest monsoon, while the opposite occurs at La Niña, so this may be related to a weakening of the Indian monsoon [109,110]. These findings would explain the difference of rainfall erosivity between the La Niña period and El Niño period. According to the CWT results, a noticeable resonance cycle between rainfall erosivity and MEI was found in different regions of the Tibetan Plateau, but there were also significant differences in cycle duration, direction of action, and intensity. This may be due to the fact that ENSO events themselves present diversities of climatic features at each stage of occurrence, development, maturation and decline [108]. Additionally, it also suggested that global climate anomalies are an important driver of changes in the rainfall erosivity on the TP.Due to the vast expanses of land and sparse populations as well as the harsh natural conditions, the distribution of weather stations on the Tibetan Plateau is extremely irregular, which may affect the accuracy of the interpolation. Although this study improves the comprehension of the impact of ENSO on rainfall erosivity, it still lacks further explanation from a mechanistic perspective. Furthermore, in light of the known results, it is clear that not only ENSO but also topography, altitude, and microclimate are associated with rainfall erosivity, and detailed knowledge will be necessary for future studies.This study carried out an insightful analysis of the spatial, interannual, and seasonal variability of rainfall erosivity on the TP from 1971 to 2020 and its relationship with ENSO. Daily rainfall data from 91 meteorological stations were collected, and the change trend of rainfall erosivity calculated based on a daily rainfall erosivity model were detected at a regional and site-scale using methods such as Mann–Kendall test and Sen’s slope. The potential influence of ENSO on rainfall erosivity was revealed using the continuous wavelet transform method. The main findings were summarized below:Rainfall erosivity has shown a fluctuating trend of increasing over the past half century. Seasonal and monthly rainfall erosivity showed high heterogeneity, which was greatly related to heavy rainfall. The rainfall erosivity in order was summer > autumn > spring > winter. July was the most erosive month, accounting for 24.19% of the year, while November was the lowest, accounting for only 1.12%. The rainfall erosivity in spring and autumn showed a significant decreasing trend (p < 0.05), and in summer it showed an increasing trend but not significant. There was generally an obvious spatial variation in rainfall erosivity on the TP from 1971 to 2020, presenting a roughly spatial pattern of decreasing distribution from southeast to northwest. Annual rainfall erosivity in the south-eastern part of the plateau was mainly characterized by a slow increase, while in the north-western part annual rainfall erosivity was more unstable with mostly no significant trends.ENSO events had a significant impact on rainfall erosivity on the TP. The rainfall erosivity in the non-ENSO period was higher than that in the ENSO period, and the La Niña event was higher than the El Niño event. It was also found that there was a clear resonance cycle between rainfall erosivity and ENSO in different regions of the plateau, with an average cycle of about 3–5 years in the high energy region, but there were differences in the timing of occurrence, direction of action, and intensity of the cycle. The rainfall erosivity on the TP was relatively large during non-ENSO periods and relatively small during El Niño/La Niña periods. In addition, the response of rainfall erosivity to ENSO was spatially heterogeneous. Rainfall erosivity in the northwest of TP appears to be more serious during the La Niña event and less severe during the El Niño event. It can be concluded that soil erosion may become more intense during the La Niña event in the northwest TP. Therefore, during the La Niña event, soil protection should be enhanced to diminish soil spattering and disturbance.This study contributes to the understanding of the spatial and temporal variability of annual rainfall erosivity across the entire TP over the last half century and extends the cognition of the possible impact of changes in ENSO characteristics associated with climate change. Uncertainties may be involved due to limited data availability and interpolation bias errors. Future studies should integrate the effects of multiple factors on rainfall erosivity, more carefully relate the effects of climate extremes, and improve the insights from the mechanistic aspects of change.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111054/s1, Figure S1: Rainfall erosivity anomalies from 1971 to 2017, the blue line shows the 5–year moving average curve, Figure S2: Sequential MK test for 50-year rainfall erosivity of TP. Note: UF and UB refer to progressive and retrograde sequences within the sequence, respectively. UF > 0 indicates an increasing trend, UF < 0 indicates a decreasing trend. The mutation year exists at the intersection of UF and UB. The dashed line represents the 95% confidence interval, Figure S3: Seasonal and monthly average rainfall erosivity Mann–Kendall trends of plateau from 1971 to 2020. ** p < 0.01; * p < 0.05; + p < 0.1, Figure S4: Long–term trend pattern of rainfall erosivity at each station from 1971 to 2020. Note: DE: decreasing, ST: stagnating, IN-ST: increasing-stagnating, IN: increasing, Table S1: The basic information of 91 meteorological stations in this study. ‘No’ refers to station number, which is the same as the station number in Figure 1, Table S2: Statistical results of Z, SLOPE, and Rst associated with the rainfall erosivity trend classification at each meteorological station on the TP from 1971 to 2020, Table S3: The average heavy rainfall (AHR) and its proportion (HR) and the proportion of average heavy rain erosivity (HRE) of each station from 1971 to 2020.Y.Z. and B.C. had the original idea and designed the study. B.C. processed and analyzed the data and wrote the manuscript; Y.Z., L.L. and Z.W. had insights on the revision of the manuscript and suggestions for improvement. Z.X. provided help and guidance for data processing and model adjustment. C.G., B.W. and D.G. revised the paper and polished the language. All authors have read and agreed to the published version of the manuscript.This research was funded by the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDA20040201), the Second Tibetan Plateau Scientific Expedition and Research (Grant No. 2019QZKK0603), and the National Natural Science Foundation of China (Grant No. 41671104).Not applicable.Not applicable.All relevant data sets in this study are described in the manuscript.The authors declare no conflict of interest.Study area of Tibetan Plateau (TP) and the distribution of meteorological stations.Technical route. Note: MK-test refers to Mann–Kendall test, CV refers to the coefficient of variation.Annual variation in rainfall erosivity on the TP from 1971 to 2020, the red dotted line represents Sen’s estimate.Statistics of seasonal and monthly average rainfall erosivity and its percentage.Spatial distribution of rainfall erosivity on the TP during the period from 1971 to 2020.Spatial distribution of the coefficient of variation (CV) in rainfall erosivity during 1971–2020.Spatial pattern of rainfall erosivity from 1971 to 2020 in spring (a), summer (b), autumn (c), and winter (d). Note: Percentage refers to the proportion of seasonal rainfall erosivity to total annual erosivity.Cross wavelet transforms (XWTs) for annual rainfall erosivity and multivariate ENSO Index (MEI) in Region I (a) and Region II (b). NOTE: The thick black outline indicates the 95% significance level against red noise, the white translucent area indicates the cone of influence, and the color bar indicates the magnitude of the XWT cross spectral power. That is, red is strong and blue is weak. The arrows (vectors) designate the phase difference between rainfall erosivity and MEI. Where the left arrow indicates the opposite phase relationship between the rainfall erosivity and MEI and vice versa. The north-pointing arrow indicates that the peak rainfall erosivity are lower than the peak MEI.Wavelet transform coherence (WTC) for annual rainfall erosivity and multivariate ENSO Index (MEI) in Region I (a) and Region II (b). NOTE: The thick black outline indicates the 95% significance level, the white translucent area indicates the cone of influence, and the color bar indicates the significance level of the Monte Carlo test. That is, red means strong correction and blue is weak. where the left arrow indicates the opposite phase relationship between the two-time series, vice versa.Comparison of the proportion of heavy rain (daily rainfall ≥ 25 mm) and rainfall erosivity for 91 meteorological stations on the TP from 1971 to 2020. Note: the designation of station numbers is shown in Table S3; and the percentage of heavy rainfall represents the proportion of heavy rain amount to total annual precipitation; percentage of heavy rainfall erosivity represents the proportion of heavy rainfall erosivity to annual rainfall erosivity.Definition and indicators of rainfall erosivity trend patten.Note: The Z value indicates the MK trend detection value, the SLOPE% is the trend change rate percentage, and the RST refers to the ratio of the rainfall erosivity in the past 3 years to the maximum 3-year moving average. DE: decreasing, ST: stagnating, IN-ST: increasing-stagnating, IN: increasing.Average monthly rainfall erosivity in different regions (Region I and Region II) and TP for El Niño and La Niña events from 1971 to 2020. The bottom table provides a summary of the average monthly rainfall erosivity.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11055.txt

@@ -0,0 +1,2 @@
+Particulate matter (PM) is a complex mixture of solid particles and liquid droplets suspended in the air with varying size, shape, and chemical composition which intensifies significant concern due to severe health effects. Based on the well-established human health effects of outdoor PM, health-based standards for outdoor air have been promoted (e.g., the National Ambient Air Quality Standards formulated by the U.S.). Due to the exchange of indoor and outdoor air, the chemical composition of indoor particulate matter is related to the sources and components of outdoor PM. However, PM in the indoor environment has the potential to exceed outdoor PM levels. Indoor PM includes particles of outdoor origin that drift indoors and particles that originate from indoor activities, which include cooking, fireplaces, smoking, fuel combustion for heating, human activities, and burning incense. Indoor PM can be enriched with inorganic and organic contaminants, including toxic heavy metals and carcinogenic volatile organic compounds. As a potential health hazard, indoor exposure to PM has received increased attention in recent years because people spend most of their time indoors. In addition, as the quantity, quality, and scope of the research have expanded, it is necessary to conduct a systematic review of indoor PM. This review discusses the sources, pathways, characteristics, health effects, and exposure mitigation of indoor PM. Practical solutions and steps to reduce exposure to indoor PM are also discussed.Air pollution results from introducing various contaminants into the atmosphere that are likely to be detrimental to humans, other living organisms, and the natural environment [1]. Among a wide range of air pollutants, particulate matter (PM) is of particular concern because of its association with cardiopulmonary health disorders [2,3]. PM is a complex mixture of solid particles and liquid droplets made up of metals, organic compounds, sulfate, nitrate, ammonium, and other ions (Figure 1) [2,4]. In different indoor environments, the composition is generally the same, but due to different indoor environment types, the difference may be small, and the proportion may be significantly different [5,6,7,8,9]. Early studies have shown that PM’s composition is associated with some respiratory diseases, including asthma, chronic bronchitis, and acute bronchitis [10,11,12].In addition to physical and chemical composition, the size of PM is an essential factor related to its effect on health. Briefly, PM is broadly categorized by its “aerodynamic equivalent diameter” (AED). As shown in Figure 2, particles with diameters between 2.5 and 10 μm (PM2.5–10) are defined as “coarse”; less than 2.5 μm as “fine”; and less than 0.1 μm as “ultrafine” [13]. The main threat to health from PM is inhalable particulate matter, which can penetrate the chest area of the respiratory system and cause adverse health effects (Figure 3) [14]. For example, because PM2.5 is light, it has a higher incidence rate and deposition rate in the lungs than other particles, resulting in it staying longer in the respiratory tract [15,16]. Pope et al. found that for every 10 μg/m3 increase in PM2.5 levels, the risk of death increases by 8% to 18% [16]. According to data from the World Health Organization, about seven million people die from exposure to PM2.5 in polluted air every year [17]. Moreover, ultrafine particles cause a more significant inflammatory response than fine particles per given mass [18], and the ultrafine-particle-toxicity effect can be enhanced by a gaseous co-pollutant such as ozone [14,19].With technological advancements and lifestyle changes, more human activities, including cooking, cleaning, and indoor sports activities, are carried out indoors, all of which are likely to increase indoor PM [20]. Exposure studies show that indoor PM contributes substantially to personal exposure, and the indoor PM concentration levels may exceed those outdoors [1]. Specifically, epidemiologic studies have presented evidence that indoor PM plays a significant role in human health, such as lung malfunctioning, cardiovascular disease, respiratory symptoms, asthma, and premature births [21,22,23,24]. In addition, PM may alter the immune response by promoting immunoglobulin E, causing an inflammatory response. Nowadays, the indoor, suspended PM concentration has been identified as one criterion for evaluating indoor environmental quality [25]. For example, in 2012, Canada established the Residential Indoor Air Quality Guidelines, which state that PM2.5 needs to be monitored, with a limit of 100 µg/m3 as a 1 h average (Short-Term Exposure) and 40 µg/m3 as an 8 h average (Long-Term Exposure). The National Ambient Air Quality Standard (NAAQS), set by USEPA, stipulates 35 µg/m3 and 15 µg/m3 for 24 h and annual periods, respectively, for exposure to PM2.5. Furthermore, the World Health Organization recommended to apply to indoor spaces the same PM guidelines as for ambient air, presented on the 2005 global update, which are 25 and 50 μg m−3 for PM2.5 and PM10, respectively (over 24 h) [26].Indoor PM sources include indoor origins and outdoor infiltration. Primary indoor sources of PM result from specific activities (cooking, sweeping, dusting, candle or incense burning, using laser-printing devices, fuel combustion for heating, and smoking tobacco), the design of the house, and secondary organic aerosols [27,28,29,30]. Due to air exchange, the indoor PM also originates from outdoor sources, including natural ones (forest fires, soil dust, and sea salt) and anthropogenic ones (transport, oil combustion, and coal burning in power plants) [2,31]. The automatic monitoring methods of PM2.5 ambient air quality are the β-ray method and micro-oscillatory balance method. The effectiveness of urban ambient air quality assessment can be effectively improved by selecting suitable methods in suitable areas.To ensure the health and comfort of indoor environments, strategies should be taken to reduce indoor PM concentrations. The development of technology has allowed the new trend of operating indoor air purifiers to control indoor PM concentrations. Systems based on the principle of electrostatic precipitation and air purifiers using filtration technologies are two of the most common indoor air-purification technologies [32,33,34]. Natural ventilation with open windows is a common and economical approach to diluting indoor PM [35].Numerous studies have described sources and health risks of indoor PM, but their conclusions have been focused on a single topic, e.g., the sources and its effects on susceptible subgroups [36,37,38]. As the quantity, quality, and scope of the research have expanded, this review is necessary to discuss new information, e.g., the characteristics, distribution, and pathways. This paper conducts a systematic review of sources, pathways, characteristics, and health effects of indoor PM. In addition, practical steps to reduce exposure to indoor PM are also discussed.Indoor particulate pollution is grouped into primary and secondary PM, based on the origin of the PM. Primary indoor pollutants are directly generated from indoor domestic activities such as cooking, biomass heating, tobacco smoking, washing, cleaning, and other indoor activities. Secondary PM includes pollutants infiltrated from the outdoor environment and particles generated due to chemical reactions between indoor precursors and outdoor sources [39,40]. It is well known that ambient (outdoor) PM is a significant contributor and determining factor of indoor PM levels. Other factors, including indoor-type homes, offices, and commercial spaces; ventilation arrangements (naturally provided by windows or mechanical ventilation); occupancy rate and time; endotoxin levels; and geographical location, play a critical role in defining the chemical composition and disease burden of indoor PM [41,42]. According to the study of Sumpter and Chandramohan, the majority of lower-income groups from developing and low income countries rely on solid fuels for cooking and heating [43]. Activities such as cooking and heating using biofuels (coal and wood) can generate significant indoor PM concentrations, especially PM2.5 and ultrafine particulate matter [44]. Additionally, indoor exposure to asbestos fibers in old houses has been highlighted as a significant concern [45,46]. Due to its high tensile strength and versatility, asbestos, including crocidolite, was once widely used in construction (roofing, floors, and walls) and manufacturing household items such as fireproof curtains. Over time, weathering releases microscopic fibers, which are highly fibrogenic to the human lungs upon inhalation [45]. Furthermore, human habits, such as frequent windows and other dust-generating indoor activities, result in crucial indoor PM sources. For example, human walking is an important factor causing the resuspension of indoor PM. During the movement of humans, soles were exposed to the air. Branis et al. found that indoor human activities in the classroom could also lead to resuspension of large particulate matter, especially for PM10 [47].Indoor cooking has been a well-investigated source of PM over the past decades. Studies have shown that cooking activities enable the emission of millions of particles (~106 particles/cm3) through oil, wood, and food combustion, and most of them are ultra-fine particulates [48,49]. In addition, cooking can lead to indoor PM emissions from cooking areas in homes, restaurants, and other building types (offices, schools, etc.) because high-temperature cooking can lead to water vapor and other solid and liquid particle emissions.High emissions of indoor fine particulates (aerodynamic diameter <100 nm) occur during frying and boiling [50]. Zhang et al. showed that average concentrations of ultrafine, fine, and black carbon particles emitted during boiling and frying ranged from 1.34 × 104 to 6.04 × 105 particles/cm3, 10.0 to 230.9 μg/m3, and 0.1 to 0.8 μg/m3, respectively [51]. Chinese-style cooking has been identified as one of the major sources of indoor PM. It contributes up to 33% of indoor PM levels (PM0.5-5) [52]. The PM exposure level of indoor populations due to cooking and heating is related to fuel type, stove type, and population type [53,54,55]. Cooking behavior, such as using different types of aerosols, fuels, and exhaust fans or stove hoods, is also closely linked to the level of indoor PM pollution. A study showed that, when no stove hood was used, indoor PM2.5 was at a sufficient concentration to affect potentially the health of children, even in non-cooking times [56].Tobacco smoke from cigarettes, water pipes, and e-cigarettes is also a well-investigated source of indoor PM, which threatens the well-being of both smokers and other occupants. Drago et al. showed that concentrations of PM2.5 and several toxic trace elements were higher in smoker dwellings than non-smoker dwellings [57]. Braun et al. studied the effects of tobacco strength, measured by quantifying the amount of tar, nicotine, carbon monoxide, and different additives on the amount of PM [58]. This study included five cigarette types with different tobacco strengths, with or without additives, and a reference cigarette. Studies have found that incense is an important source of polycyclic aromatic hydrocarbons (PAHs), carbon monoxide, benzene, isoprene, PM2.5, and PM10 [5]. Lee and Wang observed that PM2.5 emission rates of different incense types varied considerably [59]. PM levels were directly proportional to concentrations of residuals in the cigarettes. Compared to tobacco smoke, incense smoke contains higher concentrations of PM. Lin et al. showed that incense burning generates PM >45 mg/g compared to 10 mg/g from cigarette burning [60]. Kumar et al. investigated PM levels generated from indoor incense-smoking activities, such as pre-burning, burning, and post-burning phases [61]. Incenses that they studied included sandalwood and floral sticks, dhoops (dhoops are an extruded incense, lacking a core bamboo stick), and mosquito coils (a mosquito coil is a mosquito-repelling incense). They showed that the mean concentrations of PM during the burning phase were highest. Concentrations ranged from 1300 to 1880 μg/m3 for dhoops and from 214 to 259 μg/m3 for mosquito coils. The burning of floral incense had a higher PM concentration (700 to 854 μg/m3) compared to the burning of sandalwood incense (99 to 114 μg/m3). With the increasing popularity of IQOS and e-cigarette devices among adolescents and adults, more and more studies have been conducted to explore the relationship between PM2.5 and e-cigarette aerosol. Studies have shown that indoor PM2.5 concentrations can rise to 197–818μg/m3 during vaping [62]. The level of PM2.5 is comparable to or even higher than those found in conventional cigarettes. IQOS smoking had little effect on indoor fine particulate matter (>300 nm) concentration or PM2.5 concentration. However, the concentration of ultrafine particles (25–300 nm) can be significantly increased [63]. Overall, PM emission during the burning of dhoops was higher than PM emission during the burning of sandalwood or floral incense and mosquito coils. This study also showed a higher PM emission during the post-burning phase of dhoops than sandalwood and floral incenses and mosquito coils, which indicated potential exposure even after the cessation of the burning phase of dhoops.Besides indoor in-situ sources, outdoor PM can be transported into indoor environments through air movement. Particles in the outdoor environment enters the room through air flow, that is, through a combination of osmosis, natural (NV) and mechanical ventilation (MV). PM originates from a range of outdoor natural sources, including forest fires, soil dust, sea salt, the presence of pets and farm animals (i.e., animal debris), pollen, spores, plant debris, and bacteria. Anthropogenic outdoor PM sources, such as evaporative gasoline emissions from transport, oil combustion, and coal burning in power plants, contribute to indoor PM concentrations [2,31].Distribution characteristics of PM include size, mass, mass, or particle concentration, and chemical or biological composition. The mass of PM is a major characteristic that affects its distribution. Microscopic particles with low density can remain airborne for extended periods and move freely from source to surrounding areas, reducing indoor and outdoor air quality [64]. Indoor PM burden may range from 15 to 259 μg/m3 and from 3 to 202 μg/m3 for PM10 and PM2.5, respectively [65]. Factors such as geographical location, air exchange efficiency, penetration and deposition rate, occupancy rate, and the particles’ presence dictate indoor PM levels. Not all indoors are affected equally. For example, residents of a building living closer to ground level experience higher PM levels compared to occupants residing at higher levels [66].Similarly, individuals located in cities with frequent dust storms have elevated indoor PM levels and are more susceptible to PM-associated diseases compared to residents of other cities [67,68]. The effect of human exposure to PM is especially significant in urban environments, where higher population density leads to higher pollutant generation and higher human exposure [69]. Densely populated urban centers curb black carbon emissions from fossil-fuel transport, household stoves, and space heating. The negative effects of this growing pollution have been enhanced by the continuous movement of people from rural to urban areas. Ventilation of the confined spaces plays a critical role in regulating indoor PM levels. Spaces equipped with mechanical ventilation or poor air conditioning systems have high infiltrated PM2.5 levels from outdoors.Chithra and Shiva Nagendra measured the temporal characteristics of PM concentrations inside a room in a naturally ventilated school building located near a roadway in Chennai, India [70]. They found that, during working hours, the number concentrations of PM inside the room were 2.4 × 105, 2.2 × 103, and 8.1 × 102 particles/dm3 in size ranges of PM0.3–1, PM1–3, and PM3–10 µm, respectively. Putaud et al. showed a difference in the chemical profiles of coarse and fine particulate fractions of aerosol PM [71]. The coarse PM was made up of mineral composites (e.g., crustal species, fly ash, and minor elements), sea salt, and black carbon. Fine particulates contained ammonium sulfate, ammonium nitrate, and organic compounds. The formation of secondary aerosols is a major factor that affects PM pollution. Both secondary organic and inorganic aerosols can be dominant components of PM. Wang et al. found secondary inorganic species, including sulfate, nitrate, and ammonium, in PM air samples collected from four regions in China [72]. They also showed the production of secondary organic aerosol species in three regions and suggested that aqueous-phase processing and photochemical reactions were occurring and they were possible causes for the production of the organic aerosols.In addition to mass and size, biological composition is another important characteristic of indoor PM distribution. Indoor PM biological compositions, also known as bioaerosols, are solid or liquid particles carrying living organisms from biological sources, with sizes ranging from 0.1 mm to 100 mm in diameter [73]. Generally, their particle size distribution varies from the nucleation mode (<30 nm in vacuum cleaning condition) to the accumulation mode (~100 nm, indoor combustion aerosols from smoking, cooking, or incense burning), and to the fine and coarse modes (>1 µm, resuspension aerosols) [74,75]. They include biological allergens (e.g., animal dander and cat saliva, house dust, cockroaches, mites, and pollen) and microorganisms (viruses, fungi, and bacteria) [76,77,78]. Biological allergens, known as antigens, originate from a number of insects, animals, mites, plants, or fungi, and will induce an allergic state in reacting with specific immunoglobulin E antibodies. Indoor sources of allergens mainly include furred pets (dog and cat dander), house dust mites, molds, plants, cockroaches, and rodents [79], and there are outdoor sources as well [80]. Humans and animals are one of the dominant sources of bacteria in indoor environments, while fungi mostly originate from the outdoor environment [81,82,83].Both indoor and outdoor environmental conditions influence indoor PM levels. Indoor factors, such as temperature, humidity, and air exchange rate and efficiency, are some of the important ones that dictate the indoor PM levels. Outdoor factors, such as weather, wind velocity, temperature, humidity, and solar radiation, influence indoor PM levels and secondary aerosol formation. For example, changes in temperature affect PM by influencing the change of chemical reaction rates and atmospheric mixing heights that affect the vertical dispersion of pollutants and modifying local wind and flow patterns that control the transportation of pollutants [84]. Meanwhile, differences in temperature indoors and outdoors also influence natural ventilation through the movement of air, and thus affecting indoor PM concentration [85]. With the increase of relative humidity, the resuspension rate of fine particulate matters decreased [86]. Several studies have demonstrated that air exchange rate has a significant effect on indoor PM concentrations under stable outdoor PM concentrations. In general, the higher air exchange rate was, the lower the indoor PM concentration was [87]. Seasonality is a major factor influencing the distribution of indoor PM. Indoor PM2.5 concentration is related to season and building type. The PM concentration in the heating season was significantly higher than that in the non-heating season [88,89]. Epidemiological data from urban and rural areas in developing and developed countries show a steep rise in airborne PM during autumn and winter compared to spring and summer. Although advances in technology have provided safe, conventional (electrical) cooking and heating methods in developed countries, the majority of the population from urban and rural areas in developing countries still relies on traditional heating (burning wood and coal) to keep homes warm. Huang et al. studied three urban areas of China (Beijing-Tianjin-Hebei) between 2013 and 2017 and showed that the average concentration of outdoor PM2.5 in the springtime (July) was 38.76 mg/m3; by winter (January), levels had risen to 133.10 mg/m3, which was 5.3 times above WHO air quality standards [90,91]. The use of conventional heating methods, such as gas and electricity, reduced indoor levels of PM2.5 in heating and non-heating seasons by 43% and 70%, respectively [54].Meteorological factors during different seasons, such as wind velocity, precipitation, air and soil temperatures, and atmospheric and soil humidity, have noticeable relationships with the PM distribution. Deng et al. measured PM at heights of 121 and 454 m on the Canton Tower in China and showed that the vertical distributions of PM decreased with height [92]. Chen et al. studied the effects of SO2 and NH3 on secondary aerosol formation from unburned gasoline vapor and found that an increase in SO2 and NH3 concentrations (from 0 to 151 ppb and from 0 to 200 ppb, respectively) promoted the formation of secondary aerosols by a factor of 1.6 to 2.6 and 2.0 to 2.5, respectively [93]. They also reported that new particle formation and particle size growth were enhanced in SO2 and NH3. Increased solar radiation and intensity promote photochemical reactions that lead to the formation of secondary aerosols, thereby accelerating PM pollution and its distribution in the environment [72].PM possesses unique characteristics, such as a small aerodynamic diameter, a large surface to volume ratio, a complex chemical profile, and a high toxicity index, which depend upon the origin and source of the PM. The large surface to volume ratio provides a natural platform for reactive chemicals and ionic species to undergo oxidative and reduction reactions (RedOx). The large surface area of the particles can harbor surrounding environmental pollutants and acts as a “carrier” for heavy metals (e.g., Cr, Pb, Hg, Cd, and As), organic pollutants such as polyaromatic hydrocarbons (PAH), heterocyclic amines, and inorganic minerals (e.g., Si, Al, Fe, Mn, Ca, Cl, and Zn) [89,94,95]. The microscopic size and aerodynamic diameter allow the particles or particle-bound contaminants to remain airborne and drift along with the winds to reach extreme distances. The chemical composition of indoor PM is complex because it comprises particles from indoor and outdoor origins. Indoor activities, such as cooking using solid fuels including coal, wood, dung, and kerosene, can liberate black-carbon soot (elemental carbon) and organic carbon (bound carbon). Other major factors affecting both indoor and personal exposure to PM include the penetration factor, air exchange rate, and particle deposition and sedimentation, as well as human behavior [96]. Hung et al. studied indoor PM2.5 and PM10 in office spaces and showed an increase of 0.211 μg/m3 and 0.226 μg/m3 per 1 μg/m3, respectively, over outdoor levels [97]. Climatic and environmental factors affect both indoor and outdoor PM levels and composition.Annual average levels of PM10, PM2.5, and PM1 at observational heights of 121 m and 454 m above ground were 44.1, 38.2, and 34.9 μg/m3 and 35.7, 30.4, and 27.5 μg/m3, respectively [98]. Spatio-temporal studies conducted on PM10, PM2.5, and PM1 levels in Seoul (Korea) metropolitan subways showed a vertical distribution (going upwards from the ground) in the sequence PM10 > PM2.5 > PM1, and levels of fine particulates were uniform compared to coarse particles [99]. Moreover, the distribution was constant in summers compared to other seasons. Air quality monitoring studies conducted in confined and crowded spaces such as underground subways and metro stations often record a several-fold increase (4- to 6-fold) in indoor PM levels compared to ambient levels [99,100,101]. Human activities, such as primary or second-hand smoking, add airborne PM and nicotine to indoor microenvironments [9].As shown in Figure 4, potential sources for indoor PM include bioaerosols (plant, animal, bacteria, fungi, and viruses), combustion of fuel used for cooking and heating, and home or personal care products [102,103,104]. The diameter of these particles is in a range of 0.001–2000 µm. Even though some consider that the amount of PM is more significant indoors than outdoors, the concentrations vary with location (e.g., urban or rural or proximity to roadsides) and with socio-economic factors of the population [104,105,106,107]. Smoking, cooking using gas and wood stoves, and cleaning are the major sources for elevated indoor levels of PM10 and PM2.5 [108]. This PM can enter the human body through inhalation, dermal absorption, or ingestion [106,109,110].Respiratory absorption is the most common route of exposure. Both in an outdoor or indoor environment, PM in the air can get into the body via breathing through the nose, which is able to filter large particles, or the mouth, which is unable to perform that particular task [111,112,113]. Ample evidence demonstrates that the “personal cloud,” which is resuspended house dust, is one of the major pathways of exposure for inhalation [113]. Passing through the pharynx and larynx, air-containing particles enter the trachea, which is connected to the left and right primary bronchi. Primary bronchi are further subdivided into bronchioles leading into alveolar ducts and sacs in the lung. It is estimated that particles up to 100 μm in aerodynamic diameter (Dae) present in the inhaled air can be deposited in the respiratory system [114]. When inhaled, the particles in the air come to equilibrium with body temperature and humidity, and, as a consequence, the movement of large particles becomes restricted [103]. Depending on the size of the inhaled particles, they are categorized as “extrathoracic,” which cannot pass the larynx; “thoracic,” which are particles that reach beyond the larynx (relates to particles with Dae < 10 μm); and “respirable,” which are particles that manage to pass into the pulmonary or alveolar region (relates to particles with Dae < 4 μm) [114]. Fine particle pollution can enter the body through inhalation airflow, cross the respiratory tract and reach the alveoli, where it triggers an inflammatory response that reduces the immune system’s ability to respond. Further, once in the lungs, PM can enter the bloodstream and spread to other organs.The direction of the airflow within the respiratory system and the velocity of inhaled air impact particle deposition and diffusion. At the point of inhalation and when air comes into the tracheobronchial region, the linear velocity of the inhaled air can be relatively high. However, compression of residual gases within the alveolar sacs slows down the airflow speed, and, therefore, the velocity of the air in the alveolar region is minimal [110]. PM is composed of both soluble and insoluble particles, and it is assumed that the sizes of the particles get altered in the alveolar region due to their hygroscopic nature. Depending on the modifications to the PM, these could either be deposited in the lung or get removed with exhaled breath [115].The United States Environmental Protection Agency (USEPA) states that only PM < 10 μm has the ability to be deposited in the trachea-bronchial and alveolar regions, and, consequently, causes the maximum danger by way of inhalation [116]. However, during exercise, breathing is mostly done by the mouth, and, as a result, larger particles (PM > 10 μm) can be deposited in the tracheobronchial airways [103]. Madureira et al. reported that 3-month old infants had 4-fold higher inhalation doses of ultra-fine particles than their mothers [112]. PM10 was predominantly deposited in the head region, whereas deposition of PM2.5 and ultra-fine particles occurred in the pulmonary area. Lower right lobes demonstrated a high susceptibility to respiratory problems because they received higher PM deposition than upper, lower left, and middle lobes. Meanwhile, the elderly is also a group susceptive to particulate matter, since they spend more time indoors. Almeida-Silva et al. used computational models to measure particle transport and deposition in the human respiratory tract of the elders [117]. The results shown that after 5 years of continuous exposure to the average particle concentration, 258 mg of all particles are deposited on the surface of the alveoli of which 79.6% are cleared, 18.8% are retained in the alveolar region, 1.5% translocate to the hilar lymph nodes, and 0.1% are transferred to the interstitium. Additionally, Segalin et al. found that respiratory deposition of PM2.5 was almost 25% higher in male than female elderly [118].Whether or not bioaerosols are inhaled depends on their infectivity, airborne concentration, immunogenicity, and particle size [119]. Carpeted homes have higher concentrations of Al, As, and transition metals such as Cd and Cr than non-carpeted homes, which indicates the possibility of inhalation exposure of these elements [113].Respiratory absorption can be summarized as follows: deposition of particulate matter (PM10) in the upper respiratory tract, fine particles (PM2.5) in the lower respiratory tract, followed by ultrafine particles (UFPs < 100 nm) in alveoli. Few scientific data exist concerning human toxicity from inhaled fungal toxins [102]. Methodologies for measuring indoor bioaerosol exposure and health risk assessment are not well standardized and, therefore, additional studies are needed on exposure-disease and dose-response relationships in humans.Skin, which covers the entire human body, is considered the largest organ of the body. It is made up of three main layers: epidermis, dermis, and hypodermis [120]. PM in indoor air can be deposited easily onto the skin or absorbed through the skin because it is exposed to the environment [121]. Dermal exposure also depends on dust adherence to skin, which is adapted from guidelines on dermal contact with soil and varies from 0.004 to 0.01 mg/cm2 for different parts of the body of children indoors and from 0.02 to 0.80 mg/cm2 for adults based on activity. Indoor environments are rich in semi-volatile organic compounds (SVOCs). Garrido et al. and Weschler et al. showed that the concentrations of SVOCs absorbed via direct air-to-skin dermal uptake could be comparable inhalation intake [39,122]. Williams et al. showed that perspiration-induced absorption of pesticides into the human body occurred through the dermal contact [123]. The epidermis stratum corneum is the outermost skin layer and is the primary barrier that prevents environmental pollutants from entering the body. Skin is the first defense immune system that protects the human body from toxic substances [120]. PM gets absorbed by percutaneous penetration, which causes local toxicity in the skin and systemic toxicity in other organs [124]. The four different ways that PM penetrates the skin are by mechanical delivery, an intracellular route, a transcellular route, and through the trans-follicular route [120]. Hair follicles that create pores in the skin help the PM to get into the body through the trans-follicular route [125].The main pathways of human exposure to PM are skin exposure to PM and particle ingestion, hand-to-mouth behavior, and food containing. However, both the rate of particle ingestion and the size of particle ingested are highly uncertain. Finer particles can stick to the hands, so both ingestion and skin contact are major problems. However, there is little consensus on how small the particulates should be to stick to your hands, since the airborne particles range in size over five orders of magnitude (from about 0.001 μm to about 100 μm) [41]. Unsurprisingly, children consumed much higher rates of particulate matter than adults because they liked to play on the floor and put their hands and non-food objects in their mouths more frequently. This behavior, combined with their smaller body size, makes exposure to chemicals through PM more important for children than for adults [126].Absorption of PM by the digestive system can occur in two ways: directly by diet (direct consumption of food and drinks that are enriched with PM) or indirectly into the gastrointestinal tract through the expulsion of particles removed from the lungs via mucociliary transport [127,128,129,130]. It is estimated that roughly 50% of the inhaled dose could reach the intestinal tract. Therefore, researchers have pointed out the necessity of recognizing ingestion as an essential human exposure pathway to PM pollution. Research needs to be conducted on both the ingestion mechanisms and constraints to ingestion [110]. The absorption of this PM can be the epithelial lining of the small intestine, colonic epithelium, or the stomach. Researchers have identified links between the ingestion of PM and different diseased conditions in the digestive tract, but, at present, data on the effect of PM on the digestive tract are elusive [128]. More research is needed to identify how the chemical nature and sizes of the particles affect the rate of absorption of PM along the digestive tract.Wang et al. estimated the ingestion of tetrabromobisphenol A and eight bisphenol analogs, including bisphenol A, in 12 countries [131]. The highest median estimated daily intake (EDI) of bisphenols through dust ingestion was observed in Greece, Japan, and the U.S. They showed that the EDI for infants and toddlers was high, which indicated that dust ingestion was a significant exposure pathway.It is difficult to quantify the amount of PM a person gets into the body by staying indoors. Whether the uptake is through inhalation, dermal exposure, or absorption through the digestive tract is a personalized matter, which varies with individual human beings and the particular indoor environment (home, school, office, or another type of working place) and the length of stay in the environment. The nature of the chemical and biological constituents of the PM also needs to be considered.The chemical composition of indoor PM is determined by the sources of PM and chemical processes that occur both indoors and outdoors. The primary constituents of PM include inorganic metallic compounds, organic compounds of biological origin, inorganic carbonaceous material (including black carbon and elemental carbon), sulfate, nitrate, ammonium, and other ions (Table 1).Chemical compounds in PM are grouped into various categories that consist of organic carbon (OC), elemental carbon (EC), carbonate carbon (CC), non-sea salt sulfate (NSS-SO42–), nitrate (NO3–), ammonium (NH4+), sea salt, mineral dust, and non-dust elements [140]. Generally, the NSS-SO42 fraction is estimated from the difference between the total sulfate and the sea-salt fraction of SO42–. Sea-salt concentrations are estimated from soluble sodium concentrations [141]. Mineral dust components are determined by summing Al2O3, SiO2, CO32–, Ca, Fe, K, Mg, and Mn. Non-dust elements include common trace elements (i.e., Cu, Ni, Pb, V, and Zn), and they are generally derived from atmospheric origin or attributed to atmospheric pollution [142]. Typically, indoor PM consists of approximately 50% organic carbon, 3% elemental carbon, 30% sulfates and nitrates, 15% ammonium ion and water, and 1% total metal content, with more than two-thirds of that being iron.Wang et al. noticed that NSS-sulfate in PM10 and PM2.5 contributed 95% of the total sulfate in Guangzhou, China, suggesting a significant anthropogenic origin of the PM [143]. The primary sources of the sulfate in Guangzhou may be attributed to the release of SOx from sulfuric-acid-manufacturing industries and the generation of sulfate compounds in coal-fired power plants and their subsequent utilization in construction and agricultural industries. Some fraction of the SOx released through anthropogenic sources is oxidized to secondary aerosols of sulfate compounds [144].The sources and composition of PM vary with PM particle size. PM10 consists predominantly of insoluble, Earth-crust-derived compounds such as iron and aluminum oxides; biological materials such as pollen, fungi, and bacteria; and sea salts. PM2.5 is derived mainly from combustion-related sources and consists of carbon, hydrocarbons, and sulfur and nitrogen [2].In addition to PM, indoor air and ambient air also contain a range of gaseous pollutants that include carbon monoxide, sulphur dioxide, and nitrogen oxides derived from both indoor (e.g., combustion) and outdoor (e.g., bushfire) activities. Indoor PM is generally enriched with chemical additives, such as phthalate plasticizers, organophosphates, brominated flame retardants, and fluorinated surfactants used in products that are part of the indoor environment. These pollutants associated with PM cause health effects when the indoor occupants inhale PM.The chemical elements in PM can be divided into two major groups: Earth-crust elements derived from soil (i.e., soil tracers) and introduced elements derived from human activities (i.e., anthropogenic tracers). Earth-crust elements include Na, Al, K, Mg, Ca, Fe, Ti, and Mn, which are derived primarily from geological sources, whereas V, Cr, Cd, Ni, Cu, Pb, Zn, As, Sn, and Se can be derived from both natural Earth-crust sources and anthropogenic sources [145]. These elements can be used to fingerprint the anthropogenic sources of indoor PM.Several methods are used to fingerprint the most likely indoor PM [146,147]. For example, enrichment factors are calculated for individual elements in terms of their average concentration in the Earth’s crust. Aluminum is commonly used as a reference for these elements because of the minor contribution of Al as a potential pollutant and its major contribution to the Earth’s crust. The enrichment factor (EF) of an element E is defined according to Equation (1):EF = (T/R)air/(T/R)crust(1)
+      where T and R represent the concentrations of the tested and the reference element, respectively, if the EF approaches 1, then the Earth’s crust is considered as the dominant source of the tested element. Considering the variation in the Earth-crust composition (Table 2), EF > 5 in PM indicates that non-crustal anthropogenic sources contribute a significant portion of PM’s element.For example, Alves et al. measured high EF values (>10) for Pb, As, Cu, and Zn in both indoor and outdoor PM, which suggested that there was a contribution of anthropogenic activities to both the outdoor and indoor environments [149]. The data also indicated that air infiltration affects the nature and composition of indoor PM. Furthermore, the mean outdoor concentrations of many of these metals for both coarse and fine PM exceed that of indoor PM, which indicates that air infiltration makes a significant contribution in lowering the enrichment of these metals in indoor PM [141].Wang et al. noticed that Cd and Se exhibited the highest enrichment factors (>10,000) in indoor PM collected from the Guangzhou region. Pb, As, Sn, and Zn also showed high enrichment factors (>100) [143]. Ni, V, Cr, and Cu appeared to be moderately enriched (10 < EF < 100). The high enrichment of these elements indicates that non-crustal sources, including pollution emissions, contribute primarily to elemental loading in indoor PM. For example, the high ER for Cr reveals a range of pollution sources, including coal combustion and tannery sludge incineration [151]. While non-crustal V is derived mainly from heavy fuel oil combustion, metal smelting, and fossil fuel combustion are the likely sources of non-crustal volatile metals such as Cd, Zn, and Pb [152]. Furthermore, elements with high EF values such as Cr and V generally have low concentrations in PM. Elements originating from the Earth’s crust and sea salt, such as Na, Mg, K, Ca, Ti, Mn, and Fe, have high concentrations in PM, but low enrichment factors (<5), which indicates an insignificant contribution from anthropogenic sources of these elements in PM; airborne Earth-crust dust is the primary source of these elements in PM [150].In the presence of indoor sources, PM levels can rise rapidly to several orders of magnitude greater than outdoor levels [153,154,155]. As a potential health hazard, indoor exposure to PM has received increased attention in recent years because people spend almost 90% of their time indoors [153]. Brauer et al. have found that 99% of the population in south and east Asia live in areas where the WHO Air Quality Guideline for PM2.5 is exceeded [156]. Exposure to indoor PM has been identified as the cause of respiratory infection, allergic symptoms, cardiovascular disease, adverse birth outcomes, and neurological and cognitive disorders [157,158,159]. Epidemiological studies have found that mortality and morbidity of respiratory diseases rose as the PM concentration increased [160,161]. Long-term exposure to PM less than 2.5 μm in diameter (PM2.5) is associated with chronic conditions such as cardiovascular and respiratory diseases and cerebrovascular complications, leading to reduced life expectancy [162]. Short-term exposure to PM2.5 can also cause a variety of health impacts including exacerbation of asthma and increases in respiratory and cardiovascular hospital admissions and mortality [162,163]. Additionally, it was also found that mortality generated by short-term PM2.5 exposure was influenced by season, region (urban and rural), and co-pollutants [164].The main components of indoor PM are inhalable, which can penetrate the chest area of the respiratory system and cause adverse health effects [13]. Exposure to indoor PM will affect lung development, especially in children, including reversible deficits in lung function, chronically reduced lung growth rate, and a deficit in long-term lung function [13,165]. Moreover, some toxic pollutants, such as heavy metals and polycyclic aromatic hydrocarbons (PAHs), are also attached to the surface of indoor PM, and they pose a severe threat to human health [166,167]. PAHs adsorbed on the surface of PM have been shown to have carcinogenic and mutagenic effects [168]. Heavy metals and PAHs, individually or in concert, damage the double helix structure of DNA, leading to genetic mutations. MAC releases cytokines, such as growth factors, which alter the cell cycle and cause cells to divide forever. Thus, tumors can form. A survey has confirmed the presence of particulate polycyclic aromatic hydrocarbons (PAHs), such as bingo pyrene, in the air of Beijing. All air samples are highly mutagenic [169]. Even at low doses, heavy metals can have severe effects on neurodevelopment [170]. To control its adverse health impacts, it is essential to explore the composition of indoor PM. Studies have found that the elderly, children, and pregnant women are more susceptible to PM than others and PM concentrations vary in different indoor environments. As such, the further discussion of this issue is of great significance for removing indoor PM and reducing the exposure of susceptible groups to PM.In general, toxicity of particulate matter refers to the absorption and distribution of chemical components of the particle, which, in addition to carcinogenicity and mutagenicity can cause adverse health effects throughout the body [171]. According to recent studies of indoor PM done at different places [172,173,174], the components of indoor PM can be roughly divided into metals (e.g., Fe, Ni, Zn, and V), inorganic compounds (e.g., sulfate, nitrate, and ammonium), and organic compounds (e.g., PAHs, volatile organic compounds (VOCs), and soot). Numerous studies have reported that some trace elements (e.g., Fe, Ni, Zn, V, Pb, As, Se, Cd, and Hg) may cause damage to cells, tissues, proteins, and DNA and change cell permeability by inducing the production of reactive oxygen species (ROS) [175,176,177]. For example, an extensive increase in ROS can be caused by Fe in vivo, causing oxidative damage and inflammation [178]. Gilli et al. found that oxidative DNA damage attributed to PM is also related to the Fe content [179]. Magnani et al. studied the effect of transition metals in air particles on pulmonary oxygen metabolism and found that Ni could cause metabolic changes [180]. Additionally, Cu, K, Mn, Zn, V, and Ni are associated with increased odds of hospital admission for cardiovascular disease, which can lead to death in severe cases [181,182,183,184].Inorganic substances are also common pollutants in indoor air, among which nitrate, sulfate, and ammonium particles are representative. Most sulfate and nitrate in PM originate from the atmospheric oxidation of SO2 and NOx emissions, mainly in the form of aerosols (e.g., (NH4)2SO4, NH4HSO4, NH4NO3, and partially neutralized salts). Ammonium ions are mainly involved in the neutralization reaction of sulfuric acid and nitric acid. Sulfate in the atmosphere can increase the deposition of toxic compounds in the lungs, affecting breathing conditions. A 1% variation in sulfate was associated with 0.117% variation in respiratory diseases [14,185]. Further, sulfates increase ROS levels, and long-term exposure to sulfates may cause oxidative stress, increasing the risk of many vascular diseases [184]. In addition to sulfates, it was found that the increase of acute cardiovascular hospitalization rate was also related to the increase of nitrate concentration [186]. Additionally, the mortality rate of the elderly was significantly associated with nitrate concentrations [187,188].In addition to the inorganic component, the organic part of PM is formed in a complex and poorly understood way, and it is also known as organic aerosols, such as PAHs, VOCs, and soot [174,189]. PAHs, which are rich in carbon and hydrophobic, easily cross cell membranes, and quickly enter cells. Then, the PAHs from a harmful intermediate inside the cell with a ring of active epoxides, and, when a gene mutates under the circumstances of a gene polymerase error, the PAHs randomly select locations in the genome, in some cases leading to cancer [190]. Long-term exposure can affect women’s reproductive health, cause proteinuria, and even lead to lung cancer [191,192,193].Another critical component, VOCs, are organic substances that can easily evaporate under ambient air conditions, and most of them are toxic. The VOCs can cause skin irritation, cancer, respiratory diseases, chronic obstructive pulmonary disease, bronchial asthma, and systolic plus diastolic hypertension [194,195,196]. The increased concentration of VOCs will lead to decreased respiratory and lung function in children and cardiopulmonary dysfunction in susceptible populations [197].The levels of PM found in the indoor environment are mainly brought in by ventilated airflows or produced by burning indoors for heating and cooking. Soot is formed through a series of reactions in which small free radicals (e.g., OH, O, H, CH, and CH2) cause chemically induced combustion and fuel decomposition, producing larger hydrocarbon free radicals and PAHs. Some soot is cytotoxic and has adverse effects on cardiovascular and lung health. PM penetrates deep into the lungs and enters the bloodstream, causing high blood pressure and damage to blood vessels. Once in the bloodstream, PM can spread to other organs, such as the heart, damaging their cellular structure and function [198].Compared with PM10, PM2.5 with smaller particle sizes has larger specific surface area and larger adsorption capacity, and toxic heavy metals are more likely to bind to PM2.5 [199], and so do acid oxides, organic pollutants and pathogenic microorganisms. Dacunto et al. found that the proportion of trace metal (transition metal) elements and carcinogenic polycyclic aromatic hydrocarbons in PM2.5 was almost twice that of PM10. At least 60% of PM2.5–PM10 is reported to be deposited on the outside of the chest [172,200].Particles with a size ranging from 1 to 2.5 μm mainly deposited in bronchial and alveolar, and some particles remained in lung tissue for a long time, forming lung interstitial lesions. PM0.1 can invade alveolar and stay in it, and then quickly enter blood circulation system through breathing, and finally flow into human kidney, liver, heart, brain, and other organs. In conclusion, PM2.5 is more harmful to human health than PM10.Because of the different occupational and physical characteristics of people, their susceptibility to PM is also different. Therefore, it is necessary to explore the different reactions of diverse populations to PM.Most epidemiological studies have focused on the health effects of indoor PM on susceptible populations, namely pregnant women, children, or the elderly [138]. For pregnant women with long-term exposure to high levels of PM, the developing human fetus may be at risks, such as adverse birth outcomes, preterm birth, term low birth weights [159,201], congenital disabilities, stillbirths, and respiratory disease [202,203,204]. Epidemiological studies have shown that exposure to pollutants in the indoor environment is associated with respiratory diseases in children, such as wheezing, asthma, and rhinitis, and it can lead to disease in later life [205,206,207]. Mousavi et al. showed that exposure to air pollution in childhood increased the susceptibility to Alzheimer’s disease and Parkinson’s disease in adulthood [208]. Not just for children, but also for the elderly, exposure to indoor PM may be the most significant public health burden in terms of risk to health [209,210]. Chen et al. studied residential areas with long-term exposure to PM2.5 and PM2.5–10 and found that the decline of lung function in the elderly (aged 65 years or older) was related to PM [211]. Recent studies suggest that PM is associated with low bone mineral density and osteoporosis-related fractures [212]. Exposure to PM is associated with cognitive deficits, oxidative stress, neuroinflammation, and neurodegeneration [213]. Several metals, including aluminum, arsenic, cadmium, lead, manganese, and mercury, have been shown to affect the nervous system, while the general accumulation of metal ions in the brain can exacerbate oxidative stress and neuronal damage [214].In addition to the relatively severe impact on susceptible populations, there is damage to people who, unavoidably, have long-term exposure to high PM concentrations. Indoor cooking has been considered one of the most important indoor PM [215]. Studies have shown that emissions from cooking can harm human health, leading to lung toxicity, immune-toxicity, genotoxicity, and potential carcinogenicity in the human body [216,217,218]. Particularly for individuals exposed to indoor cooking fumes, such as cooks, workers, and restaurant customers, health will be adversely affected [219]. Pan et al. suggested that kitchen staff is more likely to have oxidative stress than service area workers [220]. Bigert et al. found that female cooks, restaurant and kitchen assistants, and wait staff showed a statistically significant increase in myocardial infarction risk [221]. Welding can produce high amounts of fumes containing ultrafine particles with Mn [222]. Racette et al. observed that some might develop Parkinson’s disease 17 years earlier than the general population in welding populations [223].Studies have shown gender differences in indoor PM exposure. Sears et al. studied people living near coal-fired power plants with coal-ash-fly-storage facilities and found that women were more susceptible than men to impaired cognitive ability and associated with PM exposure [224]. Additionally, Gregory et al. disclosed that the distribution of multiple sclerosis prevalence was related to PM10 concentration, especially in women [225]. In offices, female employees, and particularly those suffering from allergies, reported more sick-building-syndrome symptoms (e.g., sneezing, cough, tiredness, and irritability) than their male counterparts [226]. Contrary to the above, men are more likely than women to develop symptoms. In exploring the effects on the human brain of aerosols produced by exposure to electric frying, it was found that the aerosol response of the brain to electric frying occurred in males rather than females [227]. Weichenthal et al. analyzed the relationship between PM2.5 and non-accidental cardiovascular mortality and found that male cardiovascular mortality may be related to PM2.5 exposure, while female cardiovascular mortality had no similar association [228]. The differences in PM symptoms between men and women may be related to gender-specific behavior patterns [229].The range of the effect of PM on health is broad, but it mainly affects the respiratory tract in children and cardiovascular function in the elderly. The impact of PM on different populations shows a need to improve health in the general population, to control the sources of PM, and to raise awareness of the potential impact of household pollutants.To reduce exposure to indoor PM and its adverse health impacts, many national organizations and influential worldwide committees (e.g., WHO) have stipulated mass standards and guidelines that coincide with desired indoor air quality. Indoor air standards to evaluate an acceptable quality of air are generally defined by various agencies, causing significant regional differences. Dai et al. installed long-term IAQ sensors in 117 homes in all climate zones of China in order to able to measure indoor PM2.5 and CO2 concentrations consecutively [230]. Using this indoor and outdoor data, local polynomial regression fitting was used to determine the relationship between indoor CO2 and PM2.5 concentrations and the corresponding outdoor parameters. Maleki et al. tested the potential of an artificial neural network (ANN) algorithm to estimate hourly air pollutant concentration parameters and two air quality indices, the air quality index (AQI) and the air quality health index (AQHI) [231]. Air quality often has negative health, socio-economic, agricultural, and political consequences. Meteorology and pollution sources are two basic factors affecting air quality. The method can be used to predict the spatial and temporal distribution of pollutants and air quality index. Main standards and guidelines formulated by global institutions are summarized in Table 3, and they guide the development of effective strategies.In China, the previous Indoor Air Quality Standard (GBT 17095-1997) only provided a limit for PM10 and stipulated that the daily permissible maximum concentration of PM10 should be 0.15 mg/m3. With increased knowledge about indoor air pollution, PM with sizes <2.5 and 10 µm (PM2.5 and PM10, respectively) are now both listed as common indoor air pollutants. A standard and guidelines (JGJ/T 309-2013), issued on 2 July 2013, require that the daily average concentration of indoor PM2.5 should be <75 µg/m3. As early as 1996, Singapore formulated a control standard for PM10 in office air, and the value was 150 µg/m3 [233]. However, a control standard for indoor PM2.5 in Singapore is still lacking. In 2012, Canada established the Residential Indoor Air Quality Guidelines, which state that PM2.5 needs to be monitored, with a limit of 100 µg/m3 as a 1 h average (Short-Term Exposure) and 40 µg/m3 as an 8 h average (Long-Term Exposure). The National Ambient Air Quality Standard (NAAQS), set by USEPA, stipulates 35 µg/m3 and 15 µg/m3 for 24 h and annual periods, respectively, for exposure to PM2.5. For PM10, the US EPA [236] states that the permissible value is 150 µg/m3 over an average of 24 h. The United Kingdom, Finland, and Germany do not have PM2.5 monitoring [237]. However, these European countries follow the guidelines set by the WHO with values of 25 µg/m3 and 10 µg/m3 for 24 h and annual averages, respectively. Because various international institutions set these standards and guidelines, they are not in unity. Therefore, a comprehensive investigation should be carried out to develop uniform standards and practical strategies to mitigate exposure to indoor PM. Mass concentration predictions have a key role to play in making decisions about atmospheric resources [243]. This has had adverse health effects, such as high morbidity and mortality rates from cardiovascular and respiratory diseases. For this cause, it is crucial to avoid air pollution in advance by improving air quality protection and ensuring effective environmental monitoring. This is very critical for people’s daily safety and the government’s decision-making on air quality regulation.Current particulate removal technologies mainly involve filtration, adsorption, electrostatic dust removal, and the negative ion and plasma (NIP) method (Table 4). They have been applied in various aspects of life to improve air quality where humans live and work.Generally, filtration is typically installed to eliminate PM prior to other abatement technology [253]. Five principles for removal are used in filter technology, including interception, diffusion, inertia, gravity, and the electrostatic force (Figure 5) [244]. In particulate removal by filtration, bag type dust collectors, particle layer dust collectors, and carbon-based air filters are three typical, traditional methods [245,254,255]. Adsorption is often used in conjunction with filter-particulate removal in air purifiers to reach superior indoor air quality. Disadvantages include a compromised efficiency at a high relative humidity [256], the need for periodic replacement of adsorbents to prevent re-entry of waste pollutants into the atmosphere, and that airborne bacteria may thrive on carbon sorbents.Some carbon-based materials, e.g., activated carbon fiber and granular activated carbon, with the virtues of high density, low ash content, and high absorption capacity, have been widely used for removing molecules, gases, and vapors from indoor air [248].Electric-collection technology can be applied in filtration technology to improve the efficiency of filter-particulate removal. By applying electrostatic forces, air filters can capture PM without high-density micropores, thus leading to a sharply decreased filter thickness and higher filtration removal efficiency [257]. On the basis of these previous findings, a high-efficiency rotating TENG (R-TENG) enhanced polyimide (PI) nanofiber air filter was developed for PM removal in ambient atmospheres by Gu et al. [251]. (TENG stands for triboelectric nanogenerator.) Charged positively by the R-TENG, an electric field was formed around a stainless-steel mesh and PI-nanofiber, thus significantly improving the PM-capturing efficiency of the filter due to electrostatic filtration [258]. However, electronic filters can generate hazardous charged particles, an obstacle for the extensive application of electrostatic-particulate-removal technology [259].In addition to hazardous gases, indoor particles also can carry large amounts of harmful viruses and bacteria. Therefore, to ensure indoor human health and reduce noise pollution from filtration, new particulate-removal technologies that can provide such problems are imperative. Negative ion and plasma (NIP) methods are being developed due to their excellent ultra-fine particle removal and sterilization without making noise. The principles of the NIP methods are similar. Both are charged by an external force to neutralize the particles and coalesce them to form larger particles that settle [260]. Nowadays, the main application of the NIP method is plasma dust collection [252]. However, the PM is not removed but only adheres to the nearby surface, and it is easy to raise dust again. Therefore, the widespread application of the negative ion and plasma method is restricted.Rational and effective methods need to be taken to mitigate exposure to indoor PM for the sake of human health and a comfortable indoor environment. Based on the characteristics and sources of indoor PM, control strategies can be targeted at two sources: ambient PM and domestic PM.Building ventilationVentilation of households is crucial to bring indoor PM concentrations to levels similar to those of outdoors [261]. Specific measures should be considered, based on the ventilation types of buildings, which are identified as being natural, mechanical, or a mixed type of ventilation. Buildings with mechanical ventilation systems (air conditioners) show a minimum concentration of particles, which indicates that adequate household ventilation can improve indoor air quality. Building ventilation (interventions) is beneficial to mitigate indoor PM exposure, but they always incur a high energy expenditure cost. Therefore, an optimized building intervention model is needed to establish maximum efficiency and minimum energy cost [262]. When outdoor PM2.5 concentration is high, doors and windows should be closed. The key of the method is real-time monitoring of indoor and outdoor PM2.5. Also important is improving air distribution, increasing ventilation rate, fresh air (filtration) volume, regularly cleaning and disinfecting air-conditioning system, etc. In buildings with centralized or semi-centralized air-conditioning systems, the use of fresh air systems can reduce indoor PM2.5. Some experts believe that for public buildings, air purification facilities could be equipped in existing fresh air units, return air inlet, and blast pipes, and air cleaners also could be used directly there [261].Climate and seasonIn some regions, climatic and seasonal characteristics are distinct, such as in the east and west of the USA and southeast China. The variations can be used to mitigate exposure to indoor PM [263,264,265]. In China in the summer, the prevailing wind direction from the south will bring clean air from the ocean. As such, at this time, citizens should open windows and doors frequently to promote clean air exchange. In winter, windows and doors should be closed as far as possible. Industrial emissions must be reduced. The coal should be desulfurized before burning. Boilers of 30 t/h should be phased out as soon as possible. In North China, central heating should be promoted.Traffic and industriesWith rapid urbanization and industrialization, particles generated by traffic and industries have aggravated the indoor air environment. To handle air pollution in heavily trafficked areas, enough urban greenspace should be guaranteed to reduce indoor levels of PM, especially PM2.5 [266]. In addition, vehicle emission must be controlled, and emission standards must be strictly followed. Additionally, vehicles that use new-energy technology ought to be promoted. Urban greenery and other protective procedures should be enhanced where traffic is heavy. Finally, the upgrading of petroleum refining should be accelerated to prepare for the improvement of motor gasoline and diesel.Industrial and agricultural emissions are the two primary sources of particles that need to be reduced. Increased use of clean fuels and effective dust extraction is crucial so that industries and agriculture can reduce the generation and emission of PM. Removal of industrial dust is an issue that people are increasingly concerned about because of its health effects on humans, especially workers in factories. Nanofibrous filters, produced by electrospinning, have increasingly been used in air filtration products due to their high surface area and micro-porosity, which improve the entrapment of PM [267]. Liu et al. synthesized a transparent air filter by electrospinning, which achieved high ventilation and PM2.5 filtration efficiency (>95.0%) [268].SmokingThe first strategy is to control smoking. The government and relevant departments should actively promote the smoking ban and enhance the public’s health awareness. In public places, such as train stations and restaurants, there should be smoking and non-smoking areas and a reasonable division of functional areas to avoid cross-contamination. One study showed that Ontario’s smoking ban saved five to seven non-smokers working in bars $5–6.8 million a year [269]. With the development of e-cigarettes, more teens are jumping on the bandwagon and using them; even though levels of some potentially harmful ingredients from e-cigarettes are significantly lower than combustible cigarette, this does not mean that e-cigarette aerosols are ‘‘harmless vapour” as industry has claimed in the past [270]. The effects of SHA vaping exposure are largely unknown. E-cigarettes are not emission-free and their pollutants can reduce indoor air quality because the air exhaled by e-cigarette smokers contains dangerous chemicals. There is a potential health concern of SHA exposure via both respiration and dermal absorption. In particular, ultrafine particles formed from supersaturated 1,2-propanediol vapour can be deposited in the lung, while atomized nicotine appears to increase the release of inflammatory signaling molecule NO after inhalation.Overall, the increased use of e-cigarettes is worrisome and restrictions on tobacco marketing should be implemented to better protect the health of the general public. Future research should specifically focus on the long-term adverse effects of e-cigarettes on the cardiovascular system or respiratory diseases and cancer, as there is still a lack of strong evidence. There is no doubt, however, that quitting smoking is and will continue to be the most powerful way to prevent the cardiovascular and respiratory diseases caused by smoking and to protect people’s health [271].CookingNatural gas cooking has been identified as one of the predominant sources of indoor PM in developed countries [51,215,272,273,274,275,276]. Therefore, the development of methods to control PM generated from cooking is imperative. Kitchens must be ventilated to promote air exchange during cooking. Traditional solid fuels, such as straw, coal, and asphalt, must be replaced by clean energy as much as possible. Natural gas, methane, and electricity are good clean-energy options. Stoves must be maintained in functional order, especially in rural areas [277]. Cooking habits should be improved, and exhaust hoods are crucial for control of indoor PM during cooking [261]. Different cooking methods and ingredients, such as preferring safflower oil to olive oil. Adding salt and pepper early in cooking can also reduce the emission of PM.Indoor activitiesApart from cooking, indoor activities should be carried out to reduce PM. People should clean households regularly to remove the origins of indoor PM. Regular cleaning of the house, reasonable choices of decoration, as far as possible not keeping pets, and burning incense less, are considered good habits. Branco et al. demonstrated that improving the ventilation rate with convenient cleaning decreased the indoor PM concentration in a nursery from 79 ± 22 µg/m3 to 64 ± 15 µg/m3, which showed that regular indoor cleaning could control indoor PM [278].Indoor layoutInterior decoration, such as house plants, can improve indoor air quality and reduce indoor PM [266]. Previous studies have confirmed that the impact of woodland with rough surface on PM2.5 is much greater than that of grassland, while grassland has historically had a smaller effect on PM2.5 reduction [191,279]. However, tree-grass configurations contributed much more than tree only configurations for horizontal PM2.5 reduction, indicating that grassland is playing a contributory role in reducing atmospheric PM2.5. Vegetation has the irreplaceable role of adsorbing dust and removing harmful substances in the air. Leaf structure can be chosen for the deposition of dust [280,281]. Trees remove gas pollution mainly through leaf stomata, although some of the gas is removed from the plant’s surface. Once inside the leaf, the gas diffuses into the intercellular space and may be absorbed by the water membrane to form acids or react with the inner surface of the leaf.Trees also eliminate pollution by intercepting airborne particles. Some particles can be absorbed by the tree, but most of the trapped particles remain on the surface of the plant. The intercepted particle is often suspended back to the atmosphere, washed away by rain, or dropped to the ground with leaf and twig fall. A widely accepted view is that trees with large leaf area density and the turbulent air movement caused by their structure can capture particulate matter. Leaves provide surfaces for removing pollutants through wet and dry deposition, adsorption, and absorption. Local decreases of temperatures may modify the rate of chemical reactions, leading to decreased ozone concentrations [282]. Green plant species differ in their ability to purify particulate air pollution. Appropriate indoor green plants for air purification are orchids, caladium, and red back laurel. Leaf hairs can trap and absorb floating particles and smoke in the air. Active green walls use ornamental plants growing along the vertical plane, coupled with mechanical air induction, to actively attract polluted air through the plant growth substrate and leaves. Additionally, with improved living standards, urban residents might buy air cleansers to reduce indoor PM levels and improve indoor air quality. HEPA air purifiers are effective at reducing the concentration of indoor air particles [283]. HEPA filter-type small air purifiers demonstrate relatively high particle removal performance, based on the high single-pass collection efficiency of the HEPA filters (>99.97% for 0.3 μm particles).PM includes a mixture of solid and liquid particles suspended in air, and these particles can vary in size, shape, and chemical composition. Particles that are less than 10 μm in diameter (PM10 and PM2.5) are of major concern, because they are inhalable, thereby impacting the heart and lungs, and are associated with respiratory diseases, including asthma, chronic bronchitis, and acute bronchitis. Indoor PM is more harmful to special populations, such as the elderly, children, and pregnant women. Indoor PM originates mainly from combustion activities and the regular wearing of household furniture. It also is derived from outdoor sources, including dust particles. PM can be enriched with inorganic and organic contaminants, including toxic heavy metals and carcinogenic, volatile organic compounds. To reduce exposure to indoor PM and its adverse health impacts, the government and industry should formulate detailed and uniform indoor PM control standards based on abundant research. From a practical point of view, indoor air-purifier technologies involving electrostatic precipitation and filtration, as well as natural ventilation, are most commonly adopted to control concentrations of indoor PM. Additionally, regular indoor cleaning and suitable interior decoration, such as house plants and air purifiers, also significantly influence indoor air quality and reduce indoor PM.Given the current knowledge about the sources, distribution, pathways, characteristics, and health effects of indoor PM, the following research areas should be pursued:Development and adoption of advanced technologies, such as the tapered element oscillating microbalance (TEOM), X-ray fluorescence (XRF), and inductively coupled plasma mass spectrometry (ICP-MS), to quantify and fingerprint sources of indoor PM.Characterization and monitoring of bioaccessibility of inorganic and organic contaminants in indoor PM.Studies on mucosal interactions of indoor PM and associated contaminants concerning their toxicity.Development and evaluation of advanced PM removal technologies involving electrostatic precipitation to mitigate the health impacts of indoor PM.As soon as possible, the government and industry should formulate detailed and uniform indoor PM control standards, based on many investigations.Development and adoption of advanced technologies, such as the tapered element oscillating microbalance (TEOM), X-ray fluorescence (XRF), and inductively coupled plasma mass spectrometry (ICP-MS), to quantify and fingerprint sources of indoor PM.Characterization and monitoring of bioaccessibility of inorganic and organic contaminants in indoor PM.Studies on mucosal interactions of indoor PM and associated contaminants concerning their toxicity.Development and evaluation of advanced PM removal technologies involving electrostatic precipitation to mitigate the health impacts of indoor PM.As soon as possible, the government and industry should formulate detailed and uniform indoor PM control standards, based on many investigations.Conceptualization, L.Z., C.O., Y.Y. and N.B.; writing—original draft, L.Z., C.O., D.M.-A., K.S.V., T.P., H.W. and K.M.; writing—review and editing, M.V., Y.Y., N.B. and M.B.K.; visualization, L.Z., C.O., M.V. and Y.Y.; data curation, L.Z., H.W., K.S.V. and T.P.; funding acquisition, L.Z. and C.O.; supervision, N.B. All authors have read and agreed to the published version of the manuscript.This research was funded by “National Natural Science Foundation of China, No. 51708302”, “Natural Science Foundation of Jiangsu higher Education Institution of China, No. 17KJB610008 and No. 19KJD610002”, and “Open Funds of Jiangsu Key Laboratory for Biomass-based Energy and Enzyme Technology, No. BEETKC1906”.Not applicable.Not applicable.Not applicable.The authors declare no conflict of interest.Schematic representation of PM (Source reference: [4]).Size comparison of PM2.5 and PM10 against the average diameter of a human hair (~70 μm) and fine beach sand (~90 μm) (Source reference: [1]).Deposition potential for particles of varying sizes (Source reference: [1]).The main pathways of exposure to indoor PM.The five main capturing mechanisms for various kinds of PM (Source reference: [244]).Selected references on the composition of indoor PM.Note: Carbon include organic, elemental and carbonate carbon. The non-sea salt sulfate is calculated from the measured sulfate minus the sea-salt fraction of SO42–. Sea-salt concentrations are generally calculated from soluble sodium concentrations. Mineral dust is considered as the sum of Al2O3, SiO2, CO32–, Ca, Fe, K, Mg and Mn. Non-dust elements correspond to the sum of the common measured trace elements (i.e., Cu, Ni, Pb, V, Zn) other than geological ones.Selected references on the enrichment ratio of various elements in indoor PM.Earth crust elements or soil tracers and anthropogenic tracers; Earth crust elements: Na, Al, K, Mg, Ca, Fe, Ti and Mn; Non-Earth crust or anthropogenic: V, Cr, Cd, Ni, Cu, Pb, Zn, As, Sn, and Se.Standards and guidelines for PM2.5 and PM10.1a Guidelines for good IAQ in office premises (Singapore); 1b Ceiling Level: Highest possible allowed value for exposure (US, ACGIH); 2a Exposure: It means a continual and repetitive contact with the substance over a set period (US, ASHRAE).Application of Removal Technologies.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11056.txt

@@ -0,0 +1,2 @@
+In Myanmar, the escalating prevalence of type 2 diabetes mellitus (T2DM) and impaired glucose tolerance among adults was recently reported, with the highest prevalence in the Yangon Region. The aim of the present study was to identify the risk factors in dietary habits and their relationship with T2DM in urban Myanmar residents. We conducted a case–control study recruiting 300 individuals aged 25–74 years living in the Yangon Region, consisting of 150 newly diagnosed cases attending a diabetes clinic, and 150 controls, who were community residents and free of diabetes. The case group had a significantly higher consumption of noodles, fish, beans, fermented food and pickles, dried food, topping seasonings, and non-dairy milk products than the control group, whereas they had a lower vegetable intake (more than three servings/day) and fruit intake (more than three servings/day) than the control group. Furthermore, the case group exhibited a higher frequency of some dietary behaviors than the control group, such as (1) having meals with family, (2) skipping breakfast, and (3) eating out. The final model showed that topping seasonings (adjusted odds ratio (aOR) 11.23, 95% confidence interval (CI) 3.08–40.90), more than three servings/day of vegetable intake (aOR 0.18, 95% CI 0.05–0.67), and having meals with family (aOR 2.23, 95% CI 1.05–4.71) were associated with diabetes. The study suggests that Myanmar’s characteristic dietary culture of topping their meals with salty seasonings and sauces and eating multiple dishes together as a family are risk factors associated with T2DM. Our findings may contribute recommendations and opportunities for the primary prevention of T2DM in urban Myanmar.Diabetes mellitus is a common chronic disease and a primary global health concern. The worldwide increase in diabetes mellitus has been driven by global aging, economic growth, rapid urbanization, and nutritional trends in different income level countries [1]. A total of 463 million people (9.3% of adults 20–79 years) are living with diabetes worldwide; this number is expected to increase to 578 million (10.2%) by 2030 and 700 million (10.9%) by 2045 [2,3]. The incidence of diabetes mellitus has been rising more rapidly in low- and middle-income countries than in high-income countries, and 80% of the diabetes mellitus cases worldwide live in less developed countries and areas [4,5]. Between 2010 and 2030, the number of adults with diabetes has been projected to increase by 20% in developed countries and by 69% in developing countries [6]. Myanmar is the largest country in the region of mainland Southeast Asia. Recent studies have reported a high prevalence of type 2 diabetes (T2DM) among adults in the Yangon Region, which is the largest city in Myanmar [7,8]. In 2014, the national prevalence of T2DM in the adult population aged between 25 and 64 years in Myanmar was 10.5%, while impaired glucose tolerance was nearly two times that of diabetes mellitus (19.7%) [9]. Against this background, the Ministry of Health and Sports (MoHS) formulated the National Strategic Plan for Prevention and Control of Non-communicable Diseases (NCDs), a measure designed to work more effectively on reducing NCDs [10]. The plan has strengthened the medical service system and intervention at the community level to deliver a Basic Essential Package of Health Services to all citizens. However, the specific strategies still have gaps in facilitating lifestyle-related interventions to reduce exposure to key risk factors, including unhealthy diet and physical inactivity. The current study aims to fill this gap.T2DM develops in association with insufficient insulin action related to decreased insulin secretion or insulin dysfunction through insulin resistance, which involves multiple genetic factors. Insulin resistance is linked to lifestyle habits such as overeating, lack of exercise, and consequent obesity [11,12]. A significant cause of T2DM is an energy-dense Western diet combined with a sedentary lifestyle [13,14]. Based on this evidence, numerous studies have identified the risk factors for T2DM worldwide. In particular, there is a wide range of studies reporting on dietary habits, such as the type of food, food intake, nutrients, dietary patterns, and dietary behavior. Such evidence is still limited in Myanmar, where the culinary tradition is comprised of common Asian-style rice-based meals with family and multiple unique dishes of oily foods, while instant noodles and fried snacks are popular street foods.The similarities in previous studies on the association between dietary patterns and T2DM are summarized as follows. Healthy dietary patterns characterized by high consumption of foods such as fruits, vegetables, fish, poultry, and whole grains have been associated with a reduced risk of T2DM, whereas unhealthy dietary patterns characterized by a high intake of foods such as processed and red meats, fried products, sweets and desserts, and refined grains have been connected to an increased risk of T2DM [15,16,17,18].There is increasing evidence that dietary behaviors such as skipping breakfast, speed eating, eating out, and having meals with family are directly associated with obesity and other chronic diseases, including cardiovascular disease and T2DM [19,20,21,22]. As dietary behavior is based on individual lifestyles, such as food selection, frequency, and intake, understanding the mechanism can lead to recommendations for the prevention and management of NCDs, including T2DM.Previous studies in Myanmar have provided several reports on dietary habits and NCDs, including T2DM. A study focusing on risk factors for NCDs in the Yangon Region showed metabolic risk factors, as well as moderate or high 10-year coronary heart diseases risk, were more common among urban residents, and behavioral risk factor levels tended to be higher among rural areas. In particular, it was found that rural areas consume more alcohol and consume fewer fruits and vegetables than urban areas [23]. Furthermore, in a report investigating the association between fruit and vegetable intake and risk factors for NCDs, a high intake of fruit and vegetables was associated with lower odds of hypertriglyceridemia among men and women [24]. A study comparing dietary intake, dietary patterns, and an abnormal blood glucose status of middle-aged adults in urban and suburban areas of Mandalay found that dietary intake was significantly different between urban and suburban areas. However, the association between food patterns and abnormal glycemic status was not definitively reported [25].Contrary to the potential importance of dietary nutritional behavior for the increase in T2DM in Myanmar, dietary habits have been only marginally evaluated. Therefore, we conducted a case–control study to identify the specific dietary habits of the adult residents in Yangon. The aim of the present study was to identify the risk factors in dietary habits and their relationship with T2DM in urban Myanmar residents. This was an observational study applying a case–control study design. It recruited participants who met the eligibility criteria with a ratio of 1:1 to investigate the relationship between dietary habits and T2DM in urban Myanmar. Participants in this study were recruited from December 2018 to February 2019 until the target sample size was reached. A total of 150 case participants and 150 control participants were enrolled from the Yangon Region.Case definition: cases met eligibility criteria (Figure 1) and were diagnosed within six months before data collection. Newly diagnosed diabetes patients attending diabetes clinics were identified by physicians and cross-checked with their medical records before being invited to interviews. Study sites for recruiting cases were a private diabetes center and a national general hospital in Yangon where T2DM diagnosis adhered to the World Health Organization (WHO) diagnosis criteria of diabetes [26].The eligibility criteria for the case participants were: (1) patients who were newly diagnosed with T2DM within the six months prior to data collection (after June 2018); (2) adults aged 25 to 74 years; (3) either sex; (4) diagnosed by fasting plasma glucose ≥ 126 mg/dL and/or random blood glucose (RBG) ≥ 200 mg/dL, with or without the osmotic symptoms of diabetes (polyuria, polydipsia, thirst, bodyweight loss); and (5) all races and religions, and residing in Yangon [27] (see Figure 1).Control definition: controls were community controls randomly recruited from four townships in Yangon. Controls were defined as community residents who are eligible to participate in the study according to the criteria and free of diabetes.The eligibility criteria for the control participants were: (1) adults aged 25 to 74 years; (2) either sex; (3) individuals who do not have a history of diabetes or taking any diabetes medication, such as oral antihyperglycemic agents or insulin; (4) fasting blood glucose < 110 mg/dL and random blood glucose (RBG) < 140 mg/dL); and (5) all races and religions, and residing in Yangon [27] (see Figure 1). The sample size was calculated using the following formula [28]:(1)n=(r+1r)(p¯)(1−p¯)(Zβ+Zα/2)2(p1−p2)2
+		  (2)P1=0.76 P2=0.86 P¯=0.81To compare the dietary habits of the case and control groups, the sample size of this study was estimated by reference to the 2014 results for Myanmar on the WHO STEPwise Approach to NCD Risk Factor Surveillance (STEPS) [29]. In that survey, 86.6% of respondents consumed an average of fewer than five servings of fruits and vegetables a day. Therefore, this rate was assumed to be the rate of exposure to risk factors in the case group (P1). Assuming that people without diabetes had a high rate of vegetable intake, the control group rate was set at 76%. As a result of calculating the power used as 80% and the α error as 5% on both sides, the number of samples required to obtain the difference between the case group and the control group was 242. Therefore, 300 samples were sufficient to confirm statistically significant differences between the 80% probability and the 95% confidence intervals. Moreover, the number of samples was able to compensate for the predicted number of non-responders, which was approximately 10%.Dietary habits are habitual decisions an individual or culture makes when choosing what foods to eat and how to eat them [30]. Dietary habits and choices play a significant role in human health [31]. Therefore, in this study, we defined dietary habits as actual food intake and dietary behavior.Information on dietary habits was collected using the Food Frequency Questionnaire (FFQ), which estimates the frequency of daily food intake over a period. It has been widely used internationally [32].The food items were selected based on the Association of Southeast Asian Nations (ASEAN) food composition table [33] and those foods that are commonly eaten locally, as follows: rice, bread, noodles, meat, processed meat, fish, seafood, egg, beans, nuts, dairy milk products, non-dairy milk products, deep-fried food, stir-fried food, oil, seasonings, dried food, fermented food and pickles, sweet food, soft drink, fresh fruit juice, coffee or tea, vegetables, and fruit. Participants were asked for portion sizes of vegetables, fruits, rice, bread, soft drink, fresh fruit juice, and coffee or tea. Dietary behavior included the behavioral patterns that are associated with T2DM based on previous studies: having meals with family [22]; skipping breakfast [19]; eating out [21]; eating prepared foods [34]; having snacks [35]; removing visible fat [36]; drinking alcohol [37]; using supplements [38]; using traditional medicine [39]. All participants were asked about the frequency of their dietary habits over the past week: never or very rarely; 1–2 times/week; 3–4 times/week; 5–6 times/week; or every day.For those diagnosed with fasting plasma glucose in the case group, the date of diagnosis of T2DM and the blood sampling result were confirmed from medical records by collaborators (two physicians). Participants identified as having T2DM within the last six months were selected. Fasting plasma glucose in the control group was examined by collaborators (two physicians) using blood samples from participants’ capillaries. Blood samples were taken on an empty stomach after fasting for at least 8 h. Persons with fasting plasma glucose < 110 mg/dL were selected. A single investigator took anthropometric measurements using the WHO STEPS protocol [40]. Height was measured in centimeters with participants standing without shoes (0.5 cm accuracy). Weight was measured in kilograms with participants in light clothing (0.1 kg accuracy) using a digital weight scale (OMRON Body Composition Analyzer HBF-375, Japan). Waist circumference and hip circumference were measured using non-elastic plastic tape (TAKACHIHO Medical Co., Ltd., Tokyo, M12DXB, Japan). Body mass index (BMI) was calculated by dividing weight (kg) by height squared (m). The waist-to-hip ratio (WHR) was calculated by dividing the waist circumference by the hip circumference. Blood pressure was measured using the OMRON Digital Automatic Blood Pressure Monitor (HEM-907XL, Japan) after resting for 10 minutes in a sitting position. A trained physician performed all measurements using standardized procedures. Information collected on the demographic characteristics included: age (continuous in years); gender (male, female); marital status (never married/single, currently married, separated, divorced, widowed); education status (no formal schooling, less than primary school, primary school completed, secondary school completed, high school completed, college/university completed, postgraduate degree completed); employment status (government employee, self-employed, daily wager, education or livestock worker, factory or assembly worker, professional worker (e.g., accountant, doctor, academic), skilled worker (e.g., carpenter, hairdresser, computer worker), full-time student, household worker, retired, employed (able to work), unemployed (unable to work, others); personal monthly income (less than average, middle, high, very high); and family history of diabetes (yes, no). We also collected information on behavioral habits such as: cigarette smoking (never, quit, still smoking); alcohol drinking (never, quit, still drinking); and physical activity (vigorous, moderate, walking, sedentary). The International Physical Activity Questionnaire-Short Form (IPAQ-S) [41] was used for measuring physical activity. Measured time was calculated as minutes/week for vigorous physical activity, moderate physical activity, and low physical activity, and minutes/day for sitting (sedentary).According to the WHO guidelines, the survey instrument and consent cover letter were translated from English to Burmese [42]. Prior to this survey, a pretest using the same questionnaire was conducted on 30 Myanmar nationals who were residing in Tokyo to confirm the accuracy of the translation of the questionnaire into Myanmar. Based on the results, the revised questionnaire was used in this study in Myanmar. The questionnaire survey was conducted on all participants using a structured questionnaire that employed the self-administered questionnaire method. Research assistants (two physicians) who were trained in this study responded to requests for explanations of the questionnaire survey and questions from participants at any time. The questionnaire consisted of five sections: demographic information (demographic characteristics information, social-demographic information); medical information (medical history information, diabetes medication history); behavioral measurements (tobacco and alcohol use, physical activity, dietary habit); physical measurements (height, weight, waist circumference, hip circumference, blood pressure); and biochemical measurements (blood glucose).Categorical variables were expressed as numerals and percentages and were compared with the chi-squared test. Continuous variables were expressed as mean with standard deviation (SD) and were compared using a Student’s t-test or a Mann–Whitney U test for comparisons in each group. Food intake frequency and dietary behavior were classified into dichotomous categorical variables of high frequency and low frequency: high frequency (daily); low frequency (totally or very rarely, 1–2 times/week, 3–4 times/week, 5–6 times/week); high-frequency intake of vegetables and fruit (≥3 servings/day); and low-frequency intake of vegetables and fruit (<3 servings/day). We calculated the odds ratios (OR) and 95% confidence interval (CI) using univariate logistic regression to identify factors associated with dietary habits and T2DM. The outcome variable was diabetes status as binary data. Identified confounders were controlled in multivariate analysis, using multiple logistic regression that reported adjusted odds ratio (aOR) and 95%CI for each specific dietary habit exposure variable. The selection of covariates for multivariate logistic regression analysis was based on variables that showed a p-value < 0.2 in univariate analysis and previous studies. In the first model for the food intake, we adjusted for age and gender. In the second model, marital status, education status, family history of diabetes, current alcohol drinkers, physical activity (vigorous + moderate + walking activity ≥ 420 h/week, vigorous + moderate + walking activity < 420 h/week), and BMI (≥25 to 25<) were added to model 1 for adjustment. In the final, fully adjusted model, we additionally adjusted for having meals with family (never–every day) to model 2. In the first model for dietary behavior, we adjusted for age and gender. In model 2, marital status, education status, family history of diabetes, current alcohol drinkers, physical activity, and BMI were added to model 1 for adjustment. In the final, fully adjusted model, we additionally adjusted model 2 for seasonings (never–every day), vegetables (≥3 servings/day to <3 servings/day), and fruit (≥3 servings/day to <3 serving/day). Model fitness was checked using the Hosmer–Lemeshow goodness-of-fit test. A p-value < 0.05 was considered to indicate associated factors. Multiple logistic regression and all the analyses were performed using STATA version 16 IC (Stata Corporation, College Station, TX, USA).This study was conducted in accordance with the Declaration of Helsinki principles. The protocol for the study was approved by the Juntendo University Research Ethics Committee (approval number 2017141, 18 January 2018) and the Myanmar Ethics Review Board (approval number PLRID-00625_V1, 2 July 2019). Written informed consent was obtained from all participants.We analyzed 150 newly diagnosed T2DM patients and 150 community residents without T2DM. Table 1 shows the general characteristics of the case and the control groups and their associations with other covariates. The number of diabetic females (68.7%) was greater than males (31.3%). The mean age was 55.1 ± 10.9 years in the case group and 43.3 ± 14.8 years in the control group. The case group had a significantly lower level of education, employment, and personal monthly income compared to the control group. The case group had a more secure marital status and a more significant family history of diabetes compared to the control group. There was a significant difference in BMI, WHR, and high blood pressure between the case group and the control group. Regarding physical activity, the case group spent more time walking or in moderate or vigorous physical activity than the control group. Of the diabetes participants, 86% of those in the case group were taking medicine at the time of data collection, whereas no one in the control group was on medication.Table 2 shows the comparison of food intake frequency by univariate logistic regression analysis, the intake frequency of noodles, fish, beans, fermented foods and pickles, dried foods, and topping seasonings (such as salt, soy sauce, and fish sauce) was higher in the case group than in the control group. Conversely, the intake frequency of non-dairy products, vegetables (three or more servings/day), and fruits (three or more servings/day) was lower in the case group than in the control group. Table 3 shows the relationship between a high food intake frequency and T2DM, the results of multivariate logistic regression analysis. Statistically significant association of variables such as noodles, fish, beans, fermented foods and pickles, dried food, non-dairy milk products, and fruit to the outcome disappeared after adjusting for age, gender, marital status, education status, marital status, family history of diabetes, current alcohol drinkers, physical activity, BMI, and having meals with family. The final model showed that topping seasonings (aOR 11.23, 95% CI 3.08–40.90) was associated with a higher risk of T2DM, whereas more than three servings/day of vegetables (aOR 0.18, 95% CI 0.05–0.67) was associated with lower risk of T2DM (Figure 2).Table 4 shows the comparison of the frequency of dietary behavior by univariate regression analysis. The frequency of having meals with family, skipping breakfast, and eating out was higher in the case group than in the control group. Table 5 shows the relationship between the high frequency of dietary behavior and T2DM by multivariate logistic regression analysis. The association of variables skipping breakfast and eating out turned out to be non-significant after adjusting for age, gender, marital status, education status, family history of diabetes, current alcohol drinkers, physical activity, BMI, seasonings, vegetables, and fruit. The final model showed that having meals with family (aOR 2.23, 95% CI 1.05–4.71) was associated with T2DM (Figure 2). This study aimed to identify the risk factors in dietary habits and their relationship with T2DM in urban Myanmar residents. As a result, a daily intake of seasonings was significantly associated with T2DM, and a vegetable intake of three or more servings daily was inversely associated with T2DM. Regarding dietary behavior, having meals with family was associated with T2DM.We defined seasonings as salty foods such as salt, soy sauce, fish sauce, and fish paste. According to a study in Lithuania, after adjusting for possible confounders, participants who added salt to prepared meals had about a two-fold higher risk of developing T2DM compared to participants who never added salt to prepared meals [43]. In the first systematic review and meta-analysis assessing the relationship between sodium status and T2DM in adults, patients with T2DM had significantly higher sodium levels than controls. Regarding sodium intake, patients with T2DM had substantially higher levels of sodium intake compared to non-diabetic controls. Furthermore, increased urinary sodium excretion was associated with a higher risk of developing T2DM [44]. A prospective study in Finland reported that the relationship between sodium intake and risk of T2DM based on 24-h urinary sodium excretion data showed that high levels of sodium intake, measured in the highest quartile of 24-h sodium excretion, significantly increased the risk of T2DM [45]. In a recent study by Abdulai et al. on the role of a high dietary salt intake preference and diabetes in a rural population in China, the preference for a higher intake of dietary salt was associated with undiagnosed diabetes but not prevalent diabetes [46]. Moreover, other studies have demonstrated that an excessive salt intake may increase the risk of developing T2DM, possibly through a direct effect on insulin resistance and/or by promoting high blood pressure and weight gain [46,47]. The following presumed mechanisms have been proposed for the association between salt intake and T2DM. In general, obesity is associated with overnutrition and an excessive intake of sodium-rich foods [47]. A high salt intake activates the aldose reductase–fructokinase pathway in the liver and hypothalamus, leading to leptin resistance and endogenous fructose production, which affects obesity, insulin resistance, and fatty liver disease [48]. As a result, it may increase the risk of developing T2DM by directly affecting insulin resistance and/or promoting hypertension and weight gain [47,48]. A recent study by the Karolinska Institute for Environmental Medicine in Sweden (2017) found that for every 2.5 g of salt intake, the risk of developing T2DM increased by 65%. This study also revealed that the risk of developing T2DM is increased by 72% in people who consume more salt (7.3 g/day or more) than those who take less salt (less than 5.8 g/day) [47]. Nga Pi, a paste made from salted fermented fish or shrimp, is the main ingredient in a traditional food in Myanmar and is used as a condiment or additive in most dishes. It is a versatile food with many uses, such as in soup bases, salads, main dishes, and condiments [49]. Furthermore, Myanmar has a variety of other salty seasonings (pastes and sauces). While our study has not identified the kind of seasoning the participants added, the previous studies mentioned above support our result that Myanmar’s unique dietary habit of adding these salty seasonings to cooked meals may contribute to an increased prevalence of T2DM.Patients with diabetes are sensitive to salt, and salt intake is a major risk factor for increasing blood pressure. Moreover, hypertension is one of the leading causes of the accelerated progression of nephropathy [50]. Therefore, salt reduction education for patients with diabetes mellitus may lead to the prevention of diabetic nephropathy [51], and it can be necessary to promote it.Vegetables and fruit are rich in antioxidants, such as polyphenols, carotenoids, and vitamin C, which have been associated with a decreased risk of T2DM [52,53]. A meta-analysis of prospective studies found an association between vegetable and fruit intake and a lower risk of T2DM [52,54]. In addition, there are reports that green leafy vegetables reduce the risk of developing T2DM [55]. On the other hand, another prospective study found no association between vegetable and the risk of T2DM [56]. Other studies showed that the intake of vegetables and fruit combined, vegetables only, and fruit only were not significantly associated with the risk of T2DM [55,57]. Similarly, our results showed that only a daily vegetable intake of three servings or more was inversely associated with the risk of T2DM.A previous study followed the WHO STEPwise approach to the surveillance of chronic disease risk factors in Yangon [58]. The study reported the behavioral risks of NCDs, such as a high alcohol intake and a low vegetable and fruit intake in rural areas [23]. According to a study comparing the results of the 2009 and 2014 WHO STEPS surveys, the number of individuals with low vegetable/fruit consumption (<5 servings) has declined (4.3% RC), yet a high percentage of individuals were still observed as having <5 servings of vegetables and/or fruit per day (90.5% in 2009 and 86.6% in 2014) [59]. Another study found that those with at least two servings per day of vegetables and fruit had lower odds than others for hypertriglyceridemia among men and women in the Yangon Region and also had lower levels of total cholesterol among women [24]. Hypertriglyceridemia is one of the most common lipid abnormalities encountered in clinical practice. The most common causes of hypertriglyceridemia are obesity and uncontrolled diabetes [60]. Therefore, these results may support our findings that a high vegetable intake is inversely associated with the risk of T2DM among Yangon residents. We found that having meals with family was significantly associated with an increased risk of T2DM. Previous studies have emphasized the health benefits of frequently having meals with family to children [61], adolescents [62], adults [63,64], and older adults [22]. The outcomes of these studies include promoting a healthy diet and reducing the risk of overweight and obesity. Meanwhile, Horikawa et al. reported that energy intake was significantly higher in diabetic patients who ate at least once a month with their families than in those who did not [65]. In addition, Jeong et al. showed that triglycerides and fasting blood glucose in older males were likely to decrease as the frequency of having meals with family increases, conversely revealing conflicting results with those of elderly females. Elderly females had a greater total energy as a result of the frequency of having meals with their family; further, it showed an increased intake of nutrients, except for fat [22]. Those findings may support our results. However, in our study, the case group had a higher proportion of females, older ages, and married participants compared to the control group. In Asia, including Myanmar, females are generally responsible for most of the household chores. In preparing meals, females repeatedly taste food and take care of the food leftover after eating [22]. Hence, it is undeniable that the difference in social background between the case group and the control group affected our results.Moreover, the traditional Myanmar dining style is to sit on a mat around the table and share a meal with the family [49], a custom that may have led to excessive dietary intake in participants who frequently ate with their families. Therefore, the evidence of a relationship between having meals with family and health outcomes is inconsistent [62,63]. Accordingly, the mechanism that explains the association between the frequency of having meals with family and adult health status is unclear. However, it can be hypothesized that healthier meals served with family members helped to improve the nutritional intake and health quality of the adult participants. Moreover, we found that the education status was significantly lower (<primary school) in the case group than in the control group. This result implies that in order to stop the increase in T2DM in Yangon, it is necessary to promote adequate nutrition education simultaneously, rather than focusing solely on the frequency of having meals with family. In short, it can be assumed that having meals with family and sharing meals can create more meaningful opportunities for family members to share knowledge and information regarding daily food habits, which may lead to behavioral changes to acquire healthy lifestyle habits.In summary, overnutrition as a result of having meals with family may have induced obesity, a major risk factor for T2DM. Fulkerson et al. suggested that eating with family, friends, or neighbors in a home environment may differ from eating with others in a congregate meal setting or alone [63]. In short, this context may have a larger effect on dietary intake or weight status. However, our research has not yet evaluated the potential factors that are particularly related to having meals with family, such as the amount, types, and sources of foods served. Further studies are required to better understand the complex association between having meals with family and the increased prevalence of T2DM. Notably, diabetes prevention in Myanmar could apply family-based prevention approaches, such as information on how to modify meals with the family to be healthier.This study has several strengths. First, a case–control study was designed to identify the factors associated with dietary habits and T2DM, with cases selected from a hospital site and controls selected from a community site selected independently of exposure. Second, we were able to adjust for various potential confounders, including age, gender, education status, physical activity, and other health statuses that could disrupt the association between dietary habits and T2DM by multivariate logistic analysis. This study had limitations. First, there is a possibility of recall bias in the nature of the case–control study. Participants responded to questions regarding their dietary habits over the previous seven days by applying FFQ to detect their dietary habits. Second, some participants in the case group may have changed their lifestyle behaviors, including dietary and physical activity habits, during six months after the initial consultation as they may have already obtained information on lifestyle. Food habits are relatively quick to adopt. This may have reduced the difference in some food habits between the case and control groups. As is the nature of case–control studies, readers should interpret epidemiological relationships between diabetes and risk factors carefully, with biological plausibility. Finally, there was a large difference in the attributes of the case group and the control group. In Myanmar, it was difficult to freely select the target area due to the restrictions of the political system. Participants in both groups were randomly chosen to minimize the selection bias; however, significant differences in attributes occurred due to characteristic differences between hospitals and communities. The present study was the first to identify the factors behind the increased prevalence of T2DM in urban Myanmar by focusing on dietary habits. Therefore, our findings may contribute to the formation of a prevention program based on cultural dietary habits in a community setting as a health promotion intervention for T2DM. We recommend that further longitudinal studies be undertaken to investigate the association between seasoning use and T2DM, and an intervention study targeting healthy family meals in Myanmar, such as cooking classes and early nutritional education. Public health interventions, with either appropriate risk targeting or population-wide interventions, to suppress the rise in diabetes mellitus are required in Myanmar and in many countries that share similar cultures and contexts [10,66].We clarified in detail the different dietary habits of diabetes patients and non-diabetes persons in Yangon, Myanmar. The study suggests that Myanmar’s characteristic dietary custom of topping meals with salty seasonings and sauces and eating multiple dishes together as a family are risk factors associated with T2DM. The study results warn against the use of these seasonings and their effect on T2DM, and it promotes vegetable intake and the empowerment of families to cook and eat healthy meals to prevent the escalating prevalence of diabetes. Our findings contribute recommendations and opportunities for the primary prevention of T2DM in urban Myanmar.S.U., M.Y. and M.N.A. were involved in the study design, data quality control, statistical analyses, and manuscript drafting. S.U., M.Y., M.N.A. and A.I. conceived and coordinated the investigation and were involved in the study design. T.S.L., S.M. and E.T.K. coordinated the investigation and contributed to the study design and data acquisition at the survey site. S.U. contributed to data processing and interpretation of results and verified the analytical methods. M.Y. and M.N.A. advised on algorithm application and verified the analytical methods. S.U. contributed to the original draft preparation, review, and editing. T.T., M.Y., M.N.A. and S.S. advised on study design, data acquisition, interpretation of results, and manuscript drafting. M.Y., S.U. and A.I. contributed to the funding. M.Y. and M.N.A. contributed to the management of the entire project as supervisors. M.N.A. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors have read and agreed to the published version of the manuscript.This work was funded by (1) the Pfizer Health Research Foundation Grant (international joint research) in 2017 and (2) the Japan Society for the Promotion of Science (JSPS-Grants-in-Aid for Scientific Research: Grant No. 18K10110).The protocol for the study was approved by the Juntendo University Research Ethics Committee (approval number 2017141, 18 January 2018) and the Myanmar Ethics Review Board (approval number PLRID-00625_V1, 2 July 2019). The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.The author would like to thank Yoshihisa Shirayama for his expert advice and encouragement on this research. We are grateful to all study participants for their contribution. Special thanks are extended to physicians and other medical staff at all participating diabetes clinics for their kind assistance and their efforts in the recruitment of patients and data collection. The authors declare that they have no conflict of interest.Criteria to select the study population.Dietary habits associated with type 2 diabetes mellitus in a case–control study of the urban residents in Yangon, Myanmar, 2019. Note: Multivariable logistic regression analysis; p-value for the comparison between two groups: *** p < 0.001, * p < 0.05. All p-values are two-sided.General characteristics of cases and controls.Note: Student’s t-test or Mann–Whitney U test was used to compare continuous variables, and chi-squared test was used to compare categorical variables; SD: standard deviation, %: ratio of valid responses; p-value for the comparison between two groups: *** p < 0.001, ** p < 0.01, * p < 0.05. All p-values are two-sided.Comparison of food intake by high frequency between diabetes and non-diabetes participants.Note: Chi-squared test was used to compare categorical variables; High frequency: eating every day, %: ratio of valid responses; p-value for the comparison between two groups: *** p < 0.001, ** p < 0.01, * p < 0.05. All p-values are two-sided.Dietary habits associated with type 2 diabetes mellitus among the urban residents in Yangon Myanmar, 2019.Note: Multivariable logistic regression analysis; Goodness-of-fit test for a logistic regression model 3; Pearson chi-squared test (p = 0.465), Hosmer–Lemeshow chi-squared test (p = 0.106); Crude: unadjusted; Model 1: adjusted for age, gender; Model 2: additionally adjusted for Model 1 + marital status, education status, family history of diabetes, current alcohol drinkers, physical activity, BMI; Model 3: additionally adjusted for Model 1 + Model 2 + having meals with family; OR odds ratio, aOR adjusted odds ratio, CI confidence interval; p-value: *** p < 0.001, ** p < 0.01, * p < 0.05.Comparison of dietary behaviors by high frequency between diabetes and non-diabetes participants.Note: Chi-squared test was used to compare categorical variables; High frequency: doing every day; %: ratio of valid responses; p-value for the comparison between two groups: ** p < 0.01, * p < 0.05. All p-values are two-sided.The association between having meals with family and type 2 diabetes mellitus among the urban residents in Yangon, Myanmar, 2019.Note: Multivariable logistic regression analysis; Goodness-of-fit test for a logistic regression model 3; Pearson chi-squared test (p = 0.465), Hosmer–Lemeshow chi-squared test (p = 0.106); Crude: unadjusted; Model 1: adjusted for age, gender; Model 2: additionally adjusted for Model 1 + marital status, education status, family history of diabetes, current alcohol drinkers, physical activity, BMI; Model 3: additionally adjusted for Model 1 + Model 2 + seasonings + vegetables + fruit; OR odds ratio, aOR adjusted odds ratio, CI confidence interval; p-value: *** p < 0.001, ** p < 0.01, * p < 0.05.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11057.txt

@@ -0,0 +1,27 @@
+The literature has not yet well documented the relative elements of the adoption of healthy lifestyle habits (HLHs) over the long term. More especially, researchers are calling to complete the corpus with qualitative or mixed estimates that would allow them to better explain the conditions necessary for the adoption or maintenance of HLHs over the long term. The present study seeks to understand the winning conditions for the adoption and maintenance of HLHs. Semi-structured group interviews were conducted with three groups of university students (two in Bachelor’s degree in physical education (PE) and one in Master’s degree in physical activity science), all in favor with HLHs. The results identify some dominant winning conditions in the adoption and maintenance of HLHs, such as the role of the family environment, the role of diversity and choice in physical activities during youth, the physical and social environment, autonomy and also mental health, which is closely linked with physical health. Results were modelled in the form of an ideal pathway, which traces the impact of winning conditions from childhood to adulthood. The originality of this study stands out, among other things, because of its innovative methodology; therefore, this study opens the door to future qualitative research in the field. Investigating pathways, considering the different phases of development of children and adolescents to identify factors of change and maintenance of HLHs now seems to be an interesting and necessary avenue for research in the field.This study focuses on the winning conditions for the adoption and maintenance of healthy lifestyle habits (HLHs) by university students in training in the field of physical activity (physical education and kinesiology). The project uses a qualitative, backward-looking methodology to document the elements and mechanisms of their life experience that students retrospectively perceive as important in their awareness of HLHs and in the adoption and maintenance of some of them. The interest in the issue of maintenance stems from the lack of evidence in health promotion about what works in the long term to build HLHs. Therefore, it is interesting to understand what types of experience play a role in the winning conditions during the course of life. For example, are they cumulative, repeated, one-time but transformative experiences?By “winning conditions”, we refer to all kinds of factors that have a positive impact on the HLHs in the participants’ journey. The exploratory context of this study leads us not to restrict ourselves to one category of factors, but rather to be inclusive in all their variety, with the greatest possible openness. Thus, following the example of the life course theory [1], our approach seeks to “grasp the logics that structure diverse trajectories” [1] (p. 34) and to “understand the interactions that link them to one another” [1] (p. 34). This is conducted with the knowledge that individual trajectories are fragmented by moments of stability, improvement and decline in relation to HLHs [2].The growing interest in health promotion has its origins in the fight against numerous chronic diseases and the increased sedentary lifestyle of the population, because children are also more likely to remain obese as adults [3,4,5,6]. In Canada, only one-third of Canadian youth meet the recommended PAP guidelines of 60 min of moderate-to-high intensity daily physical activity participation (PAP) [3,7]. In response, several proposed interventions have been implemented to increase youths’ PAP to produce immediate or longer-term health effects [8]. In fact, although HLHs have been well documented for decades, we still do not know enough about which factors should be targeted for the promotion of health in our populations [3,9,10]. Indeed, the trend indicates that young people are living increasingly sedentary lifestyles and are less and less invested in regular PAP [11].The literature offers several definitions and components of HLHs that differ according to contexts and study objectives. Thus, authors often target determinants of health that they deem relevant to their research [12]. Consequently, it is possible to identify several theoretical models for understanding HLHs and the development of winning conditions for their adoption and maintenance. Some are predictive and others are prescriptive or are part of an ecosystem principle. First, some so-called “predictive” theories attempt to identify and explain the determinants of the adoption of a health-related behavior, for example Azjen’s theory of planned behavior [13] or Godin’s integrative model [14]. Ajzen’s [13] theoretical model focuses on the notion of intention to adopt a behavior. This behavioral intention is influenced by various factors and must first and foremost be planned and decided upon in order for change to occur. The higher the degree of strength of the intention, the more positively efforts are turned towards the commitment of the behavior, the more possible its adoption becomes [13]. In a complementary manner, Godin [15] proposes a model of a holistic view of health change. In his integrative model, similarly to Ajzen’s, intention is key. Behavioral change can occur through the perception of behavioral control when adoption does not come from the individual’s will [14]. Godin goes further with the addition of external variables [15] that include individual characteristics, such as age, gender and personality, as well as environmental characteristics (physical and social), such as culture, socioeconomic background, etc. Ultimately, Azjen’s and Godin’s theories help us understand the phenomena of intentionality related to the adoption and maintenance of HLHs, while incorporating the individual and environmental aspect of the person. Second, so-called “prescriptive” theories attempt to understand the adoption and maintenance of HLHs by promoting the implementation of a behavioral change process. Deci and Ryan’s [16] theory of self-determination proposes, for example, a continuum of motivation, ranging from amotivation (lack of motivation), through the full spectrum of extrinsic motivation, to intrinsic, or self-determined motivation [17]. One of the key propositions in this theory is that self-determined motivation satisfies three basic psychological needs in the individual, namely, autonomy, a sense of competence and social belonging, which would increase the chances of maintaining HLHs. Third, the social-ecological model attempts to understand the interrelationships between the individual and their various ecosystems. Bronfenbrenner developed the social ecological model [18], which was later taken up by Sallis and Owen [19] in HLHs and health promotion. First, this model has several levels that influence health: intrapersonal, interpersonal, organizational, community and political. Second, sociocultural factors and physical environments can influence HLHs on more than one level. Furthermore, there appears to be an interaction between these different levels of influence; the different variables work together and not in isolation.From these types of models, it is possible to differentiate several trends in the theoretical models of health promotion and HLHs. For example, it is possible to clarify the reasons for an intention to adopt a HLH by identifying the factors underlying this intentionality, such as perceived control (self-efficacy), perceived norms (peer and family influence) and attitudes (positive attitude towards the behavior) [14]. In addition, interesting links can be made with the role of peer-guided motivation, feelings of self-efficacy and autonomy, as in the case of self-determination theory [16]. Finally, the social ecological model [19] comes to integrate everything in an ecosystemic way. These theories can also be differentiated by their type of intervention and by their targets. In other words, there are theories that are more person-centered, such as Deci and Ryan’s and Ajzen’s. On the other hand, we have the socio-ecological model of Sallis and Owen, which places the environmental dimension as a priority. Between the two, we have Godin’s integrative model, which unifies these two axes in their vision of an efficient intervention.However, none of these models can fully and satisfactorily explain why an individual adopts or maintains HLHs over time. Consequently, it is illusory to think that a single model could satisfy the requirements and perfectly guide future interventions [14,20]. Added to this are the numerous limitations (e.g., lack of long-term evidence, need to complete research with mixed or qualitative designs) that have been raised in the various health promotion studies for several decades [21,22,23,24] that lead us to question research methods, as no model seems to be able to explain how HLHs are adopted and, especially, maintained [25]. One of the alternatives proposed by researchers [22,26,27] that have been little explored by the literature most likely concerns the type of methodology employed.Despite health promotion efforts, authors note a lack of sustainability and maintenance in health promotion intervention outcomes [21,22,23,24]. Current programs, in most cases, result in short-term improvements that tend to fade within a few weeks after the intervention ends [8]. Spinola and Castro’s [23] study supports this by demonstrating a lack of long-term evidence. In addition to these findings, few results address the best strategies to adopt when it comes to maintaining HLHs. Thus, the evidence for the effectiveness of health promotion strategies is weak, both in terms of implementation, policy and practice [24].The development of a HLH and its maintenance occurs in a process that may take time to settle, if only because it may involve “relapses” to old behaviors [28]. In this sense, current methodologies do not allow for relevant findings, as they do not seem to be adapted to the journey of a young person, who goes through several developmental stages and a multitude of conditions that shape his or her HLH choices [29].Furthermore, there is a need to supplement numerous quantitative studies with qualitative studies to find precisions about the phenomena and explore perspectives of HLHs. Indeed, many quantitative studies have been conducted to address the poor lifestyle habits of youths [22,27] and, generally, distinguish between experimental, cross-sectional studies, randomized controlled trials, in particular [30]; many provide data on the effectiveness of individual programs [25]. However, these studies have important limitations, as mentioned by these authors. On the one hand, they mostly focus on interventions that are very localized in space and limited to a specific intervention period. On the other hand, their essentially quantitative analyses make generalizations that are not very valid, since they rarely take into account the fact that participants are often influenced by the characteristics of the environment set up in the intervention [31]. These research designs focus on numerical (quantitative) observables, without delving qualitatively and concretely into what really matters to individuals. In particular, how to maintain a HLH engaged or enhanced by these specific time-limited interventions [9].As discussed above, current studies of HLH adoption and maintenance have methodological limitations, including a lack of long-term evidence, due to designs that are often limited to the intervention period, and predominantly quantitative approaches that do not allow researcher to attain a fine-grained understanding of the phenomena. Here, then, there are three methodological considerations in an attempt to address these limitations.It is relevant to document “what seems to work” for the adoption and maintenance of HLHs, i.e., what promotes a HLH rather than what hinders it, because it is not said that it would be enough to “reverse” the conditions that hinder it so that, all of a sudden, people adopt and maintain VHS. The approach of exploring the successes and the positive that exist in the individual has a solid scientific basis [32], which is found in the method of appreciative inquiry, for example, looking for the positive—what works and not what harms [33]. Thus, appreciative inquiry is in line with positive psychology on the idea of highlighting strengths, which are more motivating than weaknesses and more conducive to change in human beings. Cooperrider and Whitney were influenced by this positive approach in research in the fields of sports, medicine, behavioral sciences, etc., proving that positive images have a considerable impact in human psychology [34].The idea is to overcome the temporal limitation of the short term, which fails to capture the fluctuations in adoption, dropouts, resumption, or relapse that are so important to understanding the strength and dynamics of maintenance over time. This type of approach also avoids deploying complex, time-consuming and costly methodological follow-ups. Consequently, it seems relevant to study HLHs through a qualitative approach and from a long-term perspective, which has never been performed, to our knowledge. A better understanding of what promotes the maintenance of HLHs could help inform future program or policy decisions [25].In this study, the objective is not to follow people over several years, as in a longitudinal study, but rather to reconstitute the major inflections of the subjects’ experience backwards, starting from the account they are able to give, accompanied by a researcher. The analogy underlies the method of industrial or computer reverse engineering, that is, taking a finished and functional product, then dismantling it piece by piece to understand its operating principles. Thus, the study wishes to highlight triggering situations that have challenged the subjects and may have contributed to changes in or maintenance in people’s habits. The whole is inspired by the reverse engineering approach of the facilitating conditions of certain human phenomena proposed by Chaubet et al. [35].Therefore, this article aims to answer the following question: What are the winning conditions for adopting and maintaining healthy lifestyle habits according to the stories told and illustrated by students in physical and health education and in physical activity sciences?We identified three objectives to answer this question:To describe the pathways that led physical and health education (PE) and Master of Science students in Physical Activity (M) to adopt and maintain HLHs and to characterize the forms of change in or maintenance of these HLHs;To identify the winning conditions in the adoption and maintenance of HLHs among these students and characterize them;To integrate the winning conditions into a conceptual model of HLH adoption and maintenance.To describe the pathways that led physical and health education (PE) and Master of Science students in Physical Activity (M) to adopt and maintain HLHs and to characterize the forms of change in or maintenance of these HLHs;To identify the winning conditions in the adoption and maintenance of HLHs among these students and characterize them;To integrate the winning conditions into a conceptual model of HLH adoption and maintenance.This research project is based on a qualitative–interpretive methodology, that is, the elements that emerge from the data collected are derived from the participants’ understanding of the phenomenon under study, according to their experience. This inductive framework then becomes a scientific tool to shed light on the determinants of the adoption and maintenance of HLHs. It is a qualitative process inscribed in a dynamic that seeks to make sense of and understand the complexity of a phenomenon, while remaining rooted in the dialectic of interpretations and representations made by the participants [36]. The human experience, whether real or imagined, is related to a statement, which makes it possible to seek an understanding of the phenomenon [37].The intended participants were identified in a convenience sample, based on intentionality criteria [38]. Originally, four group profiles were selected: Bachelor’s degrees in PE education, kinesiology, dance education and Master’s degrees in physical activity science. The COVID-19 pandemic restricted access to individuals. For feasibility reasons, the analysis was reduced to the three groups already interviewed: two in PE (N-PE1 = 5 and N-PE2 = 7) and one in MSc (N-M = 13).Faculty members responsible for the university courses in the programs mentioned facilitated access to their group courses. The student participants share two characteristics: (1) their disciplines challenge them on issues of HLHs by requiring them to consider the physical and mental health of the individuals with whom and for whom they are training; (2) they also have an experience of personal reflection on these issues, with their program providing additional tools and opportunities for reflection. Therefore, this sample is purposive, accessible and relevant to the research purpose and question, a critical aspect for the scientific validity of a study [39].The main instrument used for this research was the semi-structured group interview, audio recorded and verbatim transcribed. A total of three interviews lasting approximately 40 min were conducted two in PE and one in M. The semi-structured interview method was chosen in order to allow participants to have the freedom to delve into their journeys, with their own set of stories and according to their sensitivity to the field. An interview guide helped to keep the exchange within the following three themes: the participants’ experiences with HLHs (their practice); behavioral changes (before and after); the facilitating elements in their journey to adopt them (what made the difference, what helped foster change) (see Appendix F).The data analysis was conducted in three stages. A first level of analysis isolated concrete experiences experienced by and exemplified by the individuals, indicating changes in participants’ behavior. From these “islands of change”, the analysis expanded its focus to the sources and/or conditions surrounding these changes [35]. Thus, an initial code grid was formed, a mixture of pre-constructed categories (change and surrounding conditions of those changes) and self-imposed categories related to the goals (form of harmful lifestyle habits, signs of maintaining HLHs, signs of not maintaining, winning condition of HLHs, etc.).The second stage was part of a completely inductive movement, consisting of a more detailed understanding of each of the phenomena identified. This inductive work, by conceptualizing categories [37], strongly inspired by the progressive implementation of a “grounded theorization”, but without successive returns to the field [40], produced, little by little, a modeling of several phenomena relevant to the study.Finally, in the third and last step, recurring elements were noted and links were established between the different accounts of the participants (which we also call “stories”, because that is often the form they take), at different moments of their experience. From there, a “linking” exercise identified similarities, dependencies and hierarchies among the elements of analysis, in order to create an organized and structured model [40] that would allow us to establish typologies and map an overall picture of the results.This section presents the results of the study by following our research objectives. Our first research objective—to describe the adoption and maintenance pathways of HLHs—was addressed during the analysis stage by isolating the concrete experiences of participants using pre-constructed categories (e.g., forms of HLHs, signs of maintenance, winning conditions of HLHs and changes).The following results correspond to the second research objective, which was to identify the types of winning conditions for the adoption and maintenance of HLHs. To achieve this, the analytical construction was carried out in three steps: (1) identify these winning conditions and their predominance for each of the profiles, in the form of key characteristics (Appendix A and Appendix B); (2) create a schematic representation with these same characteristics, but, this time, establish influences between them (Figure 1, Figure 2, Figure 3 and Figure 4); (3) synthesize these results in a conceptual map and a preliminary modeling that includes the two profiles under study (Figure 5 and Figure 6).Appendix A and Appendix B summarize the dominant characteristics of our participants’ adoption and maintenance of HLHs, merging the two profiles PE and M1. It partially addresses our second research objective—to identify and characterize the winning conditions for HLH adoption and maintenance. A table for each profile was initially created (Appendix A and Appendix B) to arrive at this overall representation (Figure 5). The main characteristics of the two profiles that emerged during the interviews are identified and illustrated by blocks of different colors, which correspond to decreasing levels of recurrence according to the number of participants (blue, green, orange, yellow and purple).There are five levels of key characteristics. The first includes the two most important key characteristics of participants’ adoption and maintenance of HLHs, the relationship between physical and mental health and the family environment. The second represents the social environment. The third includes the link among enjoyment, motivation and self-efficacy, as well as the integration of the outdoors into one’s life. The fourth level consists of the intrinsic awareness of the benefits of HLHs. Finally, the fifth level has three characteristics: the physical environment, the balance between the different spheres of our lives and the routine (categories are defined in Appendix C).Subsequently, a schematic representation of the key characteristics was developed based on the influences they have on each other to better understand the role of these winning conditions. This representation helps to address our second research objective—to identify the types of winning conditions for HLH adoption and maintenance. As with Appendix A and Appendix B, a diagram for both profiles (Appendix D and Appendix E) was initially created to arrive at this synthesis. To facilitate understanding of the synthesis and its concepts, we propose, in the next sections, a gradual schematic and conceptual construction.There are three major elements in this schematic construction (Figure 1). First, PAP and psychological health are at the very heart of the analysis, as they are the elements most exposed by our participants. A priori, they are individual dimensions of health; therefore, they are sought-after goals, in terms of HLHs. The results show that they are also winning conditions for the adoption and maintenance of HLHs, because they act in a bidirectional and cyclical way towards each other, as indicated by the arrows in the figure.
+                YÉD2: I have always found over the years that I felt better and had better overall mental health when I exercised just a little bit, every day. This is something I discovered quite early on, but I confirm it [...]
+              
+                MAK1: Sport has always been part of my life since I was little. So it’s something I do for fun, it’s a hobby, it’s a need to feel good. So it’s been integrated since I was really little.
+              The third major element in this pattern is the blue circle of family, which implies support, stability and an inspirational model. One explanation for the connection among these three elements is a family environment that has been able to provide diversity for their child.
+                SÉD1: [...] Every weekend we did something new, especially in nature. We’d go hiking, we’d go mushroom picking, unusual things, fishing, hunting. Uh they signed me up for field hockey, soccer, baseball when I went to the States. Uh... I had a chance to try out a lot of sports [...]
+              This family environment also provides ongoing support.
+                ÉD2: [...] Of course I couldn’t do all the sports at the same time, I would have liked that, but they said, Perfect, we’ll support you. And I did other sports as well and every time they were behind me.
+              The second level of recurrence is the social environment (Figure 2). It has a bidirectional link to the PAP. Some mentioned that peers pushed them to move while others invested in PAP because of their social aspect. This winning condition for adopting and maintaining HLHs also has an important link to motivation.
+                OÉD2: [...]… sometimes I was not motivated. Then my friends would tell me “oh, let’s go, let’s go do some sports” and I would go and finally I was happy to go but otherwise I wouldn’t have gone by myself.
+              The physical environment and leisure services are also winning conditions (in purple).
+                NK1: And when you go outside, well, you find a lot of... I played baseball very quickly from the age of 5–6 at the local park and that made me want to play [...] Without the social infrastructure where people were outside, there might not have been as many hooks either.
+              
+                Mk1: [...] Everyone is there at the same time: the arena, the baseball field, the park. So the kind of location creates a kind of excitement to go outside and play with our friends outside and then bike to our parks.
+              The factors rated as moderately strong are then enjoyment, sense of self-efficacy, motivation and peers (Figure 3). Three of these interact. For example, a sense of competence can lead to increased enjoyment of the PAP and thus motivation.
+                AÉD2: It was also a need to... Not a need but actually I felt successful, I felt competent in what I was doing so I was just enjoying doing the sport by myself.
+              Another winning condition judged “moderately strong” is the place of the outdoors in the lives of participants (Figure 3). It is influenced by the pleasure experienced when they integrate it into their lives and it refers as much to sports practiced outdoors as to outdoor activities such as camping. The outdoors also has links to PAL and psychological health.
+                MTK1: But what I found important to integrate is the outdoors, so to really have a contact with nature, that’s primordial and also mental health. I think that stress is such a major problem.
+              The dominant element for this level is the intrinsic awareness of the benefits of HLHs (Figure 4). It refers to the individual’s awareness of their choices in terms of HLHs, that is, they realize the importance of their actions. The word intrinsic is used because, according to the participants, it is an awareness that was once influenced by external elements, such as family, and, now, comes from within.
+                VIK1: It was just implanted, it was just natural, I didn’t really ask myself any questions. [...] And when I got to college, university, I just realized how important it was.
+              The link between intrinsic awareness and mental and physical health is also present:
+                AK1: And at the university, with kin[esiology] it was not really my appearance, but it was more when I train, I have more energy, as I feel better, I have less headache. So it seems that my evolution was more external and now it seems more internal.
+              Finally, factors strong enough to still be cited by participants were routine and balance (Figure 5). Analyses show that routine plays a role in achieving regular and/or daily PAP. As for balance, we see a direct link to psychological health and participants refer to a lifestyle that includes time for PAP, psychological health and social life.
+                GÉD1: [...] I changed that very habit. So I opened a little more time for my family, for my friends and to go towards new activities. And that, in my eyes, is a healthier way of life than what I had before, where I was focused on one thing and one thing only: sports. So, this is a turning point. [...] But I still keep in mind the learning that I did, so I always keep time for the other spheres of life.
+              Figure 5 illustrates the complete synthesis of the winning conditions for the adoption of HLH maintenance found among participants in both profiles. The different colors and thickness of the circle express the strength according to recurrence (for ease of understanding, these are the same colors as in Appendix A and Appendix B). The solid black arrows show which characteristics may have acted on another. Dotted black arrows express a bidirectional link between two or more characteristics. Finally, quotes from participants directly illuminate the type of link between concepts on the arrows.Appendix A and Appendix B and Figure 5 summarize the winning conditions for the adoption and maintenance of HLHs among the two participant profiles. Overall, the predominance of PAP belt functioning and mental health can be seen. These two dimensions of health are influenced by many winning conditions, such as family, social and physical environment, and the triangular link between pleasure, motivation and self-efficacy. Furthermore, intrinsic awareness of the benefits of a HLH and the integration of the outdoors have an impact on both PAP and psychological health. Finally, having a routine implemented in one’s schedule seems to be a winning condition for maintaining a PAP level; similarly, the balance between the spheres of one’s life helps the participants’ mental health.In this qualitative study, participants identified numerous factors that had an impact on the adoption and maintenance of their HLH that is different from and similar to numerous quantitative studies. After describing the pathways and then characterizing the forms of change that led students from two profiles (PE and M) to adopt and maintain HLHs, the winning conditions for adoption and maintenance were identified for these participants (Appendix A and Appendix B) by showing these key concepts and their influential relationships (Figure 5). The third objective was to integrate these winning conditions into a conceptual model of HLH adoption and maintenance (Figure 6). To achieve this, a conceptualization of these mechanisms was made, according to the paths of the two profiles under study. It is organized to understand their evolution and degree of importance. This modelling becomes relevant in order to shed light on the determinants on which to act and the appropriate periods to act, in order to help the adoption and maintenance of HLHs.This model (Figure 6) can be understood according to four pillars of HLH adoption and maintenance: the family environment, the social/physical environment, the rise to independence and independent functioning. First, the family environment is the starting point for the adoption of HLHs, as already shown. The level of autonomy is limited and/or partial and everything is conducted in the natural setting of the family. When participants talk about the family environment, they refer to the diversity of activities they have experienced, its support and its role as a model. The second pillar includes the social environment and the physical environment (as well as the combination of the two). It is important to mention that the impact of these environments may have occurred simultaneously with or following the family environment, as it was impossible to obtain a precise chronological track on when they appeared as facilitators [41]. The results show that the social environment, such as peers and different human models other than the family, as well as the physical environment, such as parks near one’s home, act as winning conditions for the adoption and maintenance of HLHs. Third and fourth came the evolution of the participants’ career path, an important pillar in itself, which gradually places the individual at the center of their decisions (third pillar), as previous study demonstrated [42]. Individuals then manage their HLHs with an autonomy marked by independence (fourth pillar). This independence occurs at specific moments in the lives of our participants, such as leaving the family nest or starting post-secondary studies. What we observed at this last stage is the triangular link between psychological health, physical health and intrinsic awareness of the benefits of a healthy and active lifestyle. This desire and awareness of the need to be psychologically and physically healthy emerges for some when they have experienced a prolonged break from an active lifestyle and consequences have followed (e.g., injury that prevented PAP). For some, the good that was attached to this LAP dissipated and, in contrast, they indicated the importance of moving for the body and, more importantly, for mental well-being. For others, the sheer independence and autonomy that comes with it has made them realize that they need to take action to maintain active HLHs.Our results show that the adoption and maintenance of HLHs follow the rise of autonomy and that, in parallel, several environments act at different key moments of the life course, from the family, to the community and the social and physical infrastructures, to the self. For ease of understanding, we present an ideal life course, in four major milestones, that synthesizes and illustrates what participants said.First, the family plays the largest role in the adoption of HLHs in children. Initially, the parents offer a variety of activities so that the child has a wide range of experiences related to the PAP. At this point, their autonomy is rather limited, as choices are more imposed by the family. By being active themselves, parents establish a culture of sport in the family and act as positive role models for both nutrition and physical activity. Several authors have reached the same conclusions. This is the case of Bergeron and Reyburn [43], who mention that the interactions between the family and the young person can lead to a strong identity that facilitates the adoption of HLHs. The aspect of culture is also present in other qualitative studies, such as Dagkas and Stathi’s [44], who interviewed adolescents about the factors that influence their PAP in and out of school. Their findings are consistent with those of our participants; they mention that parental culture becomes a mirror of the child’s involvement in PAP and that their support is crucial. Active parents then serve as role models for their children, which facilitates their children’s PAP [45,46,47,48]. In addition, the literature shows that parents who move with their young children also reinforce an active lifestyle in their children [45,49,50]. However, this element was not specified in many of our participants’ responses. It would have been interesting to question them to see if their parenting model involved parents who engaged in PA with them. In addition to this culture, parental support and encouragement also seem to be favorable and important in our ideal pathway. This actually agrees with Sallis et al. [51] and Hesketh et al. [52], in that these two aspects greatly influence a child’s PA level.Following this beginning of autonomy, the social and physical environment becomes increasingly important. In the case of the social environment, the young person tends to gravitate towards physical activities that he or she can practice with friends, for example, team sports that act as an incentive, as a motivator and that develop that community spirit. Salvy et al. [53] go in the same direction, mentioning that, during adolescence, friends become more important than family by greatly influencing our behaviors. To this end, their study found that, if friends practiced a PAP, it motivated the youth to practice one in turn and increased the likelihood that they would actually do so. Duncan et al. [54] reported similar results, stating that youths are more active when they are with friends. Unlike many studies that report the impact of an individual as a trigger/facilitator in PAP, such as a health and physical education teacher [55,56], our results do not identify this factor as a winning condition for the adoption and maintenance of HLHs. Could it be that our participants, who were predominantly from a family background conducive to HLH adoption, were less sensitive to the impact of a teacher solely because they were already aware of HLHs?In the case of the physical environment, it has a positive effect on HLHs if there are facilities or infrastructure close to where people live. Our participants mentioned sports fields, parks, bike paths, etc., which they could choose to use. Several studies have reported on the impact of the physical environment on youths’ PAP. Dagkas and Stathi [44] highlight the role of the neighborhood in which we live. Similar to our results, their participants mention that proximity to facilities influenced their PAP. Xu et al. [57] make the same point; the variety, quality and proximity of facilities are important to adolescents’ level of PAP outside of school. In addition, the better a park is maintained, the more people enjoy using it [43].Our participants go even further, as they mention the combination of the social and physical environment. They refer to available infrastructures, close by and used by the community and their peers; therefore, such facilities are alive with organized and animated activities, such as those related to sports leagues, for example. This means that it is not enough to have physical facilities, but rather to make them lively and to create a buzz and a community spirit around them. In the same logic, the literature review by Gadais, Boulanger, et al. [58] demonstrates the benefits of a good transportation system to facilitate the accessibility of these places. Authors report the importance of design features, such as safety, street quality, lighting, etc., in fostering an active community and developing positive perceptions of active transportation among youths [59,60]. Furthermore, authors argue that the built environment has effects on youths’ PAL based on the perceived level of social support [61]. This is also supported by D’Angelo et al. [62], who establish, in their study, a positive association between adolescents’ PAP and the role of peers, reinforced if their neighborhood holds many resources to move.This study shows that, as young people age, they become independent from their families and, thus, autonomous. With the arrival of this independence, they face completely different lifestyles, e.g., studies, work, living in an apartment, etc. Our results suggest that their decisions are now conscious, because they know that their choices influence their general health, unlike when they were young, when everything was done naturally (e.g., I used to spend my weekends with my parents playing sports, I used to go out and play with my friends, without asking myself questions). Some of our participants experienced a drop in PAP when they became independent, which caused an important trigger on the importance of having VHS (having experienced the negative effects of these changes in behavior, while realizing that it is necessary to take action to maintain HLHs). This does not appear to be a unique case. According to Leriche and Walczak [63], the decline in PAP among new CEGEP (Post-secondary level in Quebec, also known as college, preceding university by two years and giving access to it) students in Quebec is caused by poor time management. This perception of a lack of time would also be on the rise as their college career progresses. Despite this, our ideal pathway shows clear signs of maintenance (e.g., I have a routine to keep moving, I have signed up for a new sport with friends) and demonstrates a desire among participants to maintain a healthy and active lifestyle. Leriche and Walczak [63] refer to conscious, planned and purposeful choices, as in Ajzen’s and Godin’s models [13,14] with their notion of intention.Furthermore, the intrinsic awareness of the benefits of HLHs leads our participants to maintain HLHs for the positive repercussions they have on a physical level, but especially on a psychological level. Indeed, the place of psychological health in relation to PAP greatly dictates their lifestyle. To achieve this, they keep a balance between the different spheres of life (e.g., social, physical, mental) and above all, perform PAP capable of bringing them this state of well-being. Many of them perform PA outdoors, with peers, and seek to feel pleasure during the activity. This close relationship between these two dimensions of health has been noted by many models, which raise consistent links with our results. First, it is recognized that engaging in outdoor physical activity has promising effects on well-being, in addition to bringing pleasure and satisfaction [64,65]. Then, authors mention that PAP has a distancing power, develops a sense of efficacy and self-esteem and that the social context plays a role in the psychological impacts [66,67,68,69,70]. Wankel [71], on the other hand, places pleasure in PAP as positively associated with psychological well-being. Finally, the link between physical and psychological health is explained through biological and physiological mechanisms, such as the secretion of certain hormones that have a role in anxiety [72] and, more specifically, in the hormonal regulation that decreases the physiological reactivity of stress [73].Overall, our data lead us to see an evolution in terms of autonomy and environments, which suggests the “ideal pathway” that we propose. Initially, the individual is more conditioned by his or her family, with much diversity and support, but little choice. Then, with the possibility of choosing which activity to invest according to their preferences and their feeling of self-efficacy, they favor an increased and possibly intrinsic motivation, as suggested by Deci and Ryan’s theory of self-determination [16]. Indeed, this theory relates the importance of having autonomy to increase the individual’s intrinsic motivation. Having autonomy-supporting contexts allow the person to choose in which opportunities to invest and to obtain positive feedback regarding this choice promotes motivation. Then, the social and physical environment becomes very important. The diversity and proximity of infrastructure has a positive effect on healthy and active living. Even better, the data suggest that it is the social support related to the built environment that can make a difference.Finally, the ideal pathway ends with the individual at the center of their decisions, in complete autonomy. At this stage, the individual’s extensive knowledge of HLHs helps him or her make choices, while relying on the well-being that the PAP brings. The individual is still constrained by their environment; hence, the importance of using certain maintenance mechanisms, such as routine, friends, pleasure, the outdoors and balance among the spheres of life.Our results suggest that the winning conditions for the adoption and maintenance of HLHs vary according to a multitude of factors, following an evolution under several environments (e.g., family, social, physical and individual) that change over the course of the individual’s life. This leads us to ask several questions. What, of all these winning conditions, that is, of all the factors that have a positive impact on HLHs, plays the key role in long-term maintenance? Is it the accumulation of all these facilitating factors or is it the key role of some of them, arriving at the right time in a person’s life? Overall, we found many similarities between the two group profiles under study, which led us to believe that our results follow certain regularities. However, how can we know the degree of impact for these winning conditions according to the age of the individual? It would be interesting to investigate the different phases of development, from childhood to adulthood, to better understand the factors that influence healthy lifestyles in people’s lives. Studies in this area would allow us to better target the interventions to be carried out with the desired clientele in health promotion.On another note, our methodology, despite its small number of participants, has demonstrated its ability to obtain encouraging results for the field. Continuing to work on and refining qualitative designs to obtain more information on the long-term aspect seems to be a promising avenue for this type of research study. The originality of our design, in its methodology and in the preliminary modelling it proposes (storytelling), could help renew approaches in health promotion. Conducting longitudinal quantitative studies is very costly in terms of time and money. For this reason, it becomes relevant to work on qualitative specifications that open the methodology on long-term maintenance, investigating the life courses of people to understand what has mattered to them and made a difference. To this end, opting for studies with other profiles than those discussed here, in order to observe similarities or disparities, could provide a more complete picture of our understanding of the subject and would allow us to attempt modelling adapted to various social contexts.Finally, our participants’ view of their lifestyle suggests the important place that psychological health occupies in their lives. Therefore, this dimension of health should be taken into account in future research.One of the limitations of this research is the size of our sample. Initially, four group profiles were selected, but, in the end, only two profiles were selected (two groups at the Bachelor’s level in PE and 1 group at the Master’s level), for a total of 25 participants. This decline occurred as a result of the coronavirus (COVID-19) pandemic, which restricted access to our participants for group interviews. A larger sample size, as well as the more diverse profile of participants (e.g., not necessarily in favor of PAP) would have brought richness to the results, incorporating other desired profile types. Unfortunately, this more varied data collection was not possible.Another limitation is related to possible social biases among our participants, which may have encouraged their participation despite themselves. Indeed, although participation was not mandatory, the interviews were conducted at their university and the researcher was introduced by their professor, which may have encouraged some of them to participate. Another problem also arises with the responses themselves. The subject matter of the study, which asks to talk about one’s own past experiences, can be difficult, first to grasp, to remember and then to verbalize. Indeed, Brown et al. [74] point out that we find it difficult to evaluate what has made us change and learn. On the other hand, we believe that by being aware of the topics discussed because of their field of training, directly related to the topic under study, these participants are better equipped to reflect on them and to verbalize their ideas.Finally, the last limitation is related to the difficulty of generalizing our results. Indeed, our sample of participants was purposive and was highly targeted in this study. It does not necessarily represent the entire population, although the consistency of our results with those of many authors suggests a strong potential for transferability. However, we believe that this aspect of our research is a strength in itself, as the objective was to raise the elements that contribute to the adoption and maintenance of HLHs, beyond the barriers recognized by the literature. It became more than relevant to investigate the backgrounds of people who had the potential to have a significant repertoire, experience and reflection in terms of HLHs, in order to better highlight the important determinants of the adoption and maintenance of HLHs.This research, which was both inductive and exploratory in nature, sought to advance knowledge on what can lead to the adoption of HLHs and, more specifically, to their maintenance. The methodology allowed us to meet our research objectives, using a multi-step analysis, following the principles of Paillé and Mucchielli’s [37] conceptualizing categories method. At the end, an ideal pathway in the form of a preliminary modeling of the evolutionary mechanisms of HLH adoption and maintenance allowed us to answer our research question. This model suggests that a multitude of factors act on the individual and his or her lifestyle, such as several environmental factors (family, social, physical and individual), which brings into play several key phenomena in the adoption and maintenance of HLHs (e.g., physical and psychological health, peers, the outdoors, motivation, pleasure, intrinsic awareness of benefits) and this unfolds over the course of an increasing autonomy in the paths of our participants.These results may be an interesting contribution to research, considering the lack of long-term evidence in the field of health promotion. The impact of the different winning conditions on adopting and maintaining HLHs, conditions put forward in this research study, remains to be further investigated. One of the ways that seems promising to achieve this is to study them according to the phases of development, from childhood to adulthood. We believe that the present study is part of an evolution of methodological practices in the field of health promotion, with a design that tends to raise questions about the sustainability of HLHs. Therefore, we hope that this research study helps to inspire the methodology of future studies. Indeed, the conclusions drawn from this project could guide physical educators towards teaching strategies aimed at the diversity of the proposed activities, pleasure, the role of peers, motivation, etc. In addition, the creation of partnerships between the school and parents could be an interesting course of action, especially since our results show the importance of the family environment in the adoption of HLHs. In the end, we hope that this study allows the emergence of qualitative or mixed research methods in order to open the horizons in our approach to the notion of adoption and maintenance of HLHs.Conceptualization, L.C.-W., P.C. and T.G.; methodology, L.C.-W. and P.C.; validation, L.C.-W., P.C. and T.G.; formal analysis, L.C.-W., P.C. and T.G.; writing—original draft preparation, L.C.-W., P.C. and T.G.; writing—review and editing, L.C.-W., P.C. and T.G.; visualization, L.C.-W.; supervision, P.C. and T.G.; project administration, P.C.; funding acquisition, T.G. All authors have read and agreed to the published version of the manuscript.This research received external funding from the Government of Quebec and the Ministry of Education.Certificate was obtained from Ethic Committee at the University of Quebec in Montreal for this study the 2019-11-18 (#3533).Informed consent was obtained from all subjects involved in the study.Authors would like to gratefully thank all participants of this study, Antony Karelis and Kelsey Needham Dancause for the revision of the English and Sylvain Turcotte and Paquito Bernard for the revision of the content of the manuscript and the study.The authors declare no conflict of interest.Expériences des participants en lien avec des SHV (leur pratique)Les changements de comportements (des avant/après de certains éléments biographiques, ou bien encore des changements progressifs, parfois avec des retours en arrière)Moments ou périodes dans leur parcours où les changements sont survenusEn lien avec votre propre expérience, en quoi votre mode de vie est sain et actif ?Quels sont les moments ou les périodes de votre vie où vous pouvez dire que vous avez adopté de SHV ?Vous les avez maintenus ?Comment vos SHV ont- elles évolué à travers votre vie ?Les éléments facilitateurs dans leur parcours pour adopter et/ou maintenir leurs SHV
+					  -Environnement physique et social-Contexte familial-École-Caractéristiques personnelles-Etc.Environnement physique et socialContexte familialÉcoleCaractéristiques personnellesEtc.Type de changement:
+-Expérience précise et transformatrice-Expérience cumulative-Déclic-Etc.Expérience précise et transformatriceExpérience cumulativeDéclicEtc.En lien avec votre propre expérience, qu’est-ce qui a fait la différence dans le choix d’adopter des SHV ?De les maintenir ?Pouvez-vous énumérer des facteurs ayant eu un impact dans vos choix de comportements sains ?Vos changements sont dus à quels genres d’expériences ?The three most important winning conditions for the adoption and maintenance of healthy lifestyle habits among participants in both profiles (the symbol below health in Figure 1 means Physical activity practice, family and mental health, the same as below).The social environment as a winning condition for the adoption and maintenance of healthy lifestyle habits.Role of enjoyment, self-efficacy, motivation and the outdoors in the adoption and maintenance of healthy lifestyle habits.Intrinsic awareness of the benefits of physical activity as a driver for adoption and maintenance of healthy lifestyle habits.Summary of the winning conditions for the adoption and maintenance of healthy lifestyle habits.Conceptual model of evolutionary mechanisms of healthy lifestyle habit adoption and maintenance.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11058.txt

@@ -0,0 +1 @@
+The identification of COVID-19 waves is a matter of the utmost importance, both for research and decision making. This study uses COVID-19 information from the 52 municipalities of the Metropolitan Region, Chile, and presents a quantitative method—based on weekly accumulated incidence rates—to define COVID-19 waves. We explore three different criteria to define the duration of a wave, and performed a sensitivity analysis using multivariate linear models to show their commonalities and differences. The results show that, compared to a benchmark definition (a 100-day wave), the estimations using longer periods of study are worse in terms of the model’s overall fit (adjusted R2). The article shows that defining a COVID-19 wave is not necessarily simple, and has consequences when performing data analysis. The results highlight the need to adopt well-defined and well-justified definitions for COVID-19 waves, since these methodological choices can have an impact in research and policy making. The COVID-19 pandemic became a major global event since its identification in 2019. During 2020, countries around the world dealt with the crisis under a scenario of uncertainty, including what will happen in 2021 [1,2]. The second year of the pandemic brought new analyses and changes that need to be considered when looking at the impact of COVID-19 in different settings [3,4]. The availability of vaccines, the rise in new variants, and the lessons from the experiences of 2020 pose new challenges for policymakers around the world.Some of the issues that are relevant to extract lessons and think of solutions are whether there have been several waves of the disease, how to identify them, and if they are different [5,6,7,8]. From a policy perspective, the identification of “waves” is relevant to generate data and analysis for decision making [9].These differences in the evolution of the pandemic can be found between countries and within geographical regions, where COVID-19 infections can be decomposed into a set of asynchronous sub-trajectories originating from different regions within a country [10].Despite how popular the concept of “wave” has become, there is no standard definition to identify them. This lack of an operational definition can lead to confusion when studying and communicating about the pandemic. For this, it is important to dedicate efforts to establishing objective methodologies to define COVID-19 waves, allowing both researchers and decision makers to determine and debate about them.Chile represents an interesting case of study for several reasons. First, it is one of the countries that has been hit more severely by the pandemic [11]. Second, it has been actively proposing new strategies to deal with the pandemic, including the successful mass vaccination process [12,13]. Third, it is a developing, Latin American, unequal country that can be viewed as a reference for the region and other countries. Particularly, it has been shown in previous studies that the pandemic has had an uneven impact within the Metropolitan Region of Chile [14,15]. Considering these features, the goal of this study is—using data from the Metropolitan Region, Chile—to propose a particular method to identify COVID-19 waves and show the non-triviality of choosing different measures to define these waves. The results seek to highlight the importance of defining a baseline method with explicit criteria to identify these waves when performing research on COVID-19 in a determined time and location, and for evidence-based decision making. The results can be used to extend the analysis of the ongoing impact of COVID-19 in Chile, and can be applied—mutatis mutandis—in other contexts.Based on the number of weekly cases—with a PCR-positive test—for each municipality, cumulative incidence rates were calculated as the ratio between the number of cases each week and the estimated population in each municipality, using the projections from the last available census [16] per 100,000 inhabitants. For each municipality, the following thresholds were identified:Starting date: week of the first reported case.End of first wave (criterion 1): first week (since the starting date) that fulfilled the following conditions:Weekly incidence rate lower than 70 cases per 100,000 people (70/100,000 cases);Negative growth incidence rate for at least two consecutive weeks.Start of second wave (criterion 2): first week (since the end of the first wave) that fulfilled the following conditions:Weekly incidence rate higher than 70/100,000 cases;Positive growth incidence rate in at least one week in which the municipality presented over 70/100,000 cases.Average threshold (criterion 3): average week between the end of the first wave and the start of the second wave.Starting date: week of the first reported case.End of first wave (criterion 1): first week (since the starting date) that fulfilled the following conditions:Weekly incidence rate lower than 70 cases per 100,000 people (70/100,000 cases);Negative growth incidence rate for at least two consecutive weeks.Start of second wave (criterion 2): first week (since the end of the first wave) that fulfilled the following conditions:Weekly incidence rate higher than 70/100,000 cases;Positive growth incidence rate in at least one week in which the municipality presented over 70/100,000 cases.Average threshold (criterion 3): average week between the end of the first wave and the start of the second wave.The identification strategy contains the following two criteria: levels and changes. The criterion of 70 cases per 100,000 people was established following the government’s strategy to define geographical COVID-19 measures, the so-called step-by-step plan (Plan Paso a Paso), in which geographically determined restrictions were set according to epidemiological indicators [17]. According to this strategy, a daily incidence rate of 10 cases per 100,000 people was considered as positive and, consequently, it triggered advances in terms of the stringency of measures [18]. In terms of changes, the use of more than one week with low/high cases allows a trend to be identified, avoiding defining a wave based on outliers. Once these thresholds were defined, the duration of waves was calculated as the difference between the starting and end dates for each unit of analysis (municipality), using the three different criteria described above.To examine the potential effect of the choice between definitions, the cumulative incidence of confirmed cases and deaths due to COVID-19 was calculated for each municipality over the three periods. These first six variables were calculated as the number of COVID-19 confirmed cases/deaths per 100,000 people in each unit of analysis according to the period studied. Based on the above calculation, each of these was divided by the duration of the first wave in days (according to the corresponding criterion), thus obtaining variables adjusted for the duration of each wave.As an exercise to test the possible effects of the choice between the different definitions, a sensitivity analysis was performed, using the previously described variables as dependent variables in a series of multivariate linear regressions for the 52 municipalities of the Metropolitan Region, Chile. These models were estimated using ordinary least squares (OLS), where the explanatory variables correspond to different demographic, health and socioeconomic factors of each unit of analysis, calculated from information from the CASEN 2017 [19]. Then, for each model, an automatic selection algorithm (stepwise) of explanatory variables was used, simplifying the number of independent variables to the subset that minimizes the AIC criterion. In addition, for the residuals of each estimated model, the Moran’s I test was calculated to detect the existence and degree of spatial autocorrelation.Finally, the simplified models were compared with the results obtained in the previous study “COVID-19 incidence and mortality in the Metropolitan Region, Chile: Time, space, and structural factors” in order to analyze whether the results vary according to the choice of criteria used to determine the duration of the wave. The study uses the number of cases—defined as the sum of confirmed and probable cases—as the main input to identify COVID-19 waves. According to the Chilean Ministry of Health, a confirmed case corresponds to a person who has a positive result of SARS-CoV-2 from an RT-PCR test or who has a positive result from an antigen test for SARS-CoV-2, since it was a suspected case (has symptoms, possible reinfection or has a serious respiratory infection that requires hospitalization) [20,21]. The variable was collected weekly, starting on 30 March 2020 to 9 August 2021. For the study, official information coming from the Chilean Ministry of Health (MINSAL) and published by the Ministry of Science was used [22]. The number of deaths due COVID-19 was obtained from open data provided weekly by the Department of Health Statistics and Information (DEIS) [23]. All the data used for this study correspond to data coming from public and open sources, and consider information for the 52 municipalities in the Metropolitan Region, Chile. Table 1 shows the descriptive statistics of the sample, using the different criteria previously described. First, the wave’s duration varies, on average, between 130 and 300 days, using the end of the first wave or the beginning of the second as the criterion. As expected, the cases and deaths follow the same pattern. When adjusting for each municipality’s wave duration, the daily cases range between 15.27 and 25.79, and the daily deaths between 0.59 and 0.95. Finally, the presence of spatial correlation is observed in all the cases.Figure 1 shows an example of the identification strategy, using the municipality of Pudahuel (253,139 inhabitants) [16]. First, the x-axis registers the number of cases per 100,000 people in the municipality, while the weeks of the year are on the y-axis. Second, the horizontal line depicts the threshold of the 70/100,000 cases used to identify the start/end of waves. Third, the vertical lines identify the thresholds for the wave’s duration, using the following three previously described criteria: blue = criterion 1; purple = criterion 2; green = criterion 3. Finally, the colors in the bars identify the different stages of the step-by-step strategy, by level of stringency, as follows: red = stage 1 (quarantine); orange = stage 2 (transition); yellow = stage 3 (preparation); blue = stage 4 (initial opening); green = stage 5 (no restriction).Figure 2 shows this information for the 52 municipalities of the Metropolitan Region. First, it is interesting to note the different patterns arising from the analysis of different units. In terms of the evolution of cases, there are municipalities that exhibit two clearly defined modes in the distribution of cases (urban municipalities), while, in others, the identification of the end of wave 1 is unclear (cases keep moving up and down), or whether there are more than two waves over the period (rural municipalities). Finally, there are differences in the application of the step-by-step strategy, with municipalities facing early and prolonged quarantines, and others having less stringent measures, both during the first wave (2020) and the second wave (2021).Finally, Table 2 and Table 3 show the results from the regressions using the different criteria and dependent variables. In both cases, the results are contrasted against the ones reported by using a 100-day period.Table 2 exhibits the results for the number of cases and the cases adjusted by the duration of the wave. First, compared to the benchmark model (first 100 days in each municipality) variables, multidimensional poverty and the use of public transportation are also significant to explain both case rates, and the cases adjusted by duration. Interestingly, population density does not explain variation in the dependent variable when extending the period of analysis, while the distance to a health center becomes significant. In almost all models, the Moran’s I test for residuals appears as significant, showing the presence of spatial autocorrelation in the regression. In terms of the model as a whole, the adjusted R2 is larger for the benchmark model (100 days), while statistical correlation is not removed from the regression when using wave 1 (definitions using the three proposed criteria) versus the benchmark model. However, when looking at cases adjusted by duration, criterion 3 outperforms the rest.Table 3 shows the results for death-related variables. For this set of models, the results seem more stable, with the same group of variables appearing as statistically significant in every model. As with the cases, when looking at deaths, the best overall fit corresponds to the model using 100 days; while the model using criterion 3 has a larger adjusted R2 for deaths/duration. Spatial correlation of residuals is not present in any model.The study proposes a method that, based on the weekly cumulative rate of cases, allows waves of COVID-19 to be identified. The method, with its three variants, was used to define the study period when analyzing the impact of COVID-19 (using rates and duration-adjusted rates). The results show that period identification and its impact are far from simple.The different criteria used to define COVID-19 waves show that the decision is not trivial; both the impact and its determinants can be affected by this methodological decision, confirming the need to build a common definition that will serve as a basis for researchers and decision makers when determining COVID-19 waves. In terms of factors affecting COVID-19 cases and deaths, the proposed models seem to work better—in terms of overall fit—in explaining high-incidence periods (e.g., the 100 days and criterion 1 models); when taking into account duration, a conservative criterion (criterion 3) seems to be a reasonable choice to identify the “true” duration of the waves.The analysis highlights the importance of considering time as a key factor, both to understand the causes and effects of the pandemic, and to design policies to address it; for example, it is relevant to assess governments’ interventions throughout the pandemic [24]. It also emphasizes the need to acknowledge the researchers’ methodological choices and their consequences in the results [25]. Although the data were restricted to analyzing COVID-19 in a particular setting, the analysis could be used—mutatis mutandis—in other settings and for different health problems. We hope that this method will contribute to the creation of a formal and common definition of a “wave”, allowing researchers and authorities to study and communicate about the causes and consequences of the pandemic. The goal of the analysis is not to propose the method, but to highlight the need to have one method when studying and deciding about COVID-19 waves. Other approaches have also been proposed (for example, based on the effective reproduction number, R [26]); the study highlights the need to advance towards adopting an objective measure as a way to enhance academic and policy dialogue.The study has several limitations that need to be considered when interpreting the results. First, the initial motivation for the study was to highlight the need to have a “complete picture” for analyzing COVID-19 data; although the study extends the previous (restricted) definition of a period of 100 days, it needs to acknowledge that COVID-19 is still ongoing, and future data can allow new and different types of results. Second, even though the general method to define waves can be replicated in other contexts, it requires adaptation (e.g., the number of cases to define the threshold). Third, the criteria for defining waves are based exclusively on the total number of confirmed cases. As previously stated, alternatively, the analysis can be performed using different variables, such as deaths or cases by COVID-19 variants, and considering testing in the region of study.The information presented adds to the current literature on the causes and impact of COVID-19. It can be useful to monitor data; for example, to identify the rise in future waves in the country. We hope that these results motivate other researchers to assess other ways to solve these problems and contribute to fostering an evidence-based debate on COVID-19.Conceptualization, A.A., P.V.D., F.E., C.C., C.V. and M.M.; methodology, A.A., P.V.D. and F.E.; software, A.A.; validation, A.A., P.V.D. and F.E.; formal analysis, A.A.; data curation, A.A.; writing—original draft preparation, A.A. and P.V.D.; writing—review and editing, A.A., P.V.D., F.E., C.C., C.V. and M.M.; visualization, A.A.; project administration, F.E. and M.M.; funding acquisition, F.E. and M.M. All authors have read and agreed to the published version of the manuscript.This research was funded by Universidad de Santiago, DICYT Asociativo Grant 022191MH DAS. The APC was funded by Universidad de Santiago, Dirección de Investigación Científica y Tecnológica, Vicerrectoría de Investigación, Desarrollo e Innovación.Not applicable.Not applicable.Data used in the study is available at: https://github.com/MinCiencia/Datos-COVID19.The authors thank the support of DICYT Asociativo Grant 022191MH DAS to carry out the project “Understanding causes and consequences of COVID-19 and other diseases on population health: multicausality, time, and space for decision-making”.The authors declare no conflict of interest. Waves’ identification example: Pudahuel.Cases, wave thresholds, and step-by-step stages in the Metropolitan Region.Wave 1 statistics using different criteria.Significance level: *** p < 0.01, * p < 0.1.OLS regression under different scenarios: case rate.Significance level: *** p < 0.01, ** p < 0.05, * p < 0.1.OLS regression under different scenarios: death rate.Significance level: *** p < 0.01, ** p < 0.05, * p < 0.1.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11059.txt

@@ -0,0 +1 @@
+This study aimed to explore sex-specific latent class models of adolescent obesogenic behaviors (OBs), predictors of latent class membership (LCM), and associations between LCM and weight-related outcomes (i.e., weight status and unhealthy weight control behaviors). We analyzed nationally representative data from the 2019 Korea Youth Risk Behavior Survey. To identify latent classes for boys (n = 29,841) and girls (n = 27,462), we conducted a multiple-group latent class analysis using eight OBs (e.g., breakfast skipping, physical activity, and tobacco product use). Moreover, we performed a multinomial logistic regression analysis and a three-step method to examine associations of LCM with predictors and weight-related outcomes. Among both sexes, the 3-class models best fit the data: (a) mostly healthy behavior class, (b) poor dietary habits and high Internet use class, and (c) poor dietary habits and substance use class. School year, residential area, academic performance, and psychological status predicted the LCM for both sexes. In addition, perceived economic status predicted the LCM for girls. The distribution of weight-related outcomes differed across sex-specific classes. Our findings highlight the importance of developing obesity prevention and treatment interventions tailored to each homogeneous pattern of adolescent OBs, considering differences in their associations with predictors and weight-related outcomes.Adolescence is one of the critical development periods, with an increasing risk of obesity [1]. The increase in adolescent obesity has become a global public health concern [2]. For example, obesity prevalence in U.S. adolescents increased from 18.1% in 2007–2008 to 21.2% in 2017–2018 [3]. Researchers have reported on a marked increase in the prevalence of adolescents with obesity in East Asia [2]. Indeed, the prevalence of obesity in Korean adolescents has doubled from 5.3% in 2007 to 10.8% in 2018 [4]. Adolescent obesity causes physical (e.g., hypertension, insulin resistance, and gastrointestinal disease) and mental (e.g., depression, anxiety, and eating disorder) problems, besides increasing healthcare costs [5,6]. Furthermore, adolescents with obesity are more likely to be obese in adulthood, with greater risks of cardiovascular disease, type 2 diabetes, and cancer [7]. The prevention of adolescent obesity-related problems necessitates special attention to unhealthy weight-control behaviors (UWCBs) as well as weight status [8]. UWCBs (e.g., fasting, skipping meals, using weight-loss pills, and vomiting) are prevalent among adolescents [9]. For example, the prevalence of UWCBs in U.S. and Korean adolescents range from 13.6% to 15.0% [10,11]. In addition, the literature indicated that adolescents’ UWCBs are associated with increased body mass index (BMI), eating disorders, depression, and suicidal behaviors [9,11,12]. Considering the high rates and consequences of adolescent obesity and UWCB, it is imperative to develop effective strategies to prevent and treat weight-related problems. The theoretical framework guiding Project EAT provides the basis for the importance of obesogenic behaviors (OBs) for improving weight-related outcomes [13]. Adolescents’ OBs (e.g., poor dietary habits, low physical activities, sedentary behaviors, and substance use) are associated with their weight status and UWCB [14,15,16]. Furthermore, these OBs occur concurrently and interact with each other [17], besides exerting combined effects on obesity-related health [18]. Considering the limitations of single-dimensional approaches (e.g., focusing on a single OB or the number of OBs) in explaining the complexity and multidimensionality of OBs, it is necessary to identify patterns of multiple OBs and develop differential interventions reflecting the characteristics of each pattern [18].Thus, recent researchers have identified latent classes or clusters of adolescent OBs and have investigated their associations with weight-related outcomes [18,19,20]. In particular, latent class analysis (LCA), one of the person-centered approaches, is considered suitable for (a) identifying heterogeneous subgroups based on multidimensional characteristics of human behaviors [21] and (b) examining effects of tailored interventions considering characteristics of target subgroups [22]. Most LCA studies identifying patterns of adolescent OBs demonstrated that multiple OBs were divided into three to five heterogeneous subgroups [19], associated with weight status, perceived overweight, and body dissatisfaction [19,20,23,24,25]. However, there is little information available on the association between OB patterns and UWCB. Given the prevalence and detrimental consequences of UWCB [8,9,11,12], further investigation on the association between OB patterns and UWCB among adolescents is necessary.Furthermore, previous studies examining patterns of adolescent OBs are limited in two aspects. First, the majority of the existing LCA studies have limited information on sex differences in OBs [19]. It is necessary to identify sex-specific OB patterns, considering sex differences in exposure levels and the vulnerability to obesogenic environments, weight-related outcomes, and responses to obesity interventions [26]. Second, despite adolescent OBs being affected by cultural characteristics [19], all LCA studies investigating OB patterns have been conducted in Western countries [19,20,24]. Thus, this study aimed to explore sex-specific patterns of OBs among Korean adolescents, to identify predictors of latent class membership (LCM), and to investigate if distinct OB classes exert differential effects on weight status and UWCB.We used data acquired from the 15th Korea Youth Risk Behavior Survey (KYRBS) conducted in 2019. This web-based survey is conducted annually to identify health behaviors such as dietary behavior, physical activity, and substance use of Korean adolescents [27]. The 15th KYRBS used a stratified multi-stage cluster sample design to collect a nationally representative sample of 7th–12th grade students. The sample comprised 57,303 students from 400 middle schools and 400 high schools [27]. Student age ranged from 12 to 18 years old (mean = 15.08, standard deviation = 1.78). We obtained an exemption from an institutional review board prior to the initiation of this study because it used a de-identified data set (IRB No. CUIRB-2021-E008). To estimate sex-specific patterns of adolescent OBs, we selected and dichotomized eight indicators based on the literature [15,16,19,20,28]. The questionnaire consisted of three parts: dietary behavior, physical activity and sedentary behavior, and substance use. First, we assessed four dietary behaviors by inquiring about the following eating habits during the past seven days: (a) >2 days of breakfast skipping [29], (b) ≥3 days of sugar-sweetened beverage (e.g., soft drinks, carbonated drinks, juice or flavored drinks; SSB) intake [24], (c) ≥3 days of fast food consumption [15], and (d) non-daily fruit and vegetable consumption [30]. Second, we used two questions regarding their total physical activity and non-academic Internet use (NAIU) to assess their physical activity and sedentary behavior. Total physical activity was assessed by inquiring if the participants had engaged in any physical activity that increased their heart rate and caused a shortness of breath for at least 60 min each day during the past seven days [28]. To assess NAIU, we investigated the average hours of using the Internet for non-academic purposes on weekdays and weekends. We calculated the average daily hours spent in NAIU using the following formula: (5 × the average NAIU hours on weekdays + 2 × the average NAIU hours on weekends)/7. The responses were classified into “≥2 h” and “<2 h” per day [16]. Third, we assessed substance use by asking the following: (a) monthly tobacco product use (i.e., using at least one of cigarette, electronic cigarette, or heated tobacco product in the past 30 days) and (b) their monthly alcohol use (i.e., alcohol consumption at least once in the past 30 days). To identify the predictors associated with LCM, we selected factors based on the literature [18,31,32,33,34,35,36]. We included demographics (i.e., school year, area of residence, and perceived economic status), academic performance, and psychological status (i.e., stress, depressive feeling, and sleep satisfaction). School year was classified into “middle school” and “high school.” Area of residence was divided into “suburban or rural area” and “urban area.” Perceived economic status and academic performance were assessed using a 5-point Likert scale, ranging from “very high” (0) to “very low” (4). We dichotomized the responses into two categories: “high or middle” and “low.” Stress was assessed by asking participants about their usual level of stress. Possible responses ranged from “very low” (0) to “very high” (4) and were categorized into “low or average” and “high.” Depressive feeling was assessed by questioning if the participants had experienced considerable sadness or despair to interrupt their daily activities in the past two weeks. The possible responses were “yes” and “no.” We measured sleep satisfaction by inquiring how satisfied they were with their sleep to relieve fatigue during the past seven days. Possible responses ranged from “very satisfied” (0) to “very dissatisfied” (4) and were categorized into “satisfied or average” and “dissatisfied.”Weight-related outcomes included weight status based on self-reported BMI and UWCB. To assess the weight status, participants were requested to provide their height and weight. After calculating their BMI (i.e., dividing the weight in kilograms by the square of their height in meters), they were classified based on the definition of obesity suggested by Korean Society for the Study of Obesity [37]. Participants were classified according to the percentile of BMI by their sex and age: “obese” (i.e., ≥95th percentile) and “non-obese” (i.e., <95th percentile). We measured UWCB by asking if at least one of the following methods was used for weight control in the past 30 days: (a) fasting for ≥24 h, (b) reduced food intake, (c) restricting one’s diet to a specific food (e.g., egg, milk, and grapes), (d) vomiting after having meals, (e) using non-prescribed weight-loss pills, (f) taking diuretics or laxatives, and (g) consuming weight-loss supplements. The possible responses were “yes” and “no.”We performed four phases of data analysis using SAS version 9.4: identifying (a) sample characteristics, (b) sex-specific LCMs of OBs, (c) significant predictors of the LCM, and (d) class-specific distribution estimates for weight-related outcomes by sex. First, we used descriptive statistics (i.e., frequency and weighted percentages) of the sample characteristics by sex. Second, to investigate sex-specific LCMs of OBs, we conducted a multiple-group LCA using eight OBs [21]. Following the identification of the best fitted latent class model using the entire sample, we estimated a latent class model with freely estimated parameters and another with equally constrained item-response probabilities across sex. This helped us confirm the establishment of measurement invariance [21,38]. While performing the likelihood-ratio difference test to examine differences in the G2s with degrees of freedom for each model, a significant p-value indicated that the measurement invariance should be rejected. In other words, latent class models should be separately estimated by sex [38]. To select the optimal number of latent classes, we successively estimated 1- to 5-class models. The accurate number of classes were selected based on model fit indices, model identification, parsimony, and model interpretability. In relation to model fit indices, higher entropy and lower likelihood-ratio statistic (G2), the Akaike information criterion (AIC), Bayesian information criterion (BIC), adjusted BIC, and log-likelihood suggested a better fitted model [21]. Higher percentage of seeds associated with the best fitted model indicated better model identification [38]. In addition, a simpler and theoretically interpretable model was preferable [21]. Third, we performed an LCA with covariates to identify characteristics predicting sex-specific LCMs. We included seven categorical factors dummy coded in the multinomial logistic regression analysis. A p-value less than 0.050 indicates that the distribution of LCM differed across the covariate of interest [21]. Fourth, we investigated class differences in two weight-related outcomes (i.e., weight status and UWCB) using the Bolck, Croon, and Hagenaars (BCH) approach for LCA with a distal outcome [39,40]. The BCH method, one of the three-step approaches, is advantageous in calculating more accurate distal outcome estimates than merely using posterior probabilities. This is because the aforementioned method takes into account uncertainty owing to the possibility of misclassification [39]. After estimating class-conditional probabilities of the distal outcomes, we conducted a separate Wald test to compare them between each pair of latent classes [39].The total 57,303 adolescents comprised 29,841 and 27,462 boys and girls, respectively. Of these adolescents, 13.8% boys and 8.2% girls had obesity. Moreover, 26.0% and 46.8% boys and girls engaged in UWCB, respectively. Table 1 summarizes the sample characteristics by sex.Before identifying sex-specific LCMs, we estimated a latent class model using the entire sample to determine the establishment of measurement invariance according to sex. G2, AIC, BIC, and adjusted BIC continued to decrease with an increase in the number of latent classes (Table 2). However, the model identification, parsimony, and entropy values indicated that the 3-or 4-class models were most suitable. We selected the 3-class model considering a small change in the model fit indices between the aforementioned models, a higher entropy value, and higher model identification in 3-class model than 4-class model. Moreover, the 3-class model was more parsimonious and theoretically easier to interpret. To assess measurement invariance across sex, the likelihood-ratio difference test was performed between the constrained model and the unconstrained model. G2 values in the unconstrained and constrained models were 1871.18 (df = 459) and 4359.98 (df = 483), respectively. The result indicated that measurement invariance across sex was not established (ΔG2 = 2488.80, df = 24, p < 0.001); thus, sex-specific latent class models were estimated by separating data of boys and girls [38].In both samples, the 3-or 4-class models were suitable for the data. We eventually selected the 3-class model because of relatively small differences in model indices between 3- and 4-class models, higher entropy value and higher model identification, and its simplicity and easy interpretation (Table 2). The 3-class models were presented: mostly healthy behaviors (MH), poor dietary habits and high Internet use (PDHI), and poor dietary habits and substance use (PDSU) in boys and girls (Table 3). In both sexes, the MH class was the largest group (53.9% and 47.9% of boys and girls, respectively), followed by PDHI (35.4% and 44.8%) and PDSU (10.8% and 7.4%). Overall, the probabilities of non-daily fruit and vegetable consumption in all classes were higher than 85.0%. The MH class was the least likely to engage in all OBs, compared to other classes. Compared to the MH class, adolescents belonging to the PDHI class displayed higher probabilities of skipping breakfast, SSB and fast food intake, non-daily fruit and vegetable consumption, and NAIU. Particularly, the PDHI class demonstrated the highest probability of consuming SSB and using the Internet for non-academic purposes compared to other classes. The probability of consuming SSB three or more days a week was 96.6% and 79.5% for boys and girls, respectively. The probability of NAIU for two or more hours was 51.2% and 60.5% for boys and girls, respectively. Those in the PDSU class displayed highest probabilities of engagement in breakfast skipping, non-daily fruit and vegetable consumption, and monthly smoking and drinking. The PDSU classes by sex were distinctly separated by their monthly substance use characteristics. Specifically, boys in the PDSU class exhibited higher probability of monthly tobacco product use than girls (69.2% and 50.3% of boys and girls, respectively). In addition, the probabilities of fast food consumption were less than 50.0% in all classes, but more than three times higher among those in the PDHI and PDSU classes than those in the MH class.To investigate the predictors of LCM, we used the MH class as a reference (Table 4). For both sexes, school year, area of residence, perceived academic performance, stress, depressive feeling, and sleep satisfaction significantly predicted the LCMs. Moreover, perceived economic status predicted the LCM in girls. The risk of being in the PDSU class was significantly higher in high school students than in middle school students (Odds ratios (ORs) = 5.72 and 3.20 for boys and girls, respectively). However, the risk of being in the PDHI class for boys was higher in middle school students than in high school students (OR = 0.86). Among both sexes, those living in suburban or rural areas were at a higher risk of belonging to the PDSU class than their urban counterparts (ORs = 0.82 and 0.75 for boys and girls, respectively). In girls, lower perceived economic status significantly increased the risk of belonging to the PDSU class, compared to the MH class (OR = 1.47). For both sexes, lower perceived academic performance and higher stress, depressive feelings, and sleep dissatisfaction were associated with an increased risk of being in the PDHI and PDSU classes, compared to the MH class. In addition, the three psychological predictors associated with the PDHI and PDSU classes in both sexes generally displayed stronger associations with the PDSU class than the PDHI class.Sex-specific classes of OBs were compared on their weight-related outcomes (Figure 1). The prevalence of obesity in all classes did not significantly differ among boys. In girls, the PDSU class reported on the highest proportion of obesity, followed by the PDHI class; however, the difference was insignificant. The PDSU class reported on a significantly higher rate of obesity than the MH class (p = 0.012). There was no significant difference in the proportion of obesity between the MH and PDHI classes. Regarding the proportion of UWCB, the PDSU class reported on the highest among all classes in boys. Specifically, the PDSU class reported on a significantly higher rate of UWCB compared to the MH class (p < 0.001) and PDHI class (p < 0.001). There was no significant difference in the proportion of UWCB between the MH and PDHI classes among boys. In girls, the PDSU class displayed the highest proportion of UWCB, followed by the MH and the PDHI classes. The PDSU class had significantly higher UWCB compared to the MH class (p < 0.001) and PDHI class (p < 0.001). Interestingly, the MH class reported on significantly higher rate of UWCB engagement, compared to the PDHI class (p = 0.009).This study demonstrated sex-specific latent class models of adolescent OBs. For both sexes, school year, residential area, perceived academic performance, stress, depressive feeling, and sleep satisfaction significantly predicted the LCMs. Moreover, perceived economic status predicted the LCM among girls. We also found differences in sex-specific LCMs in proportions of obesity and UWCB. We identified three classes across sex, namely the MH, PDHI, and PDSU classes. Previous studies on sex-specific patterns of adolescent OBs identified three to five latent classes: (a) the healthy class with high probabilities of healthy dietary habits and adequate physical activities, (b) the physically active class with high probabilities of poor dietary habits, (c) the sedentary class with high probabilities of poor dietary habits, and (d) the substance use class with high probability of poor dietary habits and moderate physical activities [20,23]. Despite the need for cautiously comparing LCMs between studies owing to different OBs included in the LCA and the limited number of studies examining sex-specific LCMs of adolescent OBs, our results were consistent with previous studies [20,23]. OB patterns were divided into classes with the lowest likelihood of OBs, with moderate OBs, and with the most severe OBs. In addition, poor dietary habits were common features of all classes [20,23,24]. Particularly, despite measuring “daily” and “non-daily” fruit and vegetable intake using criteria less stringent than the nutritional guidelines suggested by Korean Ministry of Health and Welfare [30], we observed high probabilities of non-daily fruit and vegetable consumption across all classes. This result is not surprising, considering the previous evidence from studies conducted in the U.S. and Canada [23,24] and the literature describing poor dietary habits in adolescents [41].In our study, the distribution and characteristics of OB patterns overlapped between boys and girls. Nonetheless, some characteristics of OB patterns demonstrated differences by sex. For example, the prevalence and characteristics of PDSU differed by sex. Specifically, the prevalence of the PDSU class and their probability of monthly tobacco product use were considerably higher for boys than for girls, consistent with a recent study [20]. This result may be attributed to a higher rate of substance use in boys than girls, particularly smoking behaviors [4,42]. Our results are also supported by Fleary [20] who suggested that multiple OBs, including health promoting and risk behaviors, occur on a sex-specific continuum. Thus, the efforts to prevent and improve adolescent weight-related outcomes should be preceded by a better understanding of the multidimensional and complex characteristics of sex-specific OB patterns. Some demographic characteristics included as covariates distinguished the class memberships across sex. For example, among boys, high school students are less likely to be in the PDHI class than the MH class. This finding may be partly attributed to the relationship between Internet use and the age and sex among Korean adolescents. The risk of NAIU is higher in boys and middle school students than in girls and high school students [43]. In contrast to boys, girls who perceived lower economic status demonstrated a greater risk of belonging to the PDSU class than those who perceived middle or high economic status. This finding is consistent with the literature [35]. This may be attributed to the engagement of parents or family members of lower socioeconomic status in OBs, which may contribute to adolescent OBs (i.e., modeling), particularly in girls [35]. This phenomenon indicates that the social status of the family is one of important predictors affecting adolescent OBs.Several other demographic characteristics significantly predicted LCMs for both sexes, in accordance with the existing literature. For example, high school students were more likely to be in the PDSU class than middle school students [25,44]. This finding may be attributed to the increase in exposure to obesogenic environments (e.g., peer social pressure for poor dietary habits, greater accessibility to substances) with increasing age [45,46,47]. In addition, the risk of belonging to the PDSU class was significantly higher in suburban or rural students than in urban students. Considering that the residential area did not significantly predict the risk of belonging to the PDHI class, this finding may be partly attributed to higher rates of tobacco product and alcohol use in suburban or rural students compared to their urban counterparts [33,34,48]. The literature indicated that regional characteristics contribute to the difference in substance use among adolescents [49,50]. In rural communities, while social norms and regulations on adolescent substance use are weak, access to substances is relatively easy [49,50]. Consistent with previous studies, the lower the academic performance [36] and the poorer the psychological status [31,32,51], the greater the risk of belonging to the class with higher probabilities of OBs. Those with a greater level of stress, depressive feelings, and sleep dissatisfaction tended to display stronger associations with the PDSU class than the PDHI class. OBs in adolescence have bidirectional relationships with psychological status. Those with poorer psychological status are more likely to engage in screen-based activities and substance use to alleviate stressful and depressive feelings [31,52]. Inadequate nutrition consumption and substance use in adolescence may cause difficulties in normal brain function and mood regulation [53,54]. Moreover, sufficient sleep may not be a priority for adolescents who engage in substance use with their peers [55]. Considering the high vulnerability to psychological problems and negative prospects for OBs in adolescence, those in the PDSU class are a critical subgroup with respect to their OBs and poor psychological status. In addition, those with psychological symptoms may be a target group for early prevention of OBs.Considering the class differences in the proportion of obesity and UWCB, we identified distinct characteristics of sex-specific LCMs of OBs. For example, while there was no significant difference in obesity rates across the latent classes for boys, the obesity rate was the highest in the PDSU class for girls, consistent with previous studies [18,23]. There are two potential reasons for the weakened relationship between obesity and LCM in boys. Boys generally increase muscle mass for a masculine body shape [56] and spend more time in various physical activities than girls [57]. We also found that the PDSU class demonstrated higher UWCB prevalence than other classes, thereby suggesting dietary habits and substance use are robust predictors of UWCB in adolescence. This phenomenon could be attributed to lower body satisfaction in the class characterized by a higher probability of sedentary behavior and poor dietary habits [23]. According to the problem behavior theory, health risk behaviors in adolescence co-occur and co-vary owing to common determinants [58]. Both UWCB and substance use are associated with impulsivity-like traits [59,60]. Substance users are more likely to participate in UWCB requiring relative little effort for rapid weight control, despite the detrimental consequences of UWCB [59].In addition, class memberships in girls were approximately twice likely to adopt UWCB than boys, thus suggesting girls are more enthusiastic about losing weight [14]. This is because girls are more likely to have greater body dissatisfaction than boys [23] and are more likely to adopt UWCB to achieve a slender physique that is considered ideal socio-culturally [56].An unexpected finding was the substantial proportion of UWCB in the MH class. Despite the lowest obesity rate in the MH class for girls, approximately half of them engaged in the UWCB. Similarly, a recent study reported that the proportion of UWCB in the healthy class was lower than that in the severely unhealthy class; however, it was higher than that in the moderately unhealthy classes [20]. This phenomenon may be partly attributed to the fact that despite the lower obesity prevalence in the healthy class compared to other classes, body satisfaction was not significantly higher in the healthy class than in other classes [23]. Moreover, normal-weight girls tend to overestimate their own weight [61]. Therefore, health professionals should not consider the class characterized by healthy behavior as a low-risk group of weight-related outcomes. Instead, they should consider not only distinct OB patterns but also sex-specific associations between each OB pattern and weight-related outcomes.The results of this study should be cautiously interpreted in consideration of potential limitations. First, we could not determine the causal relationship between sex-specific LCMs and weight-related outcomes because of cross-sectional data. Second, the BMI was based on self-reported weight and height of the participants. However, BMI based on self-reported parameters accurately differentiates obese adolescents [62]. Third, as predictors of sex-specific LCMs, we did not include developmental and environmental characteristics (e.g., impulsivity, obesogenic home environment, and peer pressure on OBs) that affected weight-related outcomes [46,47,60].This study identified the 3-class models of OBs in boys and girls using LCA, namely the MH, PDHI, and PDSU classes. We found sex-specific findings on the distribution and characteristics of the three classes and the association of sex-specific LCMs with potential predictors. Rather than focusing on a single OB, health professionals should develop interventions tailored to sex-specific adolescent OB patterns [20]. This study extended previous research on adolescent OBs by examining the relationship between sex-specific latent classes and weight-related outcomes. Classes with low probabilities of OB do not necessarily have better weight-related outcomes. This necessitates providing interventions that precisely target each latent class while considering differences in its association with weight-related outcomes. For example, interventions for boys and girls in the PDSU class with a high prevalence of UWCB should not only target multiple modifiable OBs but also include healthy weight-control methods. In addition, interventions for girls in the MH class with a substantial prevalence of UWCB should include healthy weight-control methods despite their low likelihood of OBs and obesity.Future research on adolescent OBs should consider the following issues. First, longitudinal studies are required to establish causal relationships between OB patterns and later weight-related outcomes in adolescence. Second, as this is the first study to examine the association between latent classes of OBs and UWCB, repeated studies on class differences in UWCB among adolescents from different cultures are warranted. Third, further studies should include a wider variety of potentially associated factors (e.g., family, school, and peer environments, developmental characteristics, and weight-related concerns) to refine the characteristics of adolescent OB patterns [13].Conceptualization, H.L.; methodology, H.L.; formal analysis, H.L.; validation, I.-S.L.; writing—original draft preparation, H.L. and I.-S.L.; writing—review and editing, H.L. and I.-S.L. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Daegu Catholic University (CUIRB-2021-E008).Subject consent was waived due to the use of secondary data for this study.No new data were created or analyzed in this study. Data sharing is not applicable to this article.The authors declare no conflict of interest.OB: obesogenic behaviors; UWCB: Unhealthy weight-control behavior; LCA: latent class analysis; LCM: latent class membership; MH: mostly healthy behaviors; PDHI: poor dietary habits and high Internet use; PDSU: poor dietary habits and substance use; NAIU: non-academic Internet use; OR: Odds ratio.Proportion of distal outcomes by latent class (n = 57,303). Note. Estimates with the different subscripted letter indicate statistically significant differences at the 0.050 alpha level. * Indicates the proportion of obese participants.Sample characteristics (n = 57,303).* Unweighted frequency and weighted percentage.Fit statistics of latent classes of obesogenic behaviors among boys and girls (n = 57,303).Note. Bold letters indicate the best fitting models. G2 = the likelihood-ratio statistic; AIC = Akaike’s information criterion; BIC = Bayesian information criterion.Item-response probabilities of obesogenic behaviors among boys and girls (n = 57,303).Note. Bold figures indicate that the item-response probability is 0.500 or above. SSB = sugar-sweetened beverage.Predictors of latent class membership among boys and girls (n = 57,303).Note. The reference group = mostly healthy behaviors (MH). OR = odds ratio; CI = confidence interval; ref. = reference.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11060.txt

@@ -0,0 +1 @@
+This study explores patients’ opportunities for collective participation in an institution for people with substance use disorder. Patients and staff from the treatment institution cooperated with researchers to make changes in the treatment practice, using a research circle as a model for the project. In the article, we discuss the following research questions: How and in what areas did patients have the opportunity to participate collectively in the treatment institution? How did the patients experience participation in the research circle? Data consist of minutes from meetings, seminars, and focus-group interviews. The participants analysed the material together, and the authors carried out a thematic analysis after the project. The participants chose to explore how milieu therapy could build a bridge from treatment in the institution to life after treatment, a “Bridge over troubled waters”, to quote Simon and Garfunkel. Findings show that activities in the research circle led to changes at the institution, e.g., regular Sunday afternoon meetings, a weekly quiz, and less controlling procedures of substance use, and that the institutional culture in general became based more on participation and equality. Patients, staff, and researchers participated in a partnership; mutual recognition promoted cooperation and fellowship in the research circle. We conclude that the project provided the participants with opportunities for collective participation in the institution. In addition, the patients experienced partnership and empowerment in the research circle. Our attempts to change institutional practices yielded some improvements but also met with structural and cultural barriers. Thus, the project experienced challenges and obstacles mostly related to limitations in the institutional system and culture.User involvement is a widely accepted goal in health and social services. Plans and strategies for services for people with substance use problems emphasize that participation is a movement towards openness in public services, increase of legitimacy, and a prerequisite for developing good services [1]. The Norwegian Patient and User Rights Act [2] states that patients in treatment institutions are participants with the right to influence treatment and service development [2]. However, research and public documents point to several challenges in services for people with substance use problems [3,4].Firstly, research and public documents show a lack of individual participation in institutions for people with substance use problems [3,4,5,6,7]. Substance users in treatment have pointed out that they are not listened to and that power differences and hierarchy prevent participation [8,9]. Research shows that treatment does not prepare patients for transition to life outside the institution. After treatment, patients lack the skills to achieve a meaningful life, including employment, housing, and a social network [3,10]. Skudal et al. highlighted that treatment provided by substance use institutions rarely gives patients the skills they need to master life outside the institution [3]. Their study shows that 72 percent of patients experience little preparation for life after substance use treatment and that 86 percent experience poor follow-up after treatment. In another study, Brorson et al. [11] pointed out that institutions must emphasise patients’ experiences to learn what patients require to cope with life after treatment.Secondly, even if user participation and user involvement are concepts with positive connotations, the content of the concept is still blurry. Nevertheless, several studies in drug treatment services have found that participatory practices positively impact treatment outcomes and relations between patients and service providers [3,5,6,7]. Rance and Treloar found that participatory practices created opportunities for better communication and that relationships built on participation enhanced the therapeutic alliance and functioned as resources for empowering service users [7]. Karlsson and Borg found that empowerment and regaining control of one’s own life were crucial elements for patients with substance use disorder [12]. Borge and Hummelvoll concluded that milieu therapy, which they describe as “a socio-cultural learning model for understanding patients’ learning processes through therapy and participation in an institutional milieu” [13] (p. 40), can provide opportunities for reflection, understanding, and learning of skills. They emphasise that learning processes must occur in interaction and collaboration with fellow patients and staff [13]. Other practice studies show that language, culture, and discursive processes are essential in determining how patients are understood [14]. It takes time for substance users to change their perceptions from being addicted to substances to becoming participants in society [15].Thirdly, in services for people with substance use problems, scant attention is paid to collective user involvement to develop and improve services. However, in other fields, research highlights that involving experiential knowledge from users opens new perspectives when developing services [16,17,18,19,20]. There is a demand for more knowledge about how institutions practice participation and the advantages and disadvantages of different practices [21], for methods or strategies for collective user participation [22] and dialogue between professionals and users [23].To meet the need for more knowledge about collective participation in substance use treatment, we developed an action research project in collaboration with patients and staff at a substance use treatment institution. We used a research circle as a strategy for collective participation. A research circle is an arena or a forum for joint knowledge production in a dialogue between the participants in the circle, often with participants from the practice field and academia. The basic idea is to study questions from the participants as a foundation for negotiations in the circle [24].In line with current discourses in participatory practice, we understand patients as actors with resources to develop the services they use [8,9,12,25]. User participation means that patients and users of the services are involved in decision-making processes. We use the term collective participation when the purpose is to improve the collective good [22,26,27], improving services for everyone in the same situation. In the research circle project, participation has a collective goal of achieving improvement for all patients in the institution.Arnstein maintains that individual and collective participation requires participants to have power and influence decision-making [28]. Arnstein defines counselling or consultation as tokenism and emphasises that genuine participation must include a partnership with opportunities for negotiation and influence or delegated power or control [28].The term empowerment includes awareness of power relations and oppressive mechanisms and strategies for achieving power to improve living conditions. Empowerment of groups refers to raising awareness of a group’s situation and their opportunities for collective action, as in Freire’s understanding of the pedagogy of oppressed people [29]. As Gutiérrez [30] and Rappaport [31] described, empowerment can refer to processes and goals or the result itself. Collective participation and group empowerment may also result in individual empowerment because of enhanced awareness of one’s situation, resources, and values [19,30].The terms user-participation and empowerment are criticised for being tokenism and rhetoric. There is a risk in collaborations between researchers, professional practitioners and patients that empowerment and participation can reproduce oppressive practices if the collaboration does not involve reciprocity and the opportunity to influence results [6,28,32].In this article, we use the concepts of collective participation and empowerment to discuss if the patients’ involvement can be understood as genuine participation, as a partnership, or merely as tokenism without power or influence.Recognition and respectful communication are central themes in the research circle model. The concept of recognition involves having the ability to place oneself in the other’s shoes, to understand, respect, confirm, and to have emotional availability [33]. The relation is dynamic in the interaction between parties, in this context, between patients and professional practitioners, aiming to bring about the involvement of others. In collective interaction, we are seen and recognised as subjects, contributing to our awareness of ourselves. Recognition is about understanding and valuing the other’s experience to develop mutual understanding and not exercising power over others [33]. The ability to see oneself, see the other, and reflect on actions, provides the strength to draw out something in the other and negotiate [33] (p. 149). When equal parties with different experiences and knowledge talk together, different perspectives challenge and contribute to development.The culture in a treatment institution has unique characteristics and frameworks. Based on a study of psychiatric hospitals, Løchen [34] (p. 122) described a controlling and restrictive institutional culture with norms and structures that convey no interest in the patients as human beings or room for constructive relations between patients and staff. Løchen [34] refers to two different treatment ideals: Institutions with a controlling and restrictive culture and institutions with a close and interpersonal culture. He points out that treatment institutions usually are organised as hierarchies. First, power is held by doctors and management, and then by personnel in the milieu, patients are at the bottom of the hierarchy. Research on dropout from substance use treatment institutions emphasises the importance of institutional culture. McKellar et al. [35] and Brorson et al. [11] found that patients had a lower risk of dropout from environments they perceived as supportive than from environments they experienced as rigid and controlling. Brorson et al. concluded that therapists should involve patients in decisions about their treatment and establish a good alliance with them [11] (p. 1020).The concepts of recognition and institutional culture are central elements in discussing patients’ experiences of the research circle and the opportunity to implement changes in the institution.In this article, we explore the opportunities and challenges for collective participation, drawing on the experiences from the research circle project. We will discuss the following research questions: How and in what areas did patients have opportunity to participate collectively in the treatment institution? How did the patients experience participation in the research circle?The participants in the research group agreed on participatory action research (PAR) as a research approach. In PAR, knowledge production combines research activities and action for change [36,37,38]. As mentioned, we chose to use a research circle as a model for trying out collective participation in cooperation between patients and staff from a treatment institution for people with substance use problems and researchers from the Western Norway University college (HVL).A research circle is an arena in which practitioners, service users and researchers meet to immerse themselves in a problem of common interest and contribute to change [24]. The research circle project lasted for 1.5 years, from 2013 to 2015, allowing for dialogue and continuity. A total of 16 people participated in the research circle. Six institution staff, four male nurses, one female pedagog and one female psychologist participated throughout the study period. Ten patients at the treatment institution participated in the project. Due to the absence of leave, short stays at the institution, and death, the number of patients varied. The patients who participated were men aged 24–50 who had 5–37 years of experience as substance use patients. Four female researchers from HVL participated, one nurse and three social workers, including the authors.The participants in the research circle met on three occasions at two-day seminars at a hotel, including an overnight stay. The start-up seminar was followed by an exploration phase where we explored the problem areas and topics that the project could examine. After the second seminar, in the action phase, we experimented with changes at the institution. Each phase consisted of six half-day meetings at the institution.In the start-up seminar, staff, patients, and researchers were actively involved in the process of arriving at topics for the project. The group chose topics from the patients’ narratives of their daily lives outside the institution and from the patients’ and staff’s experiences inside the institution. In addition, the researchers introduced relevant theoretical perspectives into the discussion perspectives of collective participation, empowerment, and recognition. The selected theme for the exploration phase was: How can we strengthen milieu therapy through user participation? Staying together at the hotel, away from day-to-day life in the institution, had created an informal atmosphere, and we got to know each other in new ways. The atmosphere and commitment of the start-up seminar continued in the following seminars and meetings.In the mid-term seminar, members agreed on themes related to the quality of the milieu in the institution and to develop necessary skills for life after treatment. We agreed on the following vision for the action phase: to create a treatment model to improve the milieu-therapy inside the institution and learn skills that would strengthen the patients’ ability to master life outside the institution after treatment. Referencing Simon and Garfunkel, the patients called this “Bridge over troubled waters” [39].In the action phase, we attempted to implement changes at the institution. To anchor the work in the treatment institution and HVL, we made information about the proceedings in the research circle available on the participants’ networks, and we discussed proposals for changes to patients and staff in the rehabilitation department.In the final seminar, we analysed the work of the research circle and discussed how the institution could continue with patient participation.The data consist of comprehensive and detailed minutes from three seminars and 12 meetings in the exploration and action phases. In addition, the researchers at the final seminar conducted two focus group interviews with patients and staff (separately). Questions posed in the focus groups were related to evaluating the process and content of the research circle and the attempts to make changes in the institution. Questions were also related to patients’ and staff’s motives for participation and their experiences in the research circle. The researchers wrote minutes from the seminars and meetings and transcribed audio recordings from the focus group interviews.We clarified with The Norwegian Centre for Research Data (NSD) that we were not required to register the project with NSD. All participants signed a consent form that includes their approval to participate and accept their mutual duty to maintain personal information confidentiality. We have anonymised all personal information and have not published the institution’s name, even though the institution initially wanted to include the name.In PAR, all actors ideally are involved in knowledge production through participating in change processes and dialogue [40]. The participants in the research circle were involved in designing the project and developing the topics and research questions. They were actively involved in the change processes and analysis in the final seminar. The initial idea was that patients and staff also should be co-researchers throughout the analysis and dissemination processes. However, patients and staff did not want to participate in the final analysis or co-author the article. The interpretation and understandings expressed in the paper are, therefore, the authors.The participants started developing the research questions inductively at the first seminar. We carried out the first analysis when the participants discussed and approved the minutes from seminars and meetings. This round of analysis constitutes a participant validation of the material [41,42]. The questions in the exploration phase centred around user participation. New questions and themes were added and explored through discussions during the meetings in the research circle. At the final seminar, the group analysed the material from the research circle. We concluded that the central theme in the work of the research circle had been to make changes in the institutional milieu to strengthen the mastery of life outside, in the participants’ words, to create a Bridge over troubled waters.For this article, the authors applied a thematic analysis [43,44] of the two data sets: written minutes from meetings and seminars and transcribed recordings of focus group interviews with patients and staff. We followed the steps in a thematic analysis outlined by Braun and Clarke [43] (pp. 87–93). First, we reread the material, systematically coding exciting topics across the data sets. We then searched for potential themes and connected data relevant to each potential theme. Reviewing the themes, we defined and named themes and subthemes. For producing the report, we identified quotations as examples for the themes and subthemes. The final part of the analysis was also theoretically informed. We analysed the themes using the theoretical frames of collective participation, empowerment, and recognition. We identified that one central theme is opportunities for making changes in the treatment practice through collective participation, and we identified three subthemes: 1. Mastering practical skills, 2. Recognition and respectful communication, 3. Trust and cooperation. Another central theme is the patients’ experiences as participants in the research circle, sub-themes 1. Recognition and 2. Dialogue.The patients believed that milieu therapy must include learning skills to master daily life when they leave the institution. “Freedom is threatening”, the patients said. They pointed out that they had to learn many skills to master life outside the institution, one said:“I am not ready for all the situations I end up in when I come out. I am getting no help here to master the life out there. How could we train for the outside in here: tidy your room, cooking, teach me how to behave in a job interview, enjoy my own company, how to behave when I want to rent somewhere to live when taking a phone call? Instruction in everyday tasks.”Crucial areas were cooking and tidying their rooms. The patients wanted to take responsibility for the kitchen service at the institution, buy the food, cook it, lay the table, and clean up afterwards. The patients thought that the kitchen should be an area in which patients and staff cooperated. One said: “Cooking, creating a sense of community around the table is important. It is crazy that a table is hardly laid for breakfast and lunch”.The patients said it was challenging to have responsibility for specific tasks like tidying their rooms and cooking. They wanted to be challenged to take collective responsibility and to receive individual feedback on their performance. They thought it would increase their self-respect and that they would become more aware of their weaknesses and strengths.At the up-start seminar in the research circle, it emerged that recognition and respectful communication are essential elements in achieving personal mastery, security, and trust. The patients’ experience from the research circle was that everyone listened to each other irrespective of agreeing or disagreeing. As a result, patients and staff felt it was essential to transfer respectful communication to the environment in the institution.The patients explained that they lived in a hierarchy outside the institution’s environment but continued living in a hierarchy in the treatment system—again in the lowest caste. They were concerned about the lack of respectful communication and participation. The participants agreed that respectful communication should be a goal but that it should at the same time be direct and confrontational.The staff experienced the institution’s culture as more of control than trust and saw a need to change the interaction between staff and patients. The discussion about control and trust took as an example the institution’s procedure of requiring patients suspected of using substances to submit a urine sample. The staff thought it was better to be open and confront the patient directly with suspicion. The patients answered that they needed openness and directness, one saying: “We want, and we tolerate direct and challenging communication”. They emphasised that the staff should give honest and constructive feedback: “We must be able to correct each other and still maintain our relationships”.The patients believed that community and good relationships between patients and staff were vital in building and maintaining trust and cooperation. They also emphasised that patients are essential to each other in the institutional environment. They understand each other and know what can trigger unease or promote a feeling of security. Therefore, the patients in the research circle established a regular meeting for all patients every Sunday to promote a sense of community in the institutional environment. The purpose of these meetings was to share ideas about activities in the research circle, receive input from other patients, and implement proposals for change in the environment. In addition, the patients set up an ‘activity bank’ in the department to summarise how the week had gone, determine what they could do differently, and make plans for the next week.The patients felt insecure in social settings; they were not used to talking about anything other than substances and were poorly informed about current affairs. However, one patient said: “Knowing a little about what has been in the news can be useful when trying to find topics to talk about with others”. Therefore, patients and staff agreed to plan and arrange a quiz once a week. The intention was to increase everyday knowledge, promote social engagement, and create a social community among the patients and staff. One said a quiz would broaden their horizons: “It will make us open the windows a little and let the outside in.” To arrange and participate in a quiz would challenge the patients to work together and to focus and remember.In the project period, patients and staff tried out quizzes, cooking projects, and interviews when there was suspicion of intoxication. Patients and staff agreed that these changes were good examples of participation and active milieu work and could increase patients’ ability to master life outside. These changes were, however, not permanently established as practices at the institution.The patients said they soon felt that everyone recognised each other in the research circle. Being recognised and perceived as equals strengthened their self-confidence, leading to personal growth and an experience of mastery. One said:Easy to take part in equal. It was like this right from the start: managers, staff, and patients. Everyone talking, everyone respected. No matter who. A fantastic way to work. When we enter treatment, we do not expect to sit talking to the management and top academics. We have gained great self-respect from this.The patients thought the different perspectives contributed to the good dialogue. One said: “Many different people sitting together and chatting—a lot of good comes out of it!” The research circle allowed the staff to see that the patients have the resources and strength to contribute and that drawing out the patients’ perspectives is essential. “What the patients bring up here, their experiences, gives me new knowledge, experience-based knowledge that I can process and use in other settings”.Several patients said they had not previously dared to speak in groups, but participation in the research circle had changed this. The patients emphasised that they could be open and honest with other patients, they spoke the same ‘language’, and that they in the research circle experienced the same form of dialogue with the staff and researchers. They, however, reflected upon that cooperation with the staff at the institution was often based upon insecurity and misunderstandings.All patients, staff, and researchers commented that the interaction in the research circle helped remove the barriers between participants. For example, one patient described their interaction with the staff members in the research circle as being: “A fantastic way to break down barriers. That does not mean we respect them less. No one will lose their role because they have breached these boundaries”. The staff said that the close interaction in the research circle was a new experience; it had been helpful in the research circle to get close to the patients, hear their experiences and perceptions, and collaborate with them.The patients thought the conversation form in the research circle could be helpful in treatment. Daring to express oneself and getting good feedback afterwards gives a feeling of mastery. It is, however, not problem-free. As one said: “It can, at the same time, be more difficult because you are naked without intoxication, have more shame, which can make it more difficult to deal with”. All participants, patients, staff, and researchers experienced the research circle as a new experience of individual and collective participation, contributing ideas, and reflections.Did the patients have opportunities for collective participation in the treatment institution? What promoted and what hampered participation opportunities?The patients’ stories represented the starting point for choosing the themes that led to attempts for change in the institution. Their stories told about their experiences in society after previous treatment programs: their difficulties in mastering the transition to life without substances, a lack of practical day-to-day life skills, a lack of social networks and problems communicating with others. The patients had experienced that the milieu therapy provided by the institution did not help strengthen the skills they needed to master everyday life. The patients’ descriptions align with other research, showing that substance use treatment institutions do not plan well for discharge and transition from institutions to life outside [3,6,10].The patients’ stories were the starting point for discussions on changing milieu therapy in the action phase. Proposal formation was, however, a joint project in which all participants actively participated.Several change proposals were implemented in the institution to create a more supportive environment. One implemented change was a less controlling procedure when the staff suspected patients of using substances of having a dialogue with the patient instead of demanding a urine sample. The institution staff became aware, during the project, of the challenges of their role as guards, exercising control, and at the same time attempting to create good relationships and build trust.Other changes were intended to improve the sense of community and social training in the institution. Patients got responsibility for the activity bank and the Sunday evening patient meetings. Patients and staff worked together on activities, arranged the quiz, and became more present in the department. The changes provided patients with communication and interaction training during the execution of these tasks. The changes were intended to achieve closer cooperation in the patient group and between patients and staff. Delegating responsibility for tasks and cooperating in their execution positively influenced the culture of the environment and promoted personal development. Other research also emphasises the potential for learning and personal development in institution environments [11]. Therapeutic communities can create cohesion and cooperation and achieve improvement process goals [13]. Seeing each other, respecting different opinions, and recognising others as significant participants in the system are essential aspects of the change process [33]. Eide and Nesvåg pointed out that the content of the milieu therapy is essential for how patients experience day-to-day life after treatment [15].We can sum up that the exchange of experiences and dialogue in the research circle resulted in constructive proposals for change. Collective and individual participation requires the system and helpers to share power and create a culture of trust and mutual respect: To treat patients as subjects and have faith in their resources and competence to solve problems.The participants in the research circle experienced challenges when trying to implement lasting changes in the institution. The patients’ wish to learn about cooking, which the staff supported, was limited by organisational issues. It was difficult to reconcile to give patients responsibility for cooking and cleaning with the current management firm. The institution used a centralised food supply, which made patients shopping for and cooking food difficult. Making patients responsible for cleaning was challenging to combine because cleaning staff were employed to do this job. Implementing the research circle’s improvement suggestions proved challenging to achieve in practice. The patients felt that that participatory practice had a long way to go and that they still were at the bottom of the treatment institution hierarchy. Their feelings agree with Løchen’s description of a controlling and restrictive institutional culture [34].We conclude that the patients’ involvement in collective participation, only to a small extent, contributed to changes in the institution’s treatment practices. Viewed as collective participation in the institution’s context, using Arnstein’s conceptual framework [28], we do not understand the patient’s participation as a partnership but as tokenism without power or influence. The challenges presented by the attempts at making changes in the institution are related to the institution’s culture, structure, and forms of governance [35], and the inherent power relations [28] in the institution. In afterthought, we can ask if these challenges may also relate to that we had not sufficiently anchored the research project in the institution.The good relations between the participants in the research circle, based on recognition and equality, and the close and respectful communication that grew out of this, contributed to an open and constructive discussion of complicated issues. There was also space in the research circle for confrontation, criticism, honest feedback, humour, informal chat, and good stories. The relations contributed to the participants feeling that the boundaries between role and position dissolved. Other research emphasises that patients feel more recognised and valued when professionals show a more personal side [7,13,35]. Changed relations are typical experiences where users, staff, and researchers meet with a common agenda of participation [7,19,45]. The place for meetings is also essential. New surroundings outside the usual treatment setting contributed to greater openness and access to participants’ resources and knowledge [8,46,47].The sense of community and dialogue in the research circle was a distinctive strength that promoted sharing experiences. Mutual recognition and trust meant that communication expanded perspectives and provided new knowledge. The lack of power struggle is consistent with the findings of appreciative communication research [33]. The participants were open to each other’s stories and experiences. They experienced dialogical cooperation and a mutual understanding of the research circle’s experiences, perspectives, and perceptions. Having time and respect to understand the other as they experience themselves became an essential part of the collaboration. The experiences correspond with findings in the research by Hansen et al. [48] that the inner understanding in respectful communication creates the opportunity to develop shared meaning and prevent misunderstandings, providing action competence. The experiences also concur with research into interaction and respectful communication, showing that recognition, cooperation as equals, and participation can provide an impetus for change and development [7,13,33,49,50].The interaction between patients, staff, and researchers in the research circle contributed to discovering new aspects of themselves and others. The focus was listening to others, changing language, and expressing feelings without hostile reactions from the other participants. The patients felt that they had to become more aware of communicating their views and opinions in the institutional environment. They wanted to transfer their experiences from interaction in the research circle to everyday life in the institutional environment.We can summarise that the patients experienced participation in the research circle. They experienced being heard and taken seriously. Their stories and experiences formed the starting point of deciding which themes and issues would become central in the exploration phase. In the action phase, all participants influenced the proposals for change in milieu therapy. We understand the interaction between patients, staff, and researchers in the context of the research circle as a partnership where the parties involved can negotiate and influence the results [28].The patients described their experiences in the research circle as empowering, in the way that Freire [29], Gutiérrez [30], and Rappaport [31] have described the processes of empowerment. The patients experienced that their stories about substance use treatment and their needs after treatment led to proposals for changes in the institutional environment. They said that participating in the research circle contributed to increased self-confidence in raising awareness about their situation. The patients’ motives also related to the fact that the research circle should collectively lead to improvements for others: “This is the driving force behind being here. That commitment, a wish to participate goes onwards and beyond ourselves”. They also stated that this gave them hope for the future, that their efforts could help improve their situation and the situation of other patients in treatment.One limitation of the study is that the patients did not participate in the final analysis and preparation of the article. However, after the project, when two patients and three staff, participated in the dissemination of Further and master’s education in HVL, we checked that our understanding concurs with their understanding. Another limitation is that the research did not include information on implementing changes in the institution over time. Further research should include this as part of the evaluation of results.We can conclude that the activities in the research circle led to some changes at the institution: Some improvements were implemented in the milieu therapy, and the institutional culture became based more on participation and equality. In addition, the patients experienced partnership and empowerment in the research circle.To implement proposals that could lead to lasting changes in the organisation and work structures was more challenging. The organisation of the environment and treatment in the institution appeared to be fragmented and divided into different roles, different tasks, and interpretations. For milieu therapy to develop to provide the strength, substance users need to master life outside the institution, all participants in the system, patients, and staff, must work together to change attitudes and working methods. The institutional culture must change from power and control to a humane culture of closeness and participatory practice, as mentioned in other studies [7,11,35]. The previous study from the institution, Larsen and Sagvaag, described the institution as being dominated by a diagnostic culture that hindered participation [6]. They found that staff and managers believed that management permitted empowerment if they found this to be expedient, even though they believed patient participation was valuable.We conclude that there is a need to change attitudes towards patients with substance use problems, to view them as people who possess competence and resources. It is about time for a paradigm shift in policies and practice in the treatment of substance use problems and the education of the relevant professions. We emphasise three essential prerequisites for making fundamental changes in attitudes, policies, and treatment practices. Firstly, that patients are involved both in individual and collective participation. Secondly, to meet in arenas outside the institution, in places that provide space for an open dialogue between patients and staff. Thirdly, developmental and research projects must be well rooted in the institution’s management to ensure the implementation of suggestions for changes in the organisation, work structures and treatment practices. Finally, we recommend conducting research and development projects in collaboration between the practice fields, patients and professional practitioners, and research and professional education in academia.We end by quoting one of the patients: “If this can be done, be systematised, disseminated, and benefit others, then this is good. After so many years of rising and fall, this is what one hopes for”.Both authors have been involved in conceptualization, methodology, validation, analysis, writing, original draft preparation, review and editing. All authors have read and agreed to the published version of the manuscript.The research has not received special funding.Ethical review and approval from HVL were waived for this study, due to clarification with The Norwegian Centre for Research Data (NSD) that we were not required to register the project with NSD.All participants signed a consent form that includes their approval to participate and accept their mutual duty to maintain personal information confidentiality.The data are not publicly available due to ethical restrictions.We want to thank patients and staff at the substance use institution, Svanhildur Gudmundsdottir and Mariann Vigdal, for participation in the project.The authors declare no conflict of interest.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11061.txt

@@ -0,0 +1 @@
+Although digital media usage is prevalent among middle school students, the safety of digital media-based learning activities for students at risk of digital media addiction is unknown. The goal of this study was to evaluate the differences in students’ brain activity in relation to their risk of digital media addiction. The study was quasi-experimental, with a pre- to post-test control group design. The study participants included 83 middle school students who were engaged in digital learning. We measured their brainwaves to evaluate brain activity using a PolyG-I (LAXTHA Inc.). We found no statistically significant differences in the location of the attention index between the two groups before and after digital learning. However, there were statistically significant differences between the two groups in the P3, P4, and F4 locations of the relaxation index. These results indicate that students at risk of digital media addiction may experience learning difficulties. These results can be used to guide healthcare professionals in developing digital learning programs that are safe for students and to also verify the effects of these programs.New technologies have reshaped learning and education. Digital learning is widespread, facilitated by devices such as personal computers, laptops, and smartphones in school and at home [1]. Although the benefits of digital learning are appealing in terms of achieving learning goals, the guardians and teachers of students participating in digital learning are concerned about the potential side effects on these adolescents’ mental health because of the increase in device usage time [2,3]. In particular, there is increasing concern that the increase in device usage time will lead to digital dependence or addiction. Concerns related to the potential for health problems related to digital learning include physical health problems, such as musculoskeletal disorders and headache, and psychological health problems, such as depression, stress, and concentration disorders, which may emerge from early childhood onwards [4,5,6,7,8,9]. On the other hand, research on the effect of increasing digital device usage time on the health of adolescents is insufficient.Previous studies have reported that the use of digital devices can cause not only physical but also psychological symptoms. Student health problems related to the use of digital devices include physical symptoms such as eye health and musculoskeletal health [10,11]. Other studies have shown that students at risk of digital media addiction are most aware of their psychological symptoms vs. other types of symptoms [12]. Digital media addiction was recently bringing about symptoms in children and adolescents; it was excessive usage of digital devices such as tablets, smartphones, etc. [13]. Among the types of digital media addiction, cell phone addiction is related to adolescents’ mental health and physical health [14]. The act of using a digital device, watching television for example, has effects similar to those of substance abuse and increases children’s and adolescents’ passivity about receiving the attention they need [15].In addition, previous studies have shown that the increasing use of digital devices has raised concerns regarding physical and psychological health problems, and even quality of life [16]. In addition, this increase may lead to digital media addiction. For instance, people who use the internet more frequently have an increased risk of poor physical health [17]. Smartphone use can also have harmful effects on both the interpersonal relationships and the psychological health of adolescents [18]. Because the type of digital device used and the time spent on the device can affect adolescents’ health, their risk of digital media addiction and the potential implications for their health must be considered. Students may experience attention and relaxation difficulties when they use digital devices, which can be assessed by measuring the levels of brain activity when learning how to use digital devices. There are different opinions regarding digital learning and its effects; thus, it is necessary to measure brain activity using electroencephalography (EEG) while students are learning how to use digital devices to help clarify these controversies.EEG is a useful tool for measuring learning effects while students are learning in real time [19,20], as well as for measuring their levels of attention and relaxation [21]. EEG is a non-invasive, inexpensive, safe, and easily operated method frequently used to examine learning effects through brain activity in children and adolescents [22,23]. Previous studies have used EEG to measure attention and relaxation in children and adolescents with attention deficit hyperactivity disorder (ADHD) [24]. For assessing attention, a previous study measured the ratio of the sensory motor rhythm (SMR) ~mid-β waves to theta waves (RSMT) [25,26]. In addition, relaxation was measured using the ratio of α to high-β waves (RAHB) [26,27]. EEG is a suitable tool for identifying attentive and relaxed states in adolescent students’ brains during the learning process.Herein, we compared the brain activity and the levels of attention and relaxation while using a digital device for learning between students at different risk levels for digital media addiction. The risk levels for digital media addiction were assessed in the survey, and then, data on students’ brainwaves were measured by EEG.The aim of this study was to compare the brain activity and the levels of attention and relaxation while using a digital device for learning between students at risk of digital media addiction (risk group (RG)) and students not at risk (without-risk group (WRG)).This study was a case–control study comparing brain activity and levels of attention and relaxation between RG and WRG.A convenience sampling approach was used to generate the study population. A total of 83 middle school students (54 students with no risk and 29 students at risk of digital media addiction) in Korea were selected for the study. We first targeted the middle schools and then obtained the principals’ consent for the recruitment of students for this study. The inclusion criteria for participants were: (1) students with prior learning experience using a digital device such as a computer, smartphone, or television for more than one year; and (2) students who qualified for an EEG due to the absence of a cerebral disease and who did not drink coffee or soda and had not used any haircare substances within two hours prior to the procedure.The sample size was calculated using G Power 3.1.3, which was also used to determine that this study required a Wilcoxon–Mann–Whitney test (two groups) analysis with an effect size of 0.62, a power of 0.8, and a significance level of 0.05, which resulted in the requirement for a total sample size of at least 78 students (group 1 = 26, group 2 = 52). The effect size of 0.62 was chosen based on a previous research study on the baseline of neurofeedback among ADHD children [28]. In the current study, the sample size requirement was met as 83 students were enrolled. When the effect size was calculated with the number of subjects that actually participated in the study, the number of subjects was considered appropriate because the calculated effect size of 0.61 was similar to that assumed before the study was conducted.Data were collected using two methods: survey and brainwave measurement. First, data on the student characteristics and levels of digital media addiction were collected via self-report. Students were asked to complete a questionnaire on their experiences with digital learning, the time they had previously spent participating in digital learning, and their health status, and were also asked to complete the Diagnosing Smartphone Addiction [29] and Diagnosing Online Game Addiction survey [30].After the self-report survey, data on students’ brainwaves were measured using the PolyG-I computerized EEG meter (LAXTHA Inc., Daejeon, Korea). Researchers trained in EEG examination performed the measurements in experimental rooms in the schools between 15 November and 7 December 2014. During the brainwave measurements, the students sat in a room with minimal noise with the electrodes attached to their scalps. They maintained a stable state with their eyes closed during the measurements. The electrodes were attached to eight sites on the scalp as per a previous study and in accordance with the International 10/20 System of Electrode Placement [31]. The measurements consisted of five steps. In the first step, brainwaves of students in the stable state were measured for 6 min. During this step, the students maintained a stable state without moving their bodies with their eyes open. This was followed by a 1 min rest period, constituting the second step; when continuous measurements were made without a rest time, it was difficult to separate the exact interval for stimulation and response, which may create difficulties in data analysis, so the 1 min rest period was used to avoid this scenario [32]. In the third step, the students had 6 min of learning time using a digital media device, followed by a 1 min rest period as the fourth step. During the learning time, the students studied using digital textbooks, which included audiovisual images on a digital media device such as a laptop computer. The content comprised social science subjects. The students solved pop-up quizzes, but these results were not used for analysis in this study. In order to focus on the effect of the learning activity itself on brainwaves, we tried to minimize the stress and learning gap caused by pop-up quizzes for students. In the final step, the students’ brainwaves were measured in the stable state for 6 min, the same as the first step. For brainwave data analysis, we used 4 min of brainwaves out of the 6 min measured before learning. The 1 min after the measurement started and 1 min before the end of the measurement were excluded from analysis. In addition, 4 min of brainwaves was used out of the 6 min of brainwaves measured after learning. As with before learning, the 1 min after the measurement started and the 1 min before the end of the measurement were excluded from the analysis. In this way, we were able to obtain reliable data by preventing data contamination due to the inflow of artifact waves that may occur during the movement from one step to another.The participants completed a questionnaire about their characteristics, including information about their sex (51.8% male, 48.1% female), age (mean 14.69 y), digital learning status, and health status. To assess digital learning status, information regarding past experiences with digital learning, time spent participating in digital learning, and the types of digital devices used was collected. To assess health status, the students’ health status and feelings of fatigue or pain when using digital devices were investigated.The definition of digital media addiction included the overuse of smartphones and digital devices to watch television, search the internet, or play online games [33]. We defined the risk of digital media addiction as belonging to one of the potential risk groups for smartphone or online game addiction according to previous research [12]. The students were assigned to the WRG if they were not included in either the smartphone or the online game addiction groups. Students were assigned to RG if either their score for Diagnosing Smartphone Addiction (DSA) was higher than 46/60 points or their score on the Diagnosing Online Game Addiction Scale (DOGA) was higher than 46/80 points.For the allocation of participants into the groups, the risk of digital media addiction was measured using the DSA and DOGA from the National Information Society Agency of Korea (NIA) [29,30]. These instruments were developed by the NIA to consistently measure digital media addiction through self-assessment and to provide appropriate interventions at the national level.The DSA scale consists of 15 items, with a response scale ranging from 1 (strongly disagree) to 4 (strongly agree). The NIA guidelines define general users and potential at-risk users in smartphone addiction groups as those who obtain scores of less than 35 points and 36–45 points, respectively [29]. Cronbach’s α for this scale in this study was 0.86.The DOGA scale consists of 20 items, with a response scale ranging from 1 (strongly disagree) to 4 (strongly agree). The NIA guidelines define general users and potential at-risk users in online game addiction groups as those who obtain scores of less than 35 points and 36–45 points, respectively [30]. Cronbach’s α for this scale in this study was 0.91.Brainwaves were measured and the level of attention was analyzed using the ratio of SMR ~mid-β waves to theta waves (RSMT) using the following formula: level of attention = (power ratio of SMR + M-β)/θ [25]. The sensory motor rhythm (SMR) and M-β wave values reflect attention: participants who had attention deficits were also shown to have a low level of attention. The level of relaxation was analyzed using the ratio of α to high-β waves (RAHB) using the following formula: level of relaxation = (power ratio of α)/(H-β) [25]. The H-β wave showed the participants’ stress level, while the α wave showed the participants’ relaxation level. These two indicators were relative values ranging from 0 to 1. A previous study used these indicators to determine the effects of e-learning in university students [20].Data analysis was conducted using STATA version 16.0 (STATA Corp LP, College Station, TX, USA) as follows: first, the characteristics of the two groups were analyzed using descriptive statistics; categorical variables were expressed as numbers and percentages and continuous variables as mean ± SD. The differences between groups were evaluated using χ2 tests, whereas Wilcoxon rank-sum test was used to compare the differences in brainwave activity between the two groups. Statistical significance was considered in the case of p < 0.05.Ethical approval was obtained from the University Institutional Review Board (Seoul, Korea). After receiving approval, we obtained the consent of the middle schools’ principals to recruit participants for this study. We then sent a letter requesting participant consent and an explanation of the study to the students’ guardians. We received consent forms signed by the students and their guardians before the study was conducted. The students voluntarily participated in this study and provided written informed consent. The consent form stated that the participants could withdraw from the study at any time and that their information would only be used for the present study.The characteristics of the 83 students are listed in Table 1. Tests for homogeneity between the RG and WRG showed that the groups were homogenous for all variables. The WRG consisted of 54 students (28.9% male, 36.1% female), and the RG contained 29 students (22.9% male, 12.0% female). More than half of the students had no prior experience with digital learning (36.1% of the WRG, 19.3% of the RG). Over 50% of the students had experienced fatigue or pain during digital device use (34.9% of the WRG, 20.5% of the RG).Table 2 and Table 3 show the differences in the levels of attention and relaxation for each electrode position and group before and after digital learning. Table 2 shows the levels of attention of the two groups. Before digital learning, none of the electrode locations differed significantly between the two groups. Similar results were obtained after digital learning. The variances between the measures taken before and after digital learning did not differ significantly between the groups.Table 3 shows the differences between before and after digital learning in the level of attention and relaxation between the two groups. Among the variances between before and after digital learning, a statistically significant difference at location F3 (left frontal lobe; t = −2.16, p = 0.031) and F4 (right frontal lobe; t = −2.39, p = 0.017) was observed in the relaxation index between WRG and RG.In the current study, no differences in the attention levels between RG and WRG were found. Although the RG showed lower levels of attention than the WRG, the difference was not statistically significant. However, the RG exhibited significantly different levels of relaxation in the parietal lobes before and after digital learning. In future research, it will be necessary to plan a research design that also considers time.Attention is defined as the degree to which thoughts are gathered [34], which is essential for learning. In the present study, we found no statistically significant difference in attention between the two groups. However, previous studies have reported contradictory results. One study found that digital learning did not affect the attention of learners based on their brain activity [35], whereas another study reported digital learning to be more effective at encouraging students to concentrate while studying [36]. Due to the results that digital learning has a positive effect, such as increasing memory, and the results of conflicting studies on the negative effects, such as increased attention-deficit symptoms, the results of the effect of digital learning on brain health are controversial [37]. Since this is the same not only for adolescents but also for adults, further investigation is needed to identify the mechanism by which digital learning affects brain health and to determine the effect of digital learning on attention. In addition, since we compared WRG to RG, further research into the addiction group is necessary to measure the differences in attention levels.The comparison of the measurements obtained before and after digital learning showed that the levels of relaxation at the F3 (left frontal lobe) and F4 (right frontal lobe) location differed significantly between the two groups. After digital learning, the WRG showed lower relaxation levels, whereas the RG showed higher relaxation levels. Relaxation is defined as an indication of a comfortable state [34]. For levels of relaxation to be elevated, either H-β waves are inactivated when emotionally unstable, or α waves are activated when the brain is resting [17]. This means that an excessive increase in the levels of relaxation can desensitize brain activity, so it can increase the risk of learning difficulty. As the frontal lobe is associated with intelligence, concentration, and short- and long-term memory [38], this result can be interpreted as the RG being at risk of learning difficulties. Therefore, healthcare professionals need to identify students at increased risk of digital addiction and supervise them more closely to prevent the development of learning disabilities.There are some limitations to this study. First, we targeted students in one country, so the results may not be generalizable to other regions and countries. For generalizing the results to other countries, the results must be confirmed through repeated studies. In addition, brainwaves are constantly changing, so it is necessary to diversify measurement methods, such as increasing the measurement time or measuring participants several times. Education-related factors must also be considered, such as educational achievements and previous knowledge. Third, in this study, the usage time for digital devices for the WRG was greater than that for the RG. This is because learning time, as well as leisure activities, was considered to contribute to digital device usage. In further research, groups that spend a large amount of time using digital devices must be compared, such as those that use digital devices for leisure activities such as games and play with those that use them for learning. Finally, it was difficult to identify the determinants of brainwave variation because this study focused on the comparison of WRG and RG. In future research, identifying the determinants of brainwave variation may be helpful for adolescents’ learning.The results of the present study raised the necessity for further research on the relationship between digital media addiction and digital learning. These results were in close agreement with the results of previous studies, which reported that excessive media use can reduce students’ academic performance via increased behavioral problems in school [39,40]. However, since the use of digital devices has both advantages and disadvantages, the results need to be carefully analyzed. A previous study showed that the use of digital technology is controversial depending on the purpose and methods of usage [37]. As school health providers’ roles and responsibilities include supporting students’ health and advocating for prevention strategies to avoid harmful environments [41], healthcare professionals should be able to assess risk early and to systematically manage students. This management requires the development of interventions, such as digital learning programs, to provide advice to students and guide them. For this, school health providers must improve their competency in the use of neurophysical techniques such as brain activity analyses to educate patients and students [42].In this study, we compared the differences in the levels of attention and relaxation during digital learning according to the risk of digital media addiction. We identified students at risk of digital media addiction, as well as those not at risk, and compared their levels of attention and relaxation when they learned using digital devices. The results of this study suggest that digital devices alter adolescents’ brain activity, providing information for healthcare professionals on how to guide adolescents in using digital learning methods. Through the results of this study, we were able to ascertain the necessity of further research related to risk of digital media addiction and learning difficulties. Further research is necessary to clarify the relationship between digital media addiction and learning effects. Based on the results of our study, healthcare professionals can develop programs for the safe use of digital learning among students and verify the effects of these programs.Project administration, G.S. and W.N.; funding acquisition, G.S.; conceptualization, G.S. and W.N.; methodology, G.S. and W.N.; software, W.N.; formal analysis, G.S. and W.N; investigation, G.S. and W.N.; resources, G.S. and W.N.; data analysis, G.S. and W.N.; data curation, G.S. and W.N.; writing—original draft preparation, G.S. and W.N.; writing—review and editing, G.S. and W.N.; visualization, W.N.; supervision, G.S. and W.N. All authors wrote the first draft and conducted critical discussion. All authors have read and agreed to the published version of the manuscript.This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Science, ICT and Future Planning (NRF- 2013R1A1A3013229).The study was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Korea University (IRB no. 1040548-KU-IRB-14-64-A-1 and 1040548-KU-IRB-14-176-A-1).Informed consent was obtained from all subjects involved in the study.The data that support the findings of this study are available from the corresponding author upon reasonable request.The authors have no conflict of interest to declare.Participants’ general characteristics and homogeneity between without-risk and risk groups (N = 83).Differences in attention between the without-risk and risk groups (WRG and RG, respectively) (N = 83).The differences between before and after digital learning in the level of attention and relaxation between the risk and without-risk groups.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11062.txt

@@ -0,0 +1 @@
+Two authors have contributed equally to work.Oral diseases, such as periodontitis and dental caries, can cause systemic inflammation as well as local effects, which is an important contributing factor for obesity. We aimed to investigate the change in body mass index (BMI) according to the presence of periodontitis and oral hygiene indicators. This study enrolled 110,068 participants from the national health screening cohort in Korea from 2009–2010 who underwent an oral health checkup. As oral hygiene indicators, the presence of periodontitis, dental caries, tooth loss, and tooth brushing were assessed. We constructed a linear mixed model for BMI. BMI was repeatedly measured in the health examination until 2015. In the multivariate linear mixed model for BMI, the presence of periodontitis (β = 0.0687, standard error (SE) = 0.0225, p = 0.002), dental caries (β = 0.0735, SE = 0.0152, p < 0.001), and tooth loss (β = 0.1328, SE = 0.0175, p < 0.001) were positively associated with BMI. In contrast, frequent tooth brushing (≥3 times/day) was negatively associated with BMI (β = −0.2610, SE = 0.0306, p < 0.001). The presence of periodontitis, dental caries, and tooth loss may be associated with higher BMI, whereas frequent tooth brushing may be related to lower BMI. Better oral hygiene might be associated with lower BMI. Further study is needed to determine the effect of oral health behavior and dental disease on obesity.Recently, the prevalence of obesity has steadily increased, becoming a worldwide phenomenon [1]. Furthermore, obesity is an independent and strong predictor of cardiovascular diseases, even without the interaction of other risk factors [2]. Thus, because obesity is a modifiable risk factor for cardiovascular disease, several strategies to control overweight and obesity have been considered, such as low fat or proper carbohydrate diets, physical activities, lifestyle modification, and use of medications for weight reduction [3]. However, the methods for controlling overweight and obesity are limited in clinical practice.Periodontitis, dental caries, and tooth loss are common diseases reflecting poor oral conditions [4]. Periodontitis is defined as an inflammatory disease of the supporting tissues of teeth caused by specific microorganisms or groups of specific microorganisms, resulting in progressive destruction of the periodontal ligament and alveolar bone with periodontal pocket formation, gingival recession, or both [5]. Emerging evidence suggests that poor oral health and periodontitis can influence the initiation and/or progression of various diseases such as cancer, cardiovascular disease, metabolic disease, and degenerative diseases [6,7,8,9]. On the other hand, behaviors improving oral health may reduce the risk of medical illnesses [10,11]. One possible mechanism for this is that oral diseases not only induce local inflammatory effects that destroy the dentition and tooth-supporting tissues, but also cause systemic inflammation [4]. The enhanced inflammatory reaction is one of the main pathophysiologic mechanisms of obesity [12]. Dental caries are caused by a breakdown of the tooth enamel and extend to dentin and pulp tissues. This breakdown is the result of bacteria on teeth that break down foods and produce acid that destroys tooth microstructures and results in tooth decay. Dental caries or tooth loss may adversely affect nutrition intake, leading to malnutrition or obesity [13,14]. As complex chronic diseases, both periodontitis and dental caries have been indicated to share common risk factors that are inherited and acquired [15].We hypothesized that poor oral health could contribute to the development and progression of obesity. Poor oral health can be indicated by the presence of periodontitis, dental caries, tooth loss, and infrequent tooth brushing behaviors. However, research on the relationship of oral health indicators with obesity is limited, especially in longitudinal settings. Therefore, we investigated whether the oral health indicators including periodontitis, dental caries, tooth loss, and tooth brushing behaviors have associations with longitudinal values in body mass index (BMI), using a nation-wide population-based health screening cohort in Korea.Our study utilized the National Health Insurance Service-National Health Screening Cohort (NHIS-HEALS) dataset in the Republic of Korea [16]. The Republic of Korea provides a public health insurance system with a single-payer organization of NHIS. Every two years, the NHIS provides a complimentary nation-wide health checkup program to all Korean aged ≥ 40 years. The NHIS-HEALS is a sample cohort of participants who participated in the free health screening program, which is constructed to offer relevant and useful health data for political and academic research [16]. The cohort dataset included serial health checkup data of consecutive participants, including their BMI, blood and urine test results, oral health checkup questionnaire, oral health evaluated by dentists, blood pressure measurements, and lifestyle survey results. In addition, NHIS-HEALS contains personal health claims data for the hospital visit, diagnosis (using the International Statistical Classification of Diseases and Related Health Problems 10th revision (ICD-10)), prescription, and procedures. The health claim resources were available until 31 December 2015, the loss of health insurance eligibility, or death. More detailed information of NHIS-HEALS is available in a prior publication [16].This study enrolled participants from the NHIS-HEALS 2009–2010 dataset who had an oral health checkup as the baseline examination. We excluded participants with extreme BMI values (>32 kg/m2 or <17 kg/m2) and those who had missing data for at least one of the covariates in the NHIS-HEALS 2009–2010 dataset. A flow chart representing inclusion and exclusion criteria is shown in Figure 1. This study was approved by the Institutional Review Board of our institution (SEUMC 2020-08-18). The informed consent was waived due to the retrospective analysis based on the fully anonymized data.The presence of dental caries and tooth loss were investigated by dentists at the baseline oral health checkup (2009–2010). Regarding dental caries, dentists investigated whether the participants had caries of permanent teeth, and finally, those who had at least one caries of permanent teeth were classified as having dental caries. Tooth loss was evaluated regardless of the cause, and we regarded the presence of a fixed dental prosthesis, implant with an abutment, and a third molar as toothlessness [15]. The presence of periodontitis was defined as if a diagnosis of acute periodontitis or chronic periodontitis (K052–053, respectively) was made more than twice by a dentist or if the participants underwent treatment for periodontal disease (U1010, U1020, U1051-1052, U1071-1072, and U1081-1083 claim codes) with K052 or K053 diagnosis for 1 year before the baseline oral health checkup. The K052 code included acute periodontitis/periodontal abscess (parodontal abscess) of gingival origin without sinus/acute apical periodontitis of pulpal origin/periapical abscess/periapical abscess with sinus/periodontal abscess (parodontal abscess) of gingival origin with sinus/acute apical periodontitis of pulpal origin/periapical abscess/periapical abscess with sinus/acute pericoronitis/other specified acute periodontitis/acute periodontitis, unspecified. The K053 code included chronic periodontitis/chronic simplex periodontitis/chronic complex periodontitis/chronic pericoronitis/other specified chronic periodontitis/chronic periodontitis, unspecified [10,11]. The frequency of daily tooth brushing was categorized as ‘0–1 time/day’, ‘2 times/day’, and ‘≥3 times/day’, based on a self-reported questionnaire.We assessed data for sex, age, BMI, household income level (four quartiles), and presence of comorbidities (hypertension, diabetes mellitus, and chronic kidney disease) in the baseline health checkups. BMI was defined as one’s weight (kg) divided by the square of height (m2). BMI was serially measured in the health examination programs from the baseline health examination throughout the study period. Hypertension was defined when participants had health claims of blood pressure-lowering agents with ICD 10; I10–15, blood pressure ≥ 140/90 mmHg, or positive checking in a self-report questionnaire regarding hypertension. Diabetes mellitus was defined when participants had health claims of anti-diabetic drugs with diagnostic codes with ICD 10; E08–11, E13–14, fasting blood glucose >7.0 mmol/L, or positive checking in a self-report questionnaire for diabetes mellitus. Chronic kidney disease was determined by diagnostic codes of ICD 10; N18.1–N18.5 and N18.9, or an estimated glomerular filtration rate <60 mL/min/1.73 m2 [6]. From the repeated health checkups conducted during the follow-up, data for smoking status (current, former, never), alcohol consumption (frequency per week), physical activity (days per week), and laboratory findings (aspartate aminotransferase, alanine aminotransferase, and total cholesterol) were longitudinally collected [8,11,17].The comparison for categorical variables and continuous variables between the two groups were assessed using chi-square and independent t-tests, respectively. Differences between groups according to the frequency of daily tooth brushing were investigated using the chi-square test for trend or Spearman correlation analysis. To evaluate the variables related to BMI, we developed a linear mixed model with variables of fixed effects (oral health indicators, sex, age, household income, presence of comorbidities, smoking status, alcohol consumption, physical activity, aspartate aminotransferase, alanine aminotransferase, and total cholesterol). Statistical analyses were executed using R software, version 3.3.3 (R Foundation for Statistical Computing, Vienna, Austria), and SAS 9.4 version (SAS Inc., Cary, NC, USA). Two-sided P-values less than 0.05 were considered significant.Based on inclusion and exclusion criteria, 110,068 participants of a health screening program with oral health checkups in 2009–2010 were included (Figure 1). A comparison of the characteristics of participants who did and did not undergo complete oral examination in the NHIS-HEALS is shown in Table S1. Participants who underwent oral health examinations were predominantly male, younger in age, and had a higher household income as compared to those who did not. The prevalence of diabetes mellitus, hypertension, and chronic kidney disease was lower in participants who received oral health examinations. Finally, 466,753 BMI measurements were included from the repeated health examinations conducted for the included participants. Table 1 shows the serial measurements of BMI levels. The mean baseline BMI value was 23.95 (2.66) kg/m2. The median number of BMI measurements per participant was 4 (interquartile range (IQR), 3–6), and the median follow-up time was 5.14 years (IQR, 4.06–5.95 years).The baseline characteristics of the study participants are shown in Table 2. The mean age was 56.6 (7.8) years at the baseline examination, and the proportion of males was 60.0%. The prevalence of periodontitis, dental caries, and tooth loss was 13.1%, 51.1%, and 25.8%, respectively. Moreover, 49.2% of participants presented a tooth brushing frequency of ≥3 times/day (Table 2). Characteristics of participants according to the presence of periodontitis and frequency of tooth brushing are shown in Table 3 and Table S2. Participants with periodontitis were predominantly male and older, presenting the highest quartile of household income, a higher proportion of current smokers, and a higher frequency of alcohol consumption and physical activity as compared to those without periodontitis. The BMIs of participants with periodontitis were significantly higher than those of participants without periodontitis. In addition, participants with periodontitis were more commonly accompanied by comorbidities such as hypertension, diabetes mellitus, and chronic kidney disease.A multivariate linear mixed model was used to evaluate the longitudinal changes in BMI (Table 4). There was a positive relationship between presence of periodontitis and BMI (β = 0.0687, standard error = 0.0225, p = 0.002). Furthermore, the presence of dental caries and tooth loss were both positively related with BMI (β = 0.0735, standard error = 0.0152, p < 0.001; and β = 0.1328, standard error = 0.0175, p < 0.001, respectively). In contrast, there was a significant negative correlation between tooth brushing frequency ≥3 times/day and BMI (β = −0.2610, standard error = 0.0306, p < 0.001). These results mean that persons with periodontitis, dental caries, and tooth loss (positive value of β for BMI) had higher values of BMI than those without them. On the other hand, persons with frequent tooth brushing had lower values of BMI (negative value of β for BMI). In the linear mixed model for BMI, we did not find a significant interaction between time and oral health indicators (dental caries, tooth loss, periodontitis, and frequency of tooth brushing). When we evaluated the interaction effect between periodontitis and other oral health indicators on BMI in the model, no significant interaction was noted (Figure 2, all p value for interaction > 0.05). These data suggested that the significant association between the oral health indicators and body mass index were consistent regardless of the presence of periodontitis.Our study demonstrated that poor oral health-related diseases or indicators, such as periodontitis, dental caries, and tooth loss were positively associated with longitudinal changes in BMI, suggesting an increased risk of obesity. Moreover, improved oral health behavior of frequent tooth brushing was negatively associated with BMI.Previous studies have evaluated the association between periodontitis, poor oral health, and obesity. In a meta-analysis of previous cross-sectional studies, periodontitis was strongly associated with increased BMI [18]. Moreover, the severity of periodontitis positively correlated with the degree of obesity [19]. Furthermore, in previous studies, periodontitis was closely associated with oral hygiene indicators [4,5]. In our study, with respect to the association of periodontitis with BMI, there was no statistical interaction between periodontitis and other oral hygiene indicators. Therefore, our findings are meaningful in that our results showed that periodontitis was independently associated with BMI regardless of other oral hygiene indicators.Increased tooth loss, a marker of poor oral health, was also significantly associated with obesity [20]. In a study on older people in Brazil, tooth loss was positively associated with obesity [21]. However, the relationship between obesity and oral health remains inconclusive, especially in longitudinal settings. Our study is in line with these previous studies, and it is relevant because it provides evidence regarding the association of periodontitis and poor oral health indicators with the longitudinal changes in BMI.In our study, the general population presenting frequent tooth brushing habits had a low BMI, even after adjusting for the confounding factors of oral health indicators. In a previous cross-sectional study, less-frequent tooth brushing was associated obesity, even after adjusting for oral health indicators (odds ratio: 1.22 and 1.48 for tooth brushing frequency of 1 time/day and 0 time/day, respectively) [22]. In addition, a recent general population-based longitudinal study of 4537 participants in Japan demonstrated that a low frequency of tooth brushing (≤1 time/day) was associated with the occurrence of obesity (prevalence rate ratio: 1.77, 95% confidence interval: 1.12–2.80) [23]. Our research results were similar to these previous studies; particularly, this longitudinal cohort study suggested a negative association between frequent tooth brushing and BMI.Inflammation is thought to be an important factor that can explain the association between poor oral health indicators and obesity [24]. In the presence of poor oral health, including periodontitis, an inflammatory reaction occurs in the oral cavity [25]. The enhanced local inflammation can spread systemically through tooth loss or periodontal pockets, leading to increased secretion of endotoxins or inflammatory cytokines [26,27]. This systemic inflammation plays a decisive role in the pathogenesis and development of obesity [28]. The link between tooth brushing and obesity can be explained by a leptin-related pathway that controls the balance of appetite and energy. Intraoral proprioception-related pathways are linked to histamine-hypothalamic pathways, which are closely related to leptin signaling pathways [29]. Therefore, it is presumed that stimulation of the oral cavity and possible maintaining oral health by frequent tooth brushing can suppress appetite and reduce the risk of obesity. Further research in elucidating the mechanism of oral diseases including periodontitis, dental caries, and tooth loss should be followed.Our study had some limitations. First, our dataset included only Asian people, mainly Koreans. Second, our dataset may have recall bias because our tooth brushing dataset was collected using a self-reported questionnaire. Third, the burden and location of dental caries, tooth loss, and periodontitis were not investigated due to the lack of detailed data in the nation-wide oral health checkup program used. Fourth, for our dataset, our study was conducted using the 2009–2010 health screening cohort dataset. At this time, because there was a lack of consensus on periodontitis in the NHIS-HEALS, probing was not performed on all participants, and it was not possible to confirm the agreements regarding the presence or severity of periodontitis. For this reason, we defined periodontitis with the ICD-10 codes and actual treatment claim codes in our study. Further research using a clear definition of periodontitis is needed in the future. Fifth, information regarding education levels, marital status, and other biomarkers representing inflammatory reactions were not available in the NHIS-HEALS dataset. Sixth, our observational study was without any interventions, and it cannot clearly suggest the mechanisms that explain the relationship of periodontitis and oral health indicators or behaviors with longitudinal changes in BMI. Seventh, our dataset does not provide information on the causes of tooth loss or types of dental caries, that is, active and treated arrested caries. Eighth, because the design of our study was a retrospective observational study, we checked the baseline characteristics of oral health from the participants only once, and the oral health behaviors and comorbidities may have changed over time.In conclusion, the presence of periodontitis, dental caries, and tooth loss may be associated with a higher BMI, whereas frequent tooth brushing may be related to lower BMI. Better oral health might be associated with lower BMI. Further study is needed to determine the effects of oral health behavior and dental disease on obesity.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111062/s1, Table S1: Comparison between the characteristics of patients who completed oral health examination and those who did not, Table S2: Characteristics of the study participants according to the frequency of tooth brushing per day.Conceptualization, Y.C., J.J., T.-J.S. and J.K.; methodology, Y.C., T.-J.S. and J.K.; validation, Y.C., J.J., J.-W.K., T.-J.S. and J.K.; formal analysis, J.J., J.-W.K. and J.K.; investigation, J.J., J.-W.K. and J.K.; resources, Y.C., T.-J.S. and J.K.; data curation, Y.C. and J.J.; writing—original draft preparation, Y.C., J.J., J.-W.K., T.-J.S. and J.K.; writing—review and editing, Y.C., J.J., J.-W.K., T.-J.S. and J.K.; visualization, J.J. and J.-W.K.; supervision, T.-J.S. and J.K.; funding acquisition, T.-J.S. and J.K. All authors have read and agreed to the published version of the manuscript.This research was funded by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, grant number 2020R1I1A1A01060447 (J.K.) and 2021R1F1A1048113 (T.-J.S.).The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Seoul Ewha University Medical Center (SEUMC 2020-08-18).Participant consent was waived due to this being a retrospective analysis based on fully anonymized data.The data that support the findings of this study are available from NHIS-HEALS, but restrictions apply to the availability of these data, which were used under license for the current study and so are not publicly available.The study used the NHIS-HEALS dataset (NHIS-2018-2-055) created by the National Health Insurance Sharing Service.The authors declare no conflict of interest.Flow chart showing the inclusion and exclusion criteria. NHIS–HEALS, National Health Insurance Service-National Health Screening Cohort in Korea.Body mass index according to the presence of periodontitis and oral health indicators: (a) BMI of patients according to frequency of tooth brushing, (b) BMI of patients according to the presence of dental caries, (c) BMI of patients according to the presence of tooth loss. Data are estimated as mean and 95% confidence intervals based on the multivariable linear mixed models for body mass index (fixed effect variables are the same as the models in Table 4). There was no significant interaction effect between periodontitis and other oral health indicators on body mass index (p value for interaction >0.05).Serial measurement of body mass index.BMI, body mass index; SD, standard deviation.Characteristics of patients at baseline examination.Data are expressed as mean (standard deviation) or n (%). Q: quartile.Characteristics of patients according to the presence of periodontitis.Data are expressed as mean (standard deviation) or n (%). p value is derived from the independent t-test and chi-square test. Q: quartile.Association of periodontitis and oral health indicators with body mass index.Data are derived from a multivariable linear mixed model for longitudinal data of body mass index, which includes sex, age, household income, smoking status, alcohol consumption, physical activity, the presence of diabetes mellitus, hypertension, and chronic kidney disease, aspartate aminotransferase, alanine aminotransferase, and total cholesterol level as fixed effects.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11063.txt

@@ -0,0 +1 @@
+The COVID-19 pandemic has negatively impacted the physical and mental health of many and has necessitated widespread societal shifts, including changes to work and family activities. These changes have impacted individuals’ identity, including their sexual self-image and body image, yet research on perceptions of these changes is missing. This study reports on quantitative and qualitative data from an electronic survey with adults in the United States (N = 326) to examine these perceptions. Body appreciation did not significantly differ between demographic groups. Themes emerging from the qualitative results included changes in general self-image (becoming more restricted or disempowered), changes in sexual self-image (deepening, becoming more sexy/sexual, or less sexy/sexual), and changes in body image (positive, negative, and neutral). Our findings point to positive, negative, and neutral effects on sexual self-image and body image, implying that nuanced approaches are needed to understand how identity has transformed as a result of the COVID-19 pandemic.COVID-19 has direct effects on people [1] e.g., difficulty breathing, fatigue, loss of taste or smell, nausea, gastrointestinal distress; as well as indirect psychological, emotional, and relational effects. Social distancing guidelines, work from home orders, and shelter-in-place ordinances, though protective of physical health, preclude ordinary face-to-face social interaction. Preliminary research indicates that COVID-19 may increase loneliness [2], stress [3], suicidality [4,5], and interpersonal violence [6,7].One salient impact of COVID-19 is on people’s identity, or sense of self, given the isolation and disconnection that many have experienced. A few studies have specifically explored the links between COVID-19 and identity. Jaspal and Nerlich [8], for instance examined the relevance of social representations theory and identity process theory to the COVID-19 context. Other scholars have investigated how COVID-19 impacts identity in particular contexts, including teaching communities about COVID-19 prevention [9], moral injury in healthcare professionals [10], and discrimination and mental health [11]. Broadly, we know from the literature exploring the impact of cultural trauma that this type of trauma often leads to changes in core personal beliefs [12,13]. We argue that two core personal beliefs include our body image and concept of a sexual being, and that they have both been challenged as a result of the cultural trauma of the COVID-19 pandemic.Our goal in this study was to investigate the influence of COVID-19 on two specific facets of identity: sexual self-image, or how one sees oneself as a sexual being, and body image. Indeed, body image has implications for sexual functioning [14] and vice versa [15]. We use the term “sexual self-image” to distinguish from some of the closely related concepts such as “sexual self-concept,” which can be defined synonymously with sexual self-image [16,17] e.g., “the feelings a person has about themselves as a sexual being”; or more often, include sexual self-efficacy, arousal, exploration, lack of anxiety, and sexual self-esteem [18]. Due to the limited research on sexual self-image, our rationale draws from these closely related concepts. In a multidimensional view of sexual health, sexual self-concept has been positioned as foundational for other domains [19,20]. Sexual self-concept has been linked to sexual behaviors [21,22], sexual functioning [23], and sexual satisfaction [24,25] and sexual self-esteem increased sexual self-efficacy [26]. Not only does sexual self-image impact sexual health outcomes, it also is affected by sexual health. Individuals with a sexually transmitted infection (specifically herpes or HPV) were more likely to have negative sexual self-concept [16].From a qualitative, explorative standpoint, we allowed participants to determine what was salient for them about their sexual self-image rather than prescribed domains. Although most of the research related to the COVID-19 pandemic and sexuality focused on sexual behaviors, we sought to widen the scope of existing research by focusing on how people view themselves as whole sexual beings. Such a focus necessarily includes attention to individuals’ relationships with their bodies, as sexual health includes “physical, emotional, mental, and social wellbeing in relation to all aspects of sexuality and reproduction, not merely the absence of disease, dysfunction, or infirmity” [27]. Thus, our objective was to examine intersections among sexual self-image and body image.Theories of identity argue that identity formation is a fluid and dynamic process that occurs across the lifespan. Social identity theory, for example, suggests that identity is based largely on social comparison, as individuals identify with some people as ingroup members and differentiate themselves from others as outgroup members [28,29]. As social environments change, so does the negotiation of identity [30,31]. Similarly, the communication theory of identity contends that identity is multilayered, consisting of personal, enacted, relational, and communal components, all of which shift and develop over time [32,33]. Finally, identity theory claims that identity formation rests on the reciprocal relationship between social structures and the self, which is variant across the life course [34,35]. Common to all of these theories are the ideas that (a) identity formation is ongoing and (b) individuals forge their identities in connection with other people.Identity theories point to major life transitions as catalysts for identity changes. These moments have been referred to as turning points, which mark important or salient events in the life course [36]. Strauss [37] defines turning points as “critical incidences [that] force a person to recognize that ‘I am not the same as I was, as I used to be’” (p. 149). In this study, we argue that COVID-19 is likely a transition that has influenced people’s sense of self, so our focus is on how the pandemic, conceptualized as a turning point, influences sexual identity and body image.Though there are many factors that comprise identity (e.g., personal sense of self, social interactions, communal identity), we focus on sexual sense of self and body image in this study. These two components of identity are intricately related [38]. For example, negative messages from a sexual partner about one’s body decreased confidence, self-acceptance, and sexual empowerment [39]. Moreover, body image is related to sexual satisfaction [40] and arousal and orgasm [41]. Beyond their relationships to one another, sexual identity and body image are also important to overall health and wellbeing. For instance, women’s body satisfaction and sexual self-schemas were both related to life satisfaction and positive affect [42]. Within the United States, the COVID-19 pandemic has affected sexual behaviors [43,44,45], relationships and intimacy [46,47], sexual drive [48], perceptions of sexual risk [49], and technology use in sexuality [50].Relationships and intimacy have been challenged with the circumstances of the pandemic. Couples have been physically distant or confined together throughout the pandemic [46]. In addition, emotional distance or closeness has been influenced by widespread psychological distress or increased time spent with one another, among other factors [46,47]. Technology has been useful for some partners to maintain intimacy during social distancing, both through sexting (sharing sexual text or photos) as well as through video calls [46,50]. However, Banerjee and Rao [50] point out that solo sexuality has also been aided by technology use during the pandemic, with increases in masturbation and pornography use.Sexual behaviors have been a focus of early COVID-19 research. Sexual drive and desire has often been reported as decreased, with some experiencing increases or deepening their understanding of their sexuality [48]. Some scholars reported overall decreases in sexual behavior [43] or noted that frequency of sexual intercourse was generally the same, on average [27]. Others offered more nuance in their reports. Coombe et al. [51], for instance, noted that sex with a dating partner or casual hook-up was reported less frequently, whereas sex with a spouse was more common during lockdown. In the same study, the authors reported an increase in masturbation and in solo use of sex toys [51], and multiple studies identified an increase in pornography consumption [27,52]. In another report, frequency of sexual behavior was dependent, in part, on the age of children in the home; those with children under five reported increased intimate behaviors, whereas those with children aged 6–12 years reported a decrease in the same behaviors [43]. Finally, some studies identified nuances based on sex or gender. For instance, nearly 40% of Jacob et al.’s [53] sample engaged in some sexual activity at least once per week, and this was particularly true for younger, employed men. Taken together, these studies illustrate that though sexual behavior has likely changed, pinpointing a pattern in those changes is challenging because sexual behavior is dependent on so many other factors.One of those factors may be social isolation. Several scholars, for example, have argued that social isolation has a negative psychological impact, which in turn harms libido [54,55,56]. Offering empirical support for this hypothesis, Hensel et al. [43] found that depression and loneliness were related to reduced partnered sexual behaviors. Two other potential pathways are also concerned with psychological health, suggesting that depression affects sleep patterns, which can reduce desire for sexual activity [54] or that depression medication can decrease libido [54,57]. Relatedly, stress can have physiological effects on sexual functioning [27,57], and COVID-19 may decrease immune response, which alters sexual functioning [54]. A final pathway is related to risk behavior: people may not want to take on new sexual partners because of the risks associated with sex or a fear of contagion [27,55,58]. To date, these explanations are speculative, but they have received substantial attention in academic reviews and commentary related to COVID-19.Other work on COVID-19 and sexual identity focuses on intimacy, particularly in established or long-term romantic couples. Scholars argued that intimacy could be strengthened for some couples because of increased time together, fewer social or family obligations, or less work burden, but they also cautioned that several factors could damage intimacy, including stress, lack of privacy, and increased chances for interpersonal conflict [57,59]. Similarly, Ibarra et al. [27] suggested that being together all the time could exacerbate conflict and weaken couples’ relational bonds, which in turn could harm sexual behavior. There is some preliminary empirical support for these arguments. For instance, only half of Arafat et al.’s [59] sample reported positive changes to their emotional bonding. Approximately one-third of participants in Luetke et al.’s [60] study noted that they had experienced COVID-19-related conflicts in their partnership, and conflict frequency was related to less frequent sexual behaviors, decreased experiences of orgasm, and feeling emotionally distant from one’s partner during a sexual event. Clearly, there is potential harm to romantic relationships because of the challenges of COVID-19.Similar to the nascent literature focused on the relationship between COVID-19 and sexual identity, the emerging work on COVID-19 and body image hypothesizes a relationship between COVID-related stress and anxiety and increased rates of body dissatisfaction [61,62]. Recent empirical work has supported these hypotheses, e.g., [63,64,65]. For example, Swami et al. [63] sought to understand how stress and anxiety from the pandemic related to body image when controlling for other relevant psychological variables such as perceived stress, stressful life events, and trait anxiety. They found empirical support that COVID-19-related stress and anxiety were significantly associated with negative body image, and that the relationship between COVID-19-related stress and negative body image was more pronounced for women. Flaudias et al. [64] explored similar questions but focused on an undergraduate student population. They found that pandemic-related stress was related to a greater likelihood of binge eating and dietary restrictions, and that this relationship was especially true for participants who identified as female. Flaudias et al. [64] also highlighted that the pandemic may create body image and eating concerns for those already at-risk for problematic eating behaviors. Finally, Robertson et al. [65] provided a more nuanced picture in understanding perceived changes in body image in a UK population. The authors found that women, specifically young women, were more likely to report changes in their thoughts and behaviors related to their bodies, which broadly mirrors other literature focusing on the overall mental health impact of the COVID-19 pandemic [66,67]. Robertson et al. [65] also highlighted the association between changes in body image to reports of higher levels of psychological distress related to COVID-19. This may also be linked to changes in sexual identity and activity.The pandemic and stay-at-home orders, in particular, have been stressors on individuals’ daily lives, interrupting routines, exercise habits, and accessibility to food [63,68]. Additionally, social media use has dramatically increased during this time [69], with a corresponding rise in content related to stigmatizing messages about weight gain [70,71] e.g., “Quarantine 15”; body shaming, and conflicting messages regarding the relationship between weight and COVID-19 risk [65]. Indeed, Flaudias et al. [64] found that increased exposure to COVID-19-related media was significantly associated with restrictive eating behavior, which, in turn, was associated with higher levels of body dissatisfaction. Clearly, the pandemic has impacted how individuals interact with their bodies, and preliminary evidence suggests it is often detrimental to psychological well-being.In the current study, our goal is to add to the burgeoning literature on COVID-19, sexual behavior, and identity by focusing on how people have experienced changes to their sexual sense of self and body image during the pandemic. Given the evidence of impacts for both changes in sexual being and body image as a result of the pandemic, we examine them together to highlight nuances and interactions between these two domains. Whereas most prior work has focused narrowly on sexual behavior, we widened the lens in this study by asking people about their sexual sense of self more broadly, including not only behavior, but also their perceptions of their sexual selves, their ways of interacting with others, and their relationships with their own bodies. Accordingly, our study was guided by the following research question: How do people experience changes to the ways they think about themselves, including their sexual sense of self and body image, during COVID-19?We conducted an online survey with open- and closed-ended questions with individuals in the United States over the age of 18. The larger study examined the effects of the COVID-19 pandemic on sexuality and relationships. This analysis focuses on changes to self-image and body image. Participants were recruited through snowball recruitment, health-related listserv announcements, social media postings in health-related groups (including COVID-19-specific groups) and population-related groups (to include underrepresented groups such as people with disabilities and LGBTQ individuals). Participants provided electronic consent and were entered into a drawing for an e-gift card upon completion of the survey. All procedures and protocols were approved by the authors’ institutional review board. Anonymous data files are available upon reasonable requests to the authors.A total of 326 participants completed the survey. Participants varied in age (range 18–77); however, on average, participants were 30.6 years old (SD = 11.22). The majority of participants identified as women (n = 247, 76.0%), with smaller proportions of men (n = 73, 22.5%) and gender nonbinary and agender identifying individuals (n = 5, 1.5%). Participants overwhelmingly reported they identified as White (n = 246, 75.7%), with 6.5% (n = 21) who identified as Black, 6.5% (n = 21) as Hispanic/Latinx, 4.0% (n = 13) as Asian, 1.2% (n = 4) as American Indian, 0.3% (n = 1) as Middle Eastern, 5.5% (n = 18) as two or more identities, and 0.3% (n = 1) did not specify. Most participants identified as heterosexual (70.8%; n = 230), 3.7% (n = 12) as gay, 1.8% (n = 6) as lesbian, 12% (n = 39) as bisexual, 0.3% (n = 1) as pansexual, 2.5% (n = 8) as queer. 0.9% (n = 3) as asexual, 0.9% (n = 3) as other, and 7.1% (n = 23) as two or more identities. The majority of participants were in a committed relationship with one person (n = 210, 64.4%), whereas 20.2% (n = 66) of participants were single or not dating, 10.1% (n = 33) were dating casually, 1.2% (n = 4) were in a committed relationship with more than one person, 2.5% (n = 8) selected two or more statuses, and 1.5% (n = 5) reported other.The overall survey took approximately 15 min to complete via Qualtrics. Relevant open-ended questions including “How has your sexual being changed since the COVID-19 pandemic began?”; “How have those changes influenced your larger sense of self, or how you see yourself?”; and “How have those changes influenced your sense of empowerment, or feeling capable and strong?”.The 10-item Body Appreciation Scale-2 [72] (BAS-2); was used to evaluate participants’ acceptance and respect for their bodies. Participants indicated how often, on a 5-point Likert scale, 10 statements were true about them. Sample items included “I respect my body” and “I appreciate the different and unique characteristics of my body.” Responses ranged from 1 = Never to 5 = Always. Items were averaged to form a total score, with higher scores indicating higher levels of body appreciation (M = 3.59, SD = 0.82, α = 0.95). Bivariate analyses were conducted to explore potential demographic differences in body appreciation. Specifically, a correlation was conducted to determine if a relationship existed between age and body appreciation. A series of one-way ANOVAs were also conducted to determine if differences in body appreciation existed by gender, sexual orientation, or relationships status. Open-ended questions were coded in Dedoose [73]. We used inductive thematic analyses in which common ideas were grouped together to create themes [74]. All responses were coded by two coders trained in qualitative analyses. An initial codebook was created through preliminary analyses. After an initial coding of 30 participants, the codebook was augmented. The Dedoose “test” function confirmed coder interrater reliability; any codes with less than 0.8 Kappa were discussed and refined. To explore a potential bivariate relationship between age and body appreciation, a correlation between the two variables was calculated. There was not a significant relationship between age and body appreciation, r(187) = 0.02, p = 0.84. A series of one-way ANOVAs were also calculated to determine if differences in body appreciation existed by gender, sexual orientation, or relationship status, respectively (see Table 1). Body appreciation did not differ significantly by gender (F(3,182) = 1.19, p = 0.32), sexual orientation (F(8,178) = 1.29, p = 0.25), or relationship status (F(5,181) = 0.86, p = 0.51).Participants’ general self-image was more reserved (Table S1, 1a) or honed through doing things alone (Table S1, 1b,c). The isolation from others negatively affected some participants’ self-image, including their sense of empowerment (Table S1, 1d). Part of this related to individuals’ routines that were disrupted along with their sense of themselves as disciplined (Table S1, 1e). Relationships and sexuality that were limited due to the pandemic negatively affected participants’ overall self-image (Table S1, 1f,g). Some participants worried about the permanence of these changes in their self-image (Table S1, 1h).In relation to participants’ sexual self-image changes as a result of the pandemic, participants (a) deepened their understandings of themselves, (b) viewed themselves as more sexy or sexual, or (c) viewed themselves as less sexy or sexual. There were a few participants who described movement in both positive and negative ways, specifically in deepening understandings of self while at the same time having reduced opportunities for sexual engagement or feeling sexy (Table S1, 2a–c). Some participants were questioning their sexuality (mostly their sexual identity; Table S1, 2d) and learning to love themselves through their sexuality—including kinks (Table S1, 2e,f). One participant’s pride in their sexual identity off-set their insecurities about sexual inactivity (Table S1, 2g). Quarantine provided some participants the opportunity to re-examine their sexual self-image, to be more “loving and open” (Table S1, 2h) or to reframe their self-image of having a low sex drive (Table S1, 2i). The view of oneself as more sexual led some participants to more positive self-image overall (Table S1, 2j). Some participants contextualized their sexuality as just one component of their overall being (Table S1, 2k).A large subtheme focused on participants viewing themselves as less sexy or sexual. Participants pointed to mental health difficulties, specifically depression, anxiety, or distraction, as one of the drivers of this change (Table S1, 2.1a–c). For others, changes in relationships with others was the key shift—including appearance to others (Table S1, 2.1d), practicing kinks (Table S1, 2.1e), and relationship partners (Table S1, 2c).Participants drew linkages between how they viewed themselves as sexual and their gender. This was primarily true for participants identifying as cisgender women. A few participants seemingly equated being a (strong) woman with being sexual (Table S1, 2.2a,b).Participants discussed positive changes in their body image related to partners—both in spending time together (Table S1, 3a) and in sex drive (Table S1, 3b). Losing weight also increased some participants’ sense of positive body image (Table S1, 3c).Negative body image was part of some participants’ introspection during the pandemic. Participants described noticing their body image was based on looks (rather than feelings of being in the body; Table S1, 4a), the way the pandemic was affecting their eating disorder (Table S1, 4b), and attempting to make peace with their body (Table S1, 4c). A few participants noticed how age was affecting their perceptions, such as making them feel less attractive (Table S1, 4d).Participants’ negative body image was linked to sexual desirability. Participants described feeling undesirable (Table S1, 4.1a) or less sexy (Table S1, 4.1b,c). For some, this reduction in sexual desirability was tied to self-esteem and self-worth (Table S1, 4.1d,e). Reductions in positive body image were tied to (potential) partners. With quarantining and reduced abilities to interact with others, some participants felt they were taking less care of themselves (Table S1, 4.1f). With less validation from partners, some participants struggled with navigating their insecurities (Table S1, 4.1g). Decreased desirability hindered some participants’ sexual interactions with partners (Table S1, 4.1h,i).Many participants touched on body image in relation to exercise, both positive and negative. Participants’ abilities to exercise shifted during quarantine, which increased self-esteem for some (Table S1, 5a). The shift in part occurred because individuals had more time without a commute to work (Table S1, 5b). The pandemic also created opportunities for community building in relation to exercise, which facilitated some participants in working out (Table S1, 5c). However, some participants worked to navigate their exercise routines and eating in the upheaval of the pandemic (Table S1, 5d). On the negative side, some participants were not able to go to their regular gyms or exercise classes, which reduced their self-esteem (Table S1, 6a,b). This lack of physical activity reduced sex drive (Table S1, 6c). Participants ascribed lower amounts of activity to loss of self (Table S1, 6d,e).For others, body image was shifting, but the effects were neutral. Without an “easy other” in their sexuality, participants described appreciating their body more holistically (Table S1, 7a). Similarly, another participant described their confidence as unchanging, and that confidence was the core of their sexiness (Table S1, 7b). For some older participants, the pandemic caused some to notice changes in their aging bodies (Table S1, 7c).Our goal in this investigation was to gain insight into the changes people experienced to both their sexual sense of self and their body image in the midst of the COVID-19 pandemic. Overall, participants reported positive, negative, and neutral changes in these areas. Importantly, our results suggest that body image and sexual sense of self are intricately tied to one another. For instance, positive body image was related to increased sex drive, whereas participants struggled with negative body image when they felt less sexually desirable. Those who fared the best seemed to be the ones who saw their body in a holistic way, integrating confidence with sexiness and their sense of their bodies. This set of findings points to the complexities of studying body image and sexuality—because people experience these parts of their identity in tandem, they are difficult to separate for the purposes of treatment or research. Indeed, scholars have noted links between body image and sexual sense of self in a variety of contexts, including disability [75] and chronic illness [76]. Our findings echo prior research that suggests body image and sexual identity co-vary.The results of our study complement existing research about the pandemic and identity in two important ways. First, most prior research focused on changes to patterns of sexual behavior specifically; we widened the lens to focus on sexual identity more broadly. Certainly, our participants noted changes to their sex drive, the frequency of their sexual engagement with others, and their sexual functioning, in line with existing work on COVID-19 (Hensel et al., 2020; Li et al., 2020; Sansone et al., 2020). However, participants also contextualized those ideas in light of larger changes to their sense of self—including feelings of self-worth, their understanding of themselves, evolving ideas about their sexuality, and mental health struggles. Second, our findings offer an explicit link between sexual identity and body image, in contrast to existing work that examines one or the other. Thus, our work implies that body image may be another potential pathway linking COVID-19 with sexual behavior, in addition to those that have already been proposed (e.g., psychological impact, stress, risk; Abbas et al., 2020; Ibarra et al., 2020; Gaspari et al., 2020).In terms of general self-image, some participants reported concerns about the permanence of disempowerment or restriction of themselves. In addition to general concerns about COVID-19 transmission and societal effects (including employment and financial concerns, as well as social relationships), individuals may be navigating worries about their identities. Some participants contextualize their different identities as only one part of themselves, which gave them stability in their general sense of self and/or sexual self-image. These findings reaffirm a rich history in psychology that points to the importance of multiple selves or identities [77,78]. Specifically, having a number of identities that can be relied upon when one or many are being threatened has been linked to higher psychological well-being, and this appears to be especially true in moments when identities are being shifted or integrated [79]. This finding suggests that the salience and importance of an identity or self-image to an individual must be considered in order to create effective prevention or intervention messages.Participants reporting isolation or lack of social interaction often referenced negative effects on their sense of self, whereas individuals who deepened their self-awareness reported more positive effects. This may point to opportunities for cognitive reframing. Indeed, the post-traumatic growth literature points to the relationship between traumatic events (e.g., the COVID-19 pandemic) and identity, suggesting that trauma such as the pandemic can serve as a lens through which to guide identity development [80,81]. Our findings highlight the ways in which mental health practitioners and interventionists may approach the reported shifts in sexual self-image and body image due to the pandemic.Individuals in our study reported changes in how they see themselves as sexual beings. The lasting effects of these changes post-pandemic are unknown, and research may be needed to follow-up on the continued shifts or support needs for these changes. The descriptions of decreases in sexual self-image, or seeing oneself as a less sexual person, may cause psychological distress. Mental health practitioners may provide support during and post-pandemic for potential concerns.In the case of negative effects on sexual self-image, many participants pointed to connections with mental health symptoms (e.g., depression). This further highlights the overlapping and intersecting relationship between sexual self-image and mental health outcomes (Abbas et al., 2020; Gaspari et al., 2020; Sansone et al., 2020; Hensel, 2020). Societal awareness of the mental health impact of the pandemic is increasing; however, the influence on individuals’ sex lives and identity are often overlooked and this may negatively influence psychological wellbeing. Similarly, some participants discussed their struggles with eating disorders and exercise behaviors as routines, social interactions, and access to food and exercise regimens changed. Research has long emphasized a relationship between trauma and eating disorders, e.g., [82], and our results suggest that the pandemic may be a source of trauma that compounds disordered eating behaviors. Taken together, these findings highlight an intricate relationship between sexual identity, body image, and mental health that clinicians should attend to in their patients.In connecting our exploratory results of sexual self-image to established literature on sexual self-concept, we find many linkages. Within Deutch, Hoffman, and Wilcox’s [18] conceptual model, we see evidence of sexual behavior and sexual attractiveness from their domain of sexual self-esteem and exploration as well as anxiety (though not only restricted to sexual situations but regarding sexuality broadly). However, we do not see sexual conduct (feelings of adequacy about one’s behavior in a sexual situation) or sexual self-efficacy. This is likely because other dominant themes such as attractiveness or arousal were more salient and our measure using short-form questions did not explore these further.Our results also highlight the fragility of people’s relationships with their body. Many participants reported changes in the way they felt about their bodies during the pandemic. Though some of those changes were positive, most positive changes were driven by external factors including time spent with a partner or weight loss. The former encourages positive body image based on feedback from a partner, whereas the latter relies on weight-normative attitudes that stress weight loss and conformity to socially enforced body norms (Tylka et al., 2014). Similarly, some participants reported reduced feelings of self-esteem when they were unable to exercise or, in contrast, increased self-worth because they had more time to exercise. Such dynamics showcase how easily people’s relationships with their bodies can change based on factors outside their control. Overall, this set of findings highlights the value of weight-inclusive approaches to health, such as Health at Every Size [83] and intuitive eating [84], given that the assumptions of these programs stress an internally guided relationship with one’s body.Although we did not find significant differences in body appreciation scores, we did observe trends in the direction of lower scores among gender minority (nonbinary and agender) individuals and those who are single compared to those in relationships. We may not have had sufficient statistical power to detect significant differences among these groups. This trend comports with previous research that finds higher rates of body dissatisfaction among trans and nonbinary individuals compared to cisgender individuals [85,86]. Research in cisgender populations has uncovered differences among men and women, generally reporting lower body appreciation in women compared to men [87]. Our findings suggest that the COVID-19 pandemic may be negatively impacting body image in different ways for men and women. Specifically, women were more likely in open-ended responses to describe their sexual self-image as tied to their gender. Given the widespread nature of women’s sexual objectification [88], social changes in the pandemic may be disrupting the ways some women are viewing themselves. In our sample, men also reflected on negative body image, though not in context of partnerships or their gender but rather mostly just dissatisfaction with their bodies. Men’s negative body image comments coincide with literature pointing to men’s concerns about thinness, leanness, and muscularity at lower rates than women but nonetheless present [89,90,91,92].Some participants described validation and affirmation from others in assisting them in their body and sexual self-image. Single participants may particularly struggle with lacking this external input.We conducted this study relatively early in the pandemic, and follow-up studies may examine how these trends shift throughout the pandemic and when people are able to return to relative normalcy. Moreover, our sample was made up of predominantly White and cisgender women, and we were unable to examine how body and sexual self-image may have been affected for Black, Indigenous, and People of Color. Though we did observe some diversity in our sample in terms of sexual identity, research that focuses more centrally on the experiences of marginalized communities would be valuable in nuance to our results.The COVID-19 pandemic has challenged people’s sense of self in a variety of ways as they navigate social distancing guidelines, work-from-home orders, and changing relationship dynamics. In this study, we focused on two facets of identity, sexual identity and body image, and the ways that people have experienced changes in those areas in the midst of the pandemic. Participants noted positive, negative, and neutral changes to their sexual self-image and body image and note how the two are intricately related to one another. Further, we see evidence of the pandemic’s effects across a wide range of aspects within sexual self-image. Our results highlight the importance of considering sexual sense of self and body image in tandem. Given the shifts in sexual self-image, mental and sexual health practitioners may need to focus on supporting individuals in making sense of themselves during and post-pandemic.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111063/s1, Table S1: Example quotes with participant demographic information (age in years, gender, sexual identity, race/ethnicity, relationship status).Conceptualization, J.B. and E.A.M.; methodology, J.B. and E.A.M., formal analysis, J.B. and E.A.M.; resources, J.B. and E.B.; writing—original draft preparation, J.B., E.A.M., E.B.; writing—review and editing, J.B., E.A.M. and E.B. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of University of North Carolina at Charlotte (protocol code 19-0747 and 21 April 2020).Informed consent was obtained from all subjects involved in the study.Data is available upon reasonable request to the authors.We would like to acknowledge Diana Gioia for her work on making this project a success.The authors declare no conflict of interest.Body appreciation scores by demographic variables.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11064.txt

@@ -0,0 +1 @@
+Household hygiene is critical to prevent pathogen transmission at the household level. Assessing household hygiene conditions and their determinants are needed to improve hygiene conditions, especially in rural and less developed areas where the housing conditions are relatively worse than they are in urban areas. This study used data from 278 household interviews and observations in rural areas in the district of East Sumba, province East Nusa Tenggara, Indonesia. The data were analyzed using statistical methods. In general, the household hygiene conditions in the study need to be improved. The main potential sources of pathogen transmission were from the surrounding environment, i.e., non-permanent floor and garbage, and personal hygiene, i.e., handwashing facilities with water and soap were only observed in the homes of four out of ten respondents. The presence of livestock roaming freely in the house’s yard was another source of contamination. Easy access to water and wealth significantly influenced the hygiene conditions. Implementing low-cost interventions, i.e., cleaning the house of garbage and animal feces and cleaning nails, should be the priority in immediate intervention, while providing easier access to water supply, especially during the dry season, could be a long-term intervention. This paper also argues that analyzing household hygiene conditions or practices should be complemented by analyzing contextual determinants of the hygiene conditions or practices, so that we can develop more precise intervention by considering the local or household context.Appropriate water, sanitation, and hygiene (WASH) practices can prevent the transmission of various infectious diseases. Empirical evidence indicates that household water treatment and safe storage (HWTS) reduces the prevalence of diarrhea among children [1]. Handwashing can reduce cases of diarrhea [2]. Furthermore, despite variations in the results of the effect of proper sanitation practices to reduce diarrhea [3], there is a consensus among scholars that safe excreta disposal can provide a barrier to fecal contamination in the household [4].Since pathogens can reach humans by various routes, a holistic approach or multi-barrier prevention to minimize the risk of pathogens reaching a human is suggested [5,6]. Multi-barrier prevention has been applied in the context of the drinking water supply, i.e., the water safety plan [7]. This approach first starts by identifying potential sources of contamination and then develops relevant interventions to minimize or eliminate the contamination sources. In the context of pathogen transmission in a house, an article by Ercumen et al. [8], in their introduction, elaborates nicely on how water, sanitation, and hygiene practices complement each other to provide multi-barrier prevention for pathogen transmission in a house. For example, they argue that coupling HWTS and handwashing can prevent the recontamination of treated water by dirty hands.In identifying potential sources of contamination, one can use and adapt the sanitary inspection (SI) form. The SI form consists of questions on potential sources of contamination in specific settings, e.g., pipe distribution network, water sources, or household drinking water [9]. The SI was first introduced by the World Health Organization in 1997 [10]. The SI has been widely used, especially in drinking water safety programs [9]. Daniel et al. [11] suggest adapting the SI forms to “the local context to maximize their applicability”. Since there is no “standard” SI available for measuring general household hygiene conditions, the variables that are measured can be inspired by previous studies or available and relevant SI forms. There are some studies on household hygiene conditions that can inspire the questions used to measure household hygiene conditions, e.g., by [12,13]. Questions or variables from the SI form for household drinking water can also be used since they measure hygiene practices related to drinking water; see, for example, the studies of [5,11,14]. One can then use the adapted SI forms to inform which hygiene aspects should be tackled, which can then improve the general household hygiene conditions.Some hygiene-related studies have been conducted in Indonesia, for example, on handwashing practices [15,16,17] and general household hygiene conditions [5,13]. There are is also evidence that household hygiene is associated with children’s malnutrition in Indonesia [18,19]. Furthermore, children living the in rural areas of Indonesia were more likely to be stunted compared to children in urban areas [20,21]. Other studies found that disparities in the WASH conditions or facilities between urban and rural areas exist in Indonesia [22,23,24]. However, studies on the determinants of general household hygiene conditions in rural Indonesia are lacking. This study then aims to fill that gap and aims to investigate the hygiene conditions in households in rural Indonesia. Furthermore, we seek to find contextual determinants, i.e., socio-economic or socio-environmental determinants, of hygiene conditions. These are believed to influence health-related behavior [25,26], including household hygiene practices or conditions. This study provides a snapshot of the household hygiene conditions in less developed areas and indigenous communities in developing countries that are vulnerable to WASH-related diseases and insecurities, whereas reaching those left behind is the key to achieve Sustainable Development Goals 6.1 and 6.2 by 2030 [27]. Furthermore, this study may inspire potential interventions to improve the hygiene conditions of rural households, especially in Indonesia, where there are still many less developed, remote, and also indigenous communities. Finally, this study also shows how more specific interventions can be provided by considering the data of household hygiene conditions and contextual determinants, i.e., socio-economic conditions of the households or community.A cross-sectional study was conducted in the district of East Sumba, Province East Nusa Tenggara, Indonesia, in July–August 2019 (Figure 1). According to the district data, about 30% of the total population practiced open defecation in 2017 [28]. A total of 40% of the total households in this district used wells as their main water source, and only 18% had access to tap water in 2017 [29]. The tap water is distributed without treatment. Furthermore, severe drought usually occurs in April–October, resulting in many people, especially those in the rural parts of the region, face serious water supply problems [30]. An indigenous belief, called “Marapu”, is commonly practiced by people in the study area and has a large influence on the daily life of the people [31]. Moreover, poverty is a serious problem in this area, and the prevalence of child malnutrition is also high [32,33]. About 30% of the total population in that district were categorized as “poor households”, i.e., cannot afford basic food. The level of school drop-out is also high, i.e., more than 20% [34]. For this study, 328 households in 9 villages were visited. This visit in 2019 is the continuation of a previous WASH study in 2018. In 2018, the households that were visited were randomly selected during a transect walk within nine villages and enrollment was asked of every, for example, five houses. Information on village selection has been discussed in a previous study [35,36]. Data collection was conducted by six local enumerators who were trained to conduct a household interview, sanitary inspection, and water sampling. An article by Sonego and Mosler [12] indicated data collection approaches to study hygiene practice: self-reports, structured observations, and spot-checks. Moreover, Kelly et al. [9] also suggest complementing SI with water quality analysis, especially in the context of water safety. Therefore, in this study, the water quality data were included in the analysis. The drinking water samples were collected from all respondents, and we assessed the presence or absence (P/A) of fecal contamination in the drinking water storage, i.e., presence of E. coli in a 1 mL water sample. More information about the sampling procedure and analysis can be found in another study [5]. The observational data were collected using the Open Data Kit (ODK) software on a smartphone and were then transferred to a computer for analysis. This study protocol was approved by the Human Research Ethics Committee of the Delft University of Technology and the Agency for Promotion, Investment, and One-Stop Licensing Service at the district level. Participation was voluntary, and we obtained informed consent from all respondents.We used 11 variables as determinants of household hygiene conditions. These variables were found significantly related or often measured to determine WASH-related behaviors or practices in developing countries, e.g., HWTS practices, sanitation, or handwashing behaviors: (1) wealth [37,38], (2) access to water [36,39], (3) household time allocation or spare time [39,40], (4) local beliefs [36,41,42], (5) access to market [43,44], (6) access to mass media or information [45], (7) receiving WASH promotions [45,46], (8) mother’s education [47,48], (9) father’s education [43,49], and (10) the presence of children under five-years-old [47,48,50]. The variable “wealth” was measured by household assets and conditions, e.g., the house’s floor, walls, the availability of electricity, a TV, and a car. “Access to water” was measured in five scales: “below 5 min”, “5–15 min”, “16–30 min”, “31–45 min”, and “>45 min”. The “spare time” was measured in hours and was used to measure the mother’s busyness during the day. The assumption is that a busy mother does not have time for household hygiene-related practices. The local belief was coded as following Marapu or not, i.e., beliefs other than Marapu, e.g., Christianity, Catholicism, and Islam, were coded as “not” a local or indigenous belief. This categorization was made because a previous study found that there is a significant relationship between indigenous beliefs and the HWT practices [36]. “Access to market” was coded as either “difficult access” or “easy access” based on the relative distance to the city center, i.e., Waingapu (Figure 1). The variable “access to mass media” was measured by asking the frequency at which TV was watching in a day using five frequency scales (“almost never”, i.e., score “1”, to “very often”, i.e., score “5”). For the variable “receiving WASH promotions”, the answers were “yes” or “no”, and there was no specific time frame asked for this question, e.g., promotions could have been received in the past month, past year, or even longer. The education of the mother and father were measured by year. Finally, the variable “have children under 5 years old” was coded as “yes” or “no”. There are 16 hygiene-related variables used in this study (Table 1). As mentioned in the introduction, these variables were selected based on previous household hygiene studies [5,11,12,13,14]. This study categorized those variables into four main clusters of household hygiene: sanitation, surrounding environment, drinking water, and personal hygiene (Figure 2). A variable toilet facility is related to sanitation. There are six variables related to the surrounding environment: “livestock nearby”, “floor types”, “floor cleanliness”, “feces around”, “garbage around”, “flies around”, and “food storage”. The next cluster is drinking water, which consists of variables measuring the water storage conditions (cover, crack, placement, and cleanliness), whether they practice household water treatment or not, and this was complemented with drinking water quality, i.e., whether E. coli was detected or not. Finally, the last cluster concerns personal hygiene, i.e., related to fingers, which comprises the presence of handwashing facilities and the condition of respondent’s nails. This study assumes that those variables represent general household hygiene conditions. Due to missing data in some respondents, only data from 278 respondents with complete questionnaires were analyzed in this study (85% of the total data). The wealth index was created using the principal component analysis of the household assets and conditions. The first principal component was assumed to represent the relative wealth index of the households [51]. To allow comparison between variables, we standardized the scores from “0” (worst condition) to “2” (best condition). For example, in the variable “floor cleanliness”, which has three scales, the value “0” means “dirty”, “1” means “quite dirty”, and “2” means “clean”. For the variables with two scales, i.e., binary variables (“0” or “1”), they were recoded into “0” and “2”. For example, in the variable “practice HWT”, the value “0” means “do not treat drinking water” and “2” means “treat drinking water”. In this study, all of these variables were assumed to have equal weight on general household hygiene conditions. The scores were summed to achieve the composite value of general household hygiene conditions, as implied by the sanitary inspection form [9]. This means that the worst possible household hygiene conditions have a score of “0” and that the best conditions have a score of “32”. We then divided the scores into 3 levels to categorize the hygiene level of households: scores 0–10 as “poor hygiene”, scores 11–21 as “moderate hygiene”, and scores 22–32 as “good hygiene”. Bivariate nonparametric Chi-squared (X2) tests to assess potential relationships between hygiene-related variables were also conducted. The Bonferroni adjustment of the p-value was applied to reduce the false-positive result error in the Chi-squared tests. The adjusted Bonferroni p-value was 0.0035 (0.05/14). Finally, the forced-entry linear regression analysis, i.e., all independent variables were entered simultaneously into the model, using 10 potential contextual determinants as independent variables on the composite of the value of general household hygiene conditions was conducted. All statistical analyses were conducted using IBM SPSS Statistics 25. Most of the respondents were a mother (84.5%), while others were the father or household head. The proportion of households with and without children under the age of 5 was almost equal, i.e., 47.8% and 52.2%, respectively. The mean schooling time of the mother was 7.7 years (SD = 3.6), while that of the household head or father was 7.4 (SD = 3.6). About one-fourth of the respondents (26.3%) practiced the local belief “Marapu”. The mean spare time of the respondents was 2.9 h in a day (SD = 2.1). In terms of the accessibility of the location of the respondent’s house, about 54% of the respondents were located in an area where it was relatively difficult to access the market. The majority of the respondents had a non-permanent wall (87.8%), e.g., bamboo or wood, and a non-permanent floor (72.3%), e.g., bamboo, earthen, or compacted soil, while only 7.9% had a non-permanent roof, e.g., straw. Having livestock was common, i.e., 89.9% of respondents had at least one pig, and 25.5% had a cow(s). A total of 78.8% of households were observed to cover their food. Figure 3 shows examples of the house kitchen conditions in the study area. Thirty-three percent of the respondents drew water from the tap, which was either a private (16.2%) or a public tap (16.5%), while the majority (57.0%) relied on surface water, e.g., a shallow dug well, river, or spring, and 10.1% bought commercial water, e.g., from a refill water station or water truck. More than half of the respondents (51.8%) had a water source that at a close enough distance that it could be reached within a five-minute walk per trip, while 21.5% of the respondents needed to make a walking round trip that was more than a half an hour to acquire water. Open defecation was still practiced by 31.3% of the respondents. Handwashing facilities with water and soap were only available in 39.6% of total households. Only 59.7% of respondents reported that they treat their water. About 87.4% of the respondents mentioned that they have received or participated in promotional WASH activities. The mean of the composite score was 19.8 (SD = 4.8, range values = 4–31) out of the highest possible value of 32. The households were then categorized into three categories based on their scores, i.e., poor hygiene to good hygiene (see Section 2.4). The percentages of households categorized as “poor hygiene”, “moderate hygiene”, and “good hygiene” were 4.0%, 57.2%, and 38.8%, respectively. The mean values of the 16 variables related to the hygiene conditions are shown in Figure 4. The individual variable scores were the lowest on average for “floor types”, followed by “handwashing facilities” (with water and soap), and “garbage around”, and they were the highest for “storage cracked” followed by “water quality” and “food storage”.The relationship tests between 15 hygiene variables are shown in Table 2. The relationships between the first six variables are related to the surrounding environment, i.e., from “livestock nearby” to “flies around”, were in a positive direction, i.e., the better the condition in one variable, the better the condition in other variables were. For example, no or few livestock nearby the house was associated with fewer flies around. These five variables were also significantly related to the house having no permanent floor. Furthermore, the relationship tests indicate that the poor conditions of these first five variables were associated with having no food cover, i.e., people did not cover their food even though the surrounding environment was not hygienic or clean. The relationships between “handwashing facilities” and “respondent’s nails” with “livestock nearby”, “floor cleanliness”, “feces around”, and “garbage around” were in a positive direction. Households who practiced HWT tended to cover their water storage. Furthermore, there was no variable related to “water quality”, i.e., detected E. coli in the drinking water, “storage cleanliness”, and “storage cracked”. The regression analysis indicated that significant determinants were “wealth” and “access to water” (p-value < 0.05; Table 3). The wealthier the households were, the better the household hygiene conditions were, i.e., a positive regression coefficient. On the other hand, difficult access to water was associated with lower hygiene conditions. The mean values of composite hygiene and household categorization indicate that most households in the study area did not practice proper hygiene practices in their houses. The majority of the respondents were categorized as having “moderate hygiene”, i.e., 57.2%. There are many potential sources of contamination at the household level and these put them at risk for the spread of pathogens in their houses. The results indicate that the main contamination pathways were through the environment, i.e., non-permanent or earthen floor and garbage, as well as personal hygiene, i.e., handwashing. These findings are in line with other studies that found that indigenous communities often lag behind in terms of WASH facilities or practices [52,53,54]The risk of pathogen transmission through a non-permanent floor cannot be overlooked, considering that 72.3% of the respondents had a non-permanent floor. Moreover, the non-permanent household floor was associated with a dirty floor or the presence of garbage and flies around. Previous studies indicated that an earthen floor possesses a higher chance of pathogen transmission in Bangladesh, Kenya, and Mozambique [55,56]. This suggests the need to improve the condition of the floors in the houses in the study area. However, low-income households may not be able to afford a permanent floor. In this case, they need a subsidy from the village office. This kind of subsidy, i.e., one for repairing or improving house conditions, is common in the study area but only reaches a few poor houses per year. One of the key barriers to prevent contamination and infection is handwashing [57]. However, it seems that people in the study area do not value handwashing as being important. Moreover, the mean value of the variable “respondent’s nails” was also low, indicating that people in the study area were less aware of the importance of personal hygiene. A previous study stated that “general hygiene practice was correlated with commitment to hygiene, indicating a strong association to psychosocial determinants” [12]. Future studies conducted in this area should then investigate the key psychosocial determinants of handwashing that can be targeted in health promotion activities.There were significant relationships between the six variables related to the surrounding environment, i.e., floor types, floor cleanliness, and presence of livestock, feces, garbage, and flies around. The presence of livestock and allowing them roam freely could be the reason why there are a lot of feces and flies around the house as well as dirty floors. A previous study indicated that fecal contamination in a house is associated with the presence of livestock [14,58,59]. Another study found that the presence of livestock is associated with diarrhea and malnutrition in children [60]. Additionally, considering the fact that East Sumba has one of the highest prevalences of stunting in Indonesia [32], corralling livestock separately from the house’s main areas should be conducted to minimize direct or indirect fecal contamination from livestock [14]. A previous study in Bangladesh found that corralling livestock improves the drinking water quality and reduces cases of diarrhea in children [61]. However, interventions related to livestock should also consider the cultural aspect of livestock in the Sumbanese culture. That is because livestock is a symbol of social status [62]. Moreover, some locals said that they prefer to keep their livestock as close as possible to the house to avoid livestock theft. All of these aspects should be considered in the intervention strategies. Furthermore, this study also emphasizes that culture should be taken into account in WASH programs in indigenous communities, i.e., they should avoid altering the local culture [53,63]. Furthermore, Figure 3 implies that there is not much difference regarding earthen floors with or without animal feces. This may explain the relationships between not having a permanent or earthen floor and those five variables. People, i.e., those who have an earthen floor, may keep their current unhygienic practices because they have become used to their homes looking dirty, as this may have been the case for a long period of time. The results imply that access to water is critical to having proper hygiene practices at home. A previous HWT study in this area stated that “households who need more time to collect water perceived lower levels of ability and self-regulation to operate HWT technologies” [36], which may apply to the context of hygiene practices. Some locals also said that they stopped using a toilet, i.e., they went back to open defecation, because sufficient water is unavailable. Moreover, the health effects of hygiene practices depend on having access to a sufficient water supply [3]. The lack of water in East Sumba is a result of complex environmental conditions. The mean annual rainfall in the study area is about 830 mm/year [30], which is below the mean annual rainfall in the country, i.e., about 2700 mm/year [64]. Moreover, the soil structure complicates people having access to groundwater [63]. The national and local authorities, i.e., village office and municipality, should then focus on delivering water to areas in need. Special attention should be given to the drought period in April–October to ensure that people still have an adequate amount of water to perform proper hygiene behaviors, e.g., delivering water by water trucks during the drought period. Building or subsidizing a rainwater harvesting tank in which to store water during the rainy season could also be another solution. The significant and positive influence of wealth on hygiene conditions is consistent with the literature [37,38]. This could be because having more money allows people to purchase items and practice proper hygiene practices. This finding also indicates that low-income households are more vulnerable to contracting diseases due to improper hygiene conditions and practices. Considering the economic conditions of households in the study area, low-cost interventions should be the short-term priority, e.g., regularly cleaning the house of garbage and animal feces as well as cleaning the nails. Furthermore, high-cost interventions, e.g., water provision, a latrine, or a permanent-concrete house floor, can be conducted using a turn-based system using a subsidy from the village office, i.e., these high-cost interventions can be a long-term program. This study shows that by including the contextual determinants of hygiene conditions in the analysis, we could design more specific types of intervention. For example, we could urge the target group to engage in handwashing activities regularly, but without analyzing the contextual determinants, we would not have been able to know that one of the possible reasons for irregular handwashing is because there is not a sufficient amount of water, i.e., water needs to be provided first before the suggestion to increase handwashing is given. Analyzing these contextual determinants may help us to understand the “root causes” of health-related behaviors because without tackling the problem at the “root”, we may not be able to transform people’s unhealthy behaviors. This study also indicates that rural and less developed areas in Indonesia, i.e., where the majority of households have low income and education levels, need a special focus on hygiene interventions. Multiple stakeholders at the district level, e.g., district health agency, social agency, NGOs, etc., should work together to tackle this issue. Health posts in the sub-district level and in the pre- and postnatal healthcare information (called “Posyandu” in Bahasa) in the village level can be the hygiene promotion center. Moreover, based on our discussion with some of the stakeholders in that area, support from higher-level institutions, e.g., agencies at the provincial or national level, is urgently needed, especially regarding high-cost interventions, e.g., water supply [63]. This study has some limitations. First, even though enumerator training and the pilot study were conducted before the real data collection period, there is the potential for inconsistency regarding the assessment of the hygiene variables among the different enumerators [65]. Second, we relied on self-reported answers o regarding water treatment which may be a source of bias, as found in another HWT in Cambodia [66]. Third, the water quality analyses were only conducted at a single time point and only presence-absence (P/A) tests were conducted. Thus, temporal variation in the water quality was not recorded, and P/A test may not fully capture the actual water quality. Fourth, seasonal hygiene conditions, i.e., those in the rainy and dry seasons, should also be conducted since access to water is abundant in the rainy season but limited in the dry season. Fifth, future studies should also investigate the relationship between general household conditions and water-related diseases in children, e.g., diarrhea and malnutrition. Sixth, the hygiene variables used in this study may not contribute equally to the general household hygiene. Future studies should investigate this so that we can have a better proximation of general household hygiene conditions. Moreover, a behavioral study should be conducted to understand the underlying perceptions or psychological factors behind current hygiene conditions, including the investigation of other contextual factors that may influence hygiene conditions, e.g., policy, regulations, WASH-related institutional performance, etc. These findings could help to design the behavioral change interventions that are needed in the study area. This study examines the general household hygiene conditions in less developed rural areas in Indonesia. In general, the hygiene conditions were at a “moderate” level, indicating that people in the study area were at risk for poor hygiene-related diseases. Two important contamination pathways were the surrounding environment, i.e., a non-permanent floor and the presence of garbage, and personal hygiene., i.e., handwashing. There is a need to improve non-permanent floors to permanent floors, but a subsidy is needed for low-income households. The presence of livestock roaming freely in the yard is one of the reasons for the dirty surrounding environment. However, the cultural aspect of livestock ownership should be taken into account in intervention strategies. Intervention should also target handwashing behaviors and nail cleaning to prevent the spread of contaminants through the fingers. Having access to water positively influences hygiene conditions. This is a serious challenge in this area because people regularly face water scarcity 7–8 months a year. Wealth also significantly influences household hygiene conditions, indicating that poor households should be the main target group WASH interventions. Low-cost interventions can be conducted in terms of immediate interventions, e.g., regularly cleaning the house of garbage and cleaning one’s nails. Finally, this study can be seen as a snapshot of the household hygiene conditions in less developed and rural areas in Indonesia and as a trigger to improve household health in those areas. This research was funded by Indonesia Endowment Fund for Education (LPDP) for the author’s PhD studies at the Delft University of Technology. The Delft University of Technology provided funds to support the field logistics and research activities.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Delft University of Technology (No. 805 and date of approval: 19 July 2019).Informed consent was obtained from all subjects involved in the study.The data presented in this study are openly available in the repository Narcis at https://doi.org/10.4121/uuid:9fb7a522-06a1-457c-9c69-3b2529c38d92.The author thanks the respondents for their participation in the study as well as all of the enumerators: Antonia Djarawula, Primus Lede, Selia Nangi, Jems Ndapangadung, Dominggus Wulang, and Yemima Amah, and Yayasan Anugerah Anak Sumba for the assistance during the data collection period. The author is grateful for the support and supervision from Saket Pande and Luuk Rietveld during his research at the Delft University of Technology.The author declares no conflict of interest.Location of the study area in the district of East Sumba.The conceptual model of four clusters of the determinants of household hygiene.Examples of observable household hygiene conditions in the study area: (a) drinking water is stored in a jerry can or a bucket; (b) earthen or not permanent kitchen floor and chicken inside the kitchen; (c) dirty bamboo kitchen floor with chicken’s feces; (d) food is stored inside of the food cover.The numbers indicate mean values of hygiene variables (n = 278, min = 0, max = 2), with the standard deviation bars.Household hygiene-related variables used in this study.* ”2” indicates binary variables, i.e., “0” and “1”, and “3” means that the variable is measured in three levels, i.e., 0, 1, and 2.Relationship between 15 hygiene-related variables. Green cells indicate significant relationships (p ≤ 0.0035 after the Bonferroni adjustment).Regression analysis of determinants of household hygiene conditions.n = 278; Adj. R2 = 0.111.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11065.txt

@@ -0,0 +1 @@
+Co-first author: Rodrigo M. Dias and Íbis A. P. Moraes should be considered joint first author (This article belongs to the Special Issue Physical, Psychological, and Social Health in Youth through Exercise and Healthy Behaviors).Background: Sedentary lifestyles are increasingly common amongst children, and insufficient physical activity is a global epidemic estimated to contribute to future incapacities and potential deaths. Objective: We aimed to increase the amount of evidence concerning the effect of chronic exposure to exercise on heart rate variability in children and adolescents affected by obesity. Methods: A systematic review commenced following the PRISMA guidelines developed by Web of Science, Virtual Health Library, PubMed, Cochrane Library, Embase, Ovid, Medline Complete, and Scopus using keywords obtained from the Descriptors in Health Sciences and Medical Subject Headlines (MeSH) terms. We considered (1) Population: Pediatric individuals affected by obesity; (2) Intervention: Exercise; (3) Control: Pre-intervention and sedentary; (4) Outcomes: Clearly presented primary parameters; and (5) Studies: Clinical trials, case controls, case reports, and case series. Results: 11 articles were involved and predominantly included procedures observed during approximately 12 weeks with a distribution of three sessions per week, each session being 30–60 min of aerobic exercise; additionally, the exercise grades were typically completed at a percentage of subjects’ maximum heart rates. The meta-analyses displayed a significant effect on the domains of time (R-R interval, SDNN, rMSSD), frequency (HF ms2, HF (n.u.), LF/HF), and the non-linear index (SD1). Conclusions: Chronic exposure to exercise influences heart rate variability in children and adolescents affected by obesity by elevating the variability and parasympathetic activity and improving the sympathetic-vagal balance. Exercises should be recommended for the improvement of cardiac autonomic modulation to prevent the likelihood of further chronic diseases.Obesity is problematic and affects a large section of the global population [1]. The World Health Organization (W.H.O.) states that amongst children and adolescents, obesity has increased tenfold in the last four decades, and by 2022, there will be more children and adolescents affected by obesity than presenting with malnutrition [2]. Childhood obesity has been linked to the early onset of various chronic conditions, such as type 2 diabetes and systemic arterial hypertension [3]. Additionally, research demonstrates that obesity is often attended by earlier dysfunction of the autonomic nervous system [4]. Adverse obesity-related health consequences have been associated with possible cognitive decline and increased body mass index, while central adiposity has been shown to be correlated with performance impairments under task conditions that require executive control [5]. Young individuals affected by obesity suffer from a decline in cardiopulmonary function, poor exercise tolerance, and low self-esteem [6].Although treatment choices include medication, diet changes, intensive behavioral modification, and behavioral therapy, the use of exercise provides an interestingly different impact [7,8]. Sedentary lifestyles are increasing amongst children, and insufficient physical activity is a global epidemic estimated to contribute to future incapacities and potential deaths [9,10].Gonzáles–Ruiz’s [11] systematic review of the use of exercise to reduce fat in those with pediatric obesity supports the current approval for physical, mainly aerobic exercise as an effective intervention against non-alcoholic fatty liver disease progression by targeting hepatic lipid composition and visceral and subcutaneous adipose tissue. Likewise, the impact of exercise causes enhancements in autonomic modulation [12,13,14], and this can affect obesity [15,16]. Although there is an existing association between autonomic modulation, exercise, and obesity, the application of exercise as a treatment method for probable cardiac autonomic dysfunction generated by obesity in pediatric individuals necessitates further study [17].One of the most effective ways to judge these autonomic dysfunctions is via the study of heart rate variability (HRV), an important predictor of cardiovascular health [18,19] above being a simple, reliable, and non-invasive way of scrutinizing autonomic function [20]. Similarly, cardiac autonomic dysfunction, as indicated by a reduced HRV, can promote inferior cardiovascular fitness outcomes and diminished parasympathetic function [3,21,22].A systematic review of cross-sectional studies was performed by Oliveira et al. [23] in an attempt to determine whether cardiac autonomic function is linked to cardiorespiratory fitness and physical activity in children and adolescents. They demonstrated that physical activity was positively associated with parasympathetic activity but that most studies had a small amount of evidence. Consequently, there is a need for studies that elevate the level of knowledge concerning the effect of chronic exposure to exercise on the HRV of pediatric individuals affected by obesity. According to Farrell and Turgeon [24], after 30 days of exposure to exercise, there are chronic physiological adaptations. Based on the above-mentioned information, the study objective here was to evaluate the influence of chronic exposure to exercise on the HRV of children and adolescents affected by obesity and to specify exercise as an enhancer for combating potential autonomic dysfunctions caused by obesity.The review was conducted according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [25,26,27,28,29,30,31]. The protocol for this review was registered in PROSPERO under registration number CRD42020178328.For this systematic review, studies were investigated in the databases Web of Science (WOS), Virtual Health Library (Biblioteca Virtual em Saúde-BVS), PubMed, Cochrane Library, Embase (Elsevier), Ovid, Medline Complete (EBSCO), and Scopus (Elsevier) during January 2020 using keywords obtained from the Descriptors in Health Sciences (DeCS) of the Virtual Health Library and MeSH. These database searches were organized according to the PICOS model (Population, Intervention, Control, Outcome, Study design) [32]. The search was performed using the terms “Heart Rate Variability AND Exercise AND Obesity AND (Child OR Children OR Adolescent OR Adolescents)” and “Heart Rate Variability AND Exercise AND (Pediatric Obesity OR Childhood Obesity)” (Table S1—Supplementary Materials).These articles were selected according to the following inclusion criteria: (1) Population: Pediatric individuals affected by obesity; (2) Intervention: Exercise; (3) Control: Pre-Intervention; (4) Outcomes: Evidently presented primary parameters; and (5) Studies: Clinical trials, case controls, case reports, and case series. The obesity criteria (relative to normative data for age, gender, and ethnicity) were represented by the ≥85th percentile of the updated growth norm of each different country and/or the z-BMI score from 2 to 3 [33,34]. There were no restrictions of dates in the inclusion of publications.Accordingly, the following items were omitted: (1) abstracts and expanded abstracts; (2) complete articles that were not written in English, Portuguese, or Spanish; and (3) articles that did not refer to both exercise and pediatric obesity. In cases of discrepancy, a third author was consulted. Duplicate records and studies that were unrelated to the planned objectives were excluded.The data were extracted from the included studies via an electronic spreadsheet. Data were collected on the study design, sample number, intervention time, type of exercise, and device used to collect HRV, all following the PICOS structure. Additionally, numerical data for further meta-analyses were extracted to consider the means and standard deviations presented by the studies. To extract data from studies that presented results in the form of graphs, WebPlotDigitizer [35] was used to find the central and dispersion measures.The parameters of interest originated from the three possibilities of comparison and outcome: (1) Post-Intervention vs. Pre-Intervention; (2) Obese with exercise vs. Obese without exercise; and (3) Obese with exercise vs. Not obese with exercise. Next were the variables related to HR and HRV in the time domains: (1) RR interval: interval between consecutive heartbeats; (2) SDNN: standard deviation of all normal RR intervals recorded in a time interval, expressed in ms; and (3) rMSSD: the root-mean-square of differences between adjacent normal RR intervals in a time interval, expressed in ms. Additionally, we defined frequency domains: (4) HF: High Frequency, extending from 0.15 Hz to 0.40 Hz, expressed in milliseconds squared and normalized units; (5) LF/HF ratio: reflects the sympathetic-vagal balance]; and the non-linear metrics; (6) SD1: represents the dispersion of points perpendicular to the line of identity, short term variability of continuous RR intervals; and (7) SD2: long-term variability of continuous RR intervals. According to Vanderlei et al. [18], these are the main parameters analyzed, and for that reason, these parameters of interest were chosen. The secondary parameters were body mass index (kg/m2) (BMI) and body fat percentage (Fat%).The PEDro scale [36,37] was obligatory in evaluating the quality of the evidence of the studies, as it is the most often used scale in the rehabilitation area. This scale was developed by the Physiotherapy Evidence Database to evaluate experimental studies that can attain an overall score of 10 points. These criteria are confined in the Delphi list and are useful in investigating items in systematic reviews consistent with their proposed methodology, considering the studies as presenting the following amounts of evidence: “excellent” being 9–10, “good” being 6–8, “reasonable” being 4–5, and “poor” being less than 4.Review Manager Software 5.3 was essential in completing the meta-analysis calculations [38]. The data used were those expressed as the mean and standard deviation. If such data were presented as median and interquartile ranges, their means and standard deviations were estimated according to the method by McGrath et al. [39]. Additionally, dispersion measures of standard error were transformed into the standard deviation.To study the inverse variances, the means and standard deviations of the results of each study were required. The fixed effects of the treatments were studied; hitherto, in the case of significant heterogeneity between studies, random effects were analyzed. Heterogeneity was considered using the Q2 and I2 tests. Studies that contained an excess of one group that used exercise as a treatment were combined into one group using the algorithm described by Cochrane [40].The literature research (5–30 January 2020) consisted of a total of 1181 potentially relevant articles, of which 283 were duplicates. That left 787 viable studies, of which 111 turned out to be relevant studies. After scrutinizing the titles, 71 studies were excluded. Twenty-four more studies were excluded after investigating the abstracts, and 12 additional studies were eliminated after analyzing the full texts. In the succeeding step, the three filters were applied, resulting in 11 articles being included in this review. The studies’ flowchart and inclusion strategy are exemplified in Figure 1. Amongst the 11 studies [3,5,6,17,21,41,42,43,44,45], a total of 398 individuals were included as a sample, comprising 176 boys and 222 girls aged 5 to 18 years, as illustrated in Table 1.The study methods are demonstrated in Table 1. Amongst the procedures, the ensuing factors were predominant: lasting approximately 12 weeks, distributed between three sessions per week, and 30 to 60 min per session. The principal type of exercise enforced was aerobic exercise, and the exercise grade was usually performed at a percentage (%) of the maximum HR. HRV collection was usually completed before and after the intervention by means of a Polar® heart rate monitor.With regard to the main outcomes achieved in the studies, significant changes in HRV were observed after exercise intervention. This factor can be perceived after evaluating the results of the HR meta-analysis and HRV indices in the time domain, frequency domain, and non-linear metrics.The classification of studies exhibited a majority to be reasonable to poor, as illustrated in Table 1.It was plausible to perform only meta-analyses of the HRV index “Post-Intervention vs. Pre-Intervention” (1) because of the ways of presenting the data of the selected articles. Yet since it is imperative to highlight in this paper, when possible, the variables also related to the other two possibilities of comparison and outcomes discussed, “Obese with exercise vs. Obese without exercise” (2) and “Obese with exercise vs. Not obese with exercise” (3).Figure 2A determines that exercise had an important effect on the HR index of individuals both post-intervention and pre-intervention, but this presented significant heterogeneity of data.Figure 2B establishes that the mean RR interval caused a significant effect (Z = 2.38 (p = 0.02)) without significant heterogeneity. For the HRV indices related to the time domain, the SDNN index in Figure 2C displayed a significant effect when comparing post-intervention and pre-intervention (Z = 3.37 (p = 0.0007)) without significant heterogeneity. In Figure 2D, the HRV rMSSD index was considered for both post-intervention and pre-intervention and established a significant effect (Z = 3.63 (p = 0.0003)) while not showing significant heterogeneity.For the HRV indices related to the frequency domain, the HF index (ms) presented a significant effect post-intervention as opposed to pre-intervention (Z = 2.38 (p = 0.02)) and also lacked significant heterogeneity (Figure 3A). The HF (n.u.) index and post-intervention vs. pre-intervention both established an important effect (Z = 3.75 (p = 0.0002)) but did not display significant heterogeneity (Figure 3B). When investigating the LF/HF ratios, in both post-intervention and pre-intervention there was an important effect (Z = 3.75 (p = 0.0002)) with an absence of significant heterogeneity (Figure 3C).The non-linear metric SD1 was equated for both post-intervention and pre-intervention, with both indicating a significant effect (Z = 2.92 (p = 0.004)) without significant heterogeneity (Figure 4A). SD2 index had a negative effect (Z = 3.80 (p = 0.0001)) for both post-intervention and pre-intervention without significant heterogeneity (Figure 4B).The secondary parameters were evaluated, and the forest plot was illustrated in Figure 5. In Figure 5A, the exercise offered a significant effect (Z = 2.80 (p = 0.005)) in the reduction in BMI (kg/m2) without significant heterogeneity. In the same way, Fat% demonstrated a significant change (Figure 5B) (Z = 2.84 (p = 0.005)) but not a significant heterogeneity (I2 = 87%; p < 0.00001)).The study objective was to identify the effect of chronic exposure to exercise on HRV in children and adolescents affected by obesity. Even though the classification of studies exhibited a majority to be reasonable to poor, the same tendencies in the results were found, in agreement with previous findings regarding the time domains of HRV. We revealed that children and adolescents affected by obesity in the intervention group have a reduced mean HR post-exercise compared with pre-intervention and that this is related to a greater mean RR interval. Moreover, children and adolescents affected by obesity in the intervention group established greater variability post-exercise within the time-series of heartbeats, as revealed by the SDNN index, and higher parasympathetic activity, as expressed by rMSSD.Similarly, in the frequency domain, when examining the HF ms2 in (n.u.), a pattern of increase was revealed in post-exercise intervention as well as a reduction in the LF/HF index post-exercise intervention when equated to pre-intervention, indicating increased parasympathetic activity and improved sympathetic-vagal balance, respectively. Still, no significant results were observed while considering the LF ms2, even if in normalized units, a reduction was identified post-exercise intervention. It is recognized that LF reflects both sympathetic and vagal influence and has been consistent with baroreflex sensitivity [47]. These results were undetected in LF (n.u.) and HF (n.u.) when related to pre-intervention.Finally, in the chaotic (or non-linear) domain, SD1 showed a pattern of increased post-exercise intervention in the exercise group when compared with pre-intervention, therefore representing an increase in parasympathetic activity since, physiologically, the transverse axis (SD1) is a measurement of short-term changes in the RR intervals, which are considered an indicator of parasympathetic activity. A contradictory pattern (decrease) was revealed in SD2. The physiological rationale of the longitudinal axis (SD2) is not as noticeable, yet it is thought that it reflects the long-term changes in RR intervals, and these assumptions are open to different explanations, such as sympathetic input with parasympathetic influences [48].Even supposing that the effects of chronic exposure to exercise on sympathetic activity remains questionable, the exercise demonstrably promoted greater parasympathetic cardiac activation. Perhaps this is the crucial indicator of positive exercise adaptation [5], demonstrated by the rMSSD, HF, and SD1 indices. In this manner, chronic exposure to exercise could impart improved cardiovascular physiological health [3,41], thereby reducing autonomic dysfunction [17], improving cardiac electrical stability, protecting against experimentally induced myocardial infarction [42], and elevating arterial baroreflex sensitivity and cardiorespiratory functional capacity [21].However, intervention programs proposing activities with lower intensity did not have such advantageous effects. This is an important issue, as exercise intensity appears to be a determinant in HRV response [3,6,22,43]. The physical activity programs of the studies included in this review enforced many types of exercise. Other than aerobic training [3,5,6,21,41,42,43,44], the studies involved sports, commonly including soccer practice [45,46], and a solitary research study that assessed resistance training, conducted by Farinatti et al. [17]. Notably, it was revealed that the training time—not only the session time but also the duration time in weeks—is foremost in maintaining the useful effect of exercise, as once the training periods ceased, the indices tended to slowly return to previous values [42].Other confounders, e.g., dietary control, appear to have not influenced the evaluated studies; in most studies, while no lifestyle or any form of nutritional counseling was provided, positive results were attained. Two studies proposed dietary counseling, yet there was no attempt to reduce energy intake [3,6]; additionally, two studies associated exercise intervention with a dietary restriction [21,44]. If obesity produces an imbalance between dietary intake and energy expenditure, it appears that the latter found positive changes that were enhanced by calorific restriction. Despite this, a lessening in BMI was achieved in almost all studies, and a decrease in fat percentage in all, so we can conclude that the physical activity program completed in the studies cut the subjects’ obesity status. Hence, this may also be a feature that influenced changes in HRV.It is important to remark that the autonomic nervous system directs voluntary and involuntary physiological processes, such as digestion, blood pressure, hormonal regulation, energy metabolism, and heart rate, and is accordingly considered an important regulator of homeostasis. It innervates fat depots, which are associated with catecholamine production. Thus, the biological mechanisms could involve adipocytokines secreted by fat cells [49]. According to these statements, studies have revealed that body composition measures are negatively associated with HRV parameters and indicators of parasympathetic activity [50]. RMSSD (parasympathetic activity) was negatively correlated with fat mass [51] and, together with weight loss [52], was linked to a decrease in HR and an increase in HRV (as indicated by SDNN). Other than that, a calorific restriction has been confirmed as an intervention that may reverse the autonomic changes [53].In the study of Veijalainen et al. [54], physical activity was related to better cardiac autonomic nervous system function in children independent of gender, adiposity, and the clustering of cardiometabolic risk factors. The authors comment that hemodynamical regulation of the human body is complex and involves neural and humoral mechanisms; hence, one of the explanations for the observed associations of lower physical activity with lower HRV could be that it decreases blood volume and left ventricular stroke volume to result in an increased heart rate on account of increased sympathetic activity.Aerobic exercise training has been stated as an integral component of interventions to reduce obesity and related co-morbidities in children and adolescents [3,6]. In addition, physical activity and dietary interventions are beneficial for glucose metabolism, skeletal muscle function, bone stability, psychological well-being, and physiological organ functions [21]. In contrast, there is evidence that lifestyle interventions that improve weight status and metabolic risk may improve autonomic dysfunction in children affected by obesity [17,44]. Despite the current evidence connecting exercise training to enhanced autonomic nervous system activity in children, the exact amount of exercise required for optimal adaptation is indeterminate.As HRV declines with age [42], it is easy to speculate that the promotion of physical activity should be recommended, as it can improve autonomic cardiac modulation. Such activity would aid in preventing chronic diseases, such as diabetes, cardiac events, stroke, and so forth, to accordingly lead to a better quality of life and extended life expectancy over time. Above and beyond these, this practice would affect reductions in public health expenditure through the reduction in the prevalence of chronic diseases.From a methodological perspective, we did not address the impact of the method of HRV capture, but we can affirm that most studies used the Polar heart rate monitor to capture RR intervals [3,6,8,17,21,40,41,42] and the Kubios HRV® software to compute the indices [3,6,44,45,46]. Additionally, the times and sample rates of recording were not standardized in the studies. This is a key fact, considering that data standardization in one study included in the meta-analysis used a different method of data conversion; in Hamila et al. [38], RMSSD, LF (n.u.), and HF (n.u.) were logarithmically transformed to normalize data. Another important point to be revealed is that the lack of knowledge about the HRV indices can generate errors in the interpretation of the data, as realized in the study of Chen et al. [5], in that the authors state that there was an increase in HF (n.u.) and LF (n.u.) after exercise. However, since these indices represent the quantity of each one (LF or HF) concerning the whole (total power), as one index increases, the other must inevitably decrease.One of the major restrictions of the current analysis is the fact that several studies had to be excluded as a result of insufficient reporting of descriptive data. Accordingly, we encourage researchers to provide descriptive statistics for HRV to further augment the existing body of scientific research. Additionally, the sample sizes of the studies were small, and there was heterogeneity in the study samples, the numbers of participants, groups, and grades of exercise intensity, making it problematic to extrapolate the data. Finally, the non-differentiation between ethnicities could be an influencing factor on the data evaluated in the studies. Therefore, more robust studies are required to establish whether these effects can be replicated.We conclude that chronic exposure to exercise appears to influence HRV in children and adolescents affected by obesity by increasing variability (SDNN index) and parasympathetic activity (expressed by rMSSD, HF, and SD1), cultivating the sympathetic-vagal balance demonstrated by a decrease in LF/HF. Nonetheless, the evidence is reasonable to poor and should be reassessed with more robust clinical studies in future research. Still, exercise can be recommended for the improvement of cardiac autonomic modulation to decrease the likelihood of further appearances of chronic diseases.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111065/s1, Table S1: Search Strategy.Conceptualization, R.M.D., Í.A.P.M., M.T.A.P.D., D.C.G.L.F. and T.D.S.; methodology, R.M.D. and Í.A.P.M.; formal analysis, R.M.D., Í.A.P.M. and T.D.S.; data curation, R.M.D., Í.A.P.M. and T.D.S.; writing—original draft preparation, R.M.D., Í.A.P.M., A.M.G.G.F., A.C.S., V.B., M.F., P.M.M. and T.D.S.; writing—review and editing, A.C.S., V.B., M.F., P.M.M., L.C.A., C.B.M.M., and D.M.G.; supervision, L.C.A., C.B.M.M., D.M.G. and T.D.S.; funding acquisition, L.C.A., C.B.M.M. and T.D.S. All authors have read and agreed to the published version of the manuscript.This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil: Finance Code 001, by Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil (CNPq) process number 442456/2016-6 and by Convênio Sesacre—FMABC (Faculdade de Medicina do ABC) under process number 007/2015.Not applicable.Not applicable.We would like to thank the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Brazil (CAPES); Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil (CNPq); and Convênio Sesacre—FMABC (Faculdade de Medicina do ABC).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.Flowchart.Forest Plot—Time domain indices. Legend—(A): Mean Heart Rate mean; (B): R-R interval; (C): SDNN; (D): rMSSD.Forest Plot—Frequency Domain indices. Legend—(A): HF ms2; (B): HF n.u.; (C): LF/HF ratio.Forest Plot—Non-linear indices. Legend—(A): SD1; (B): SD2.Forest Plot—Secondary parameters. Legend—(A): BMI; (B): Fat%.Characteristics of the studies.Legend—SD: Study Design; RCT: randomized controlled trial; CT: Clinical Trial; wk: weeks; min: minutes; d: days; BMI: Body Mass Index; HIT: High-Intensity Interval Training; LIT: Low-Impact Interval Training.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11066.txt

@@ -0,0 +1 @@
+Parents of children with a neurodevelopmental disorder (NDD) report higher levels of distress compared to those of typically developing children. Distress levels may be heightened by the restrictions associated with the COVID-19 pandemic. However, it is unclear whether distress levels of parents varied by the diagnosis of neurodevelopmental disorder in children during the COVID-19 pandemic. This study aims to investigate whether parental distress was influenced by the type of NDD. Participants were from Australia (N = 196) and Italy (N = 200); the parents of children aged 3–18 were invited to complete an online self-reported survey which included the 6-item Kessler Psychological Distress Scale (K6) to determine parental distress. The results show that intellectual or learning disorder (ILD) is a major contributor to parental distress compared to other NDDs in both Australia and Italy. Moreover, the worsening of symptomatic changes in children with NDDs was significantly associated with parental distress. The differences between the two countries in terms of the pandemic impact, however, were not statistically significant. The results suggest that intervention strategies need to be tailored for individual clinical information and factor in the society’s stringency level of anti-contagion policies to improve parental wellbeing.The COVID-19 pandemic has led to changes in many different aspects of life around the world during the year 2020. Its adverse consequences have not completely subsided in 2021 due to subsequent waves of infections and variants that lead to repeated anti-contagion measures including social distancing, home quarantine, and the closure of schools. Although these measures are effective in curbing the spread of the virus, the disruption of services, social isolation, and financial insecurities have caused a huge psychological impact worldwide [1]. In particular, a high prevalence of psychological symptoms, such as anxiety and depression, has been reported in parents of young children and adolescents [2,3].During home quarantine, family daily routines unexpectedly changed, which changed the role of parents. After the closure of schools, most parents had to combine working from home and childcare, while simultaneously homeschooling their children. Moreover, the economic implications of the COVID-19 have also been shown to negatively impact families [4,5]. The disruption of daily routines may be even more difficult for children with neurodevelopmental disorders (NDDs), as established routines are a crucial strategy to promote stability [6]. In addition, the closure of educational or other therapeutic placements meant that children with NDDs lost the professional support they need [7,8].In general, parents of children with NDDs experience higher levels of parental distress compared to those of typically developing children [9,10]. In addition, they are more likely to experience mental health problems such as anxiety, depression and reduced sleep quality [11,12,13]. Findings from existing research suggest factors such as social support, children’s behavioural problems, economic and social status and parenting stress are key factors that contribute to psychological distress in parents of children with NDDs. Studies have also suggested that parents of children with autism spectrum disorder (ASD) report higher levels of parental distress compared to other NDDs [14,15].Based on the above research, it can be expected that the COVID-19 pandemic and related restrictions to routine are likely to have a negative impact on parents of children with NDDs. Indeed, emerging evidence supports this notion; for example, data from our Australian survey of parents of children with NDDs found that COVID-19 had an adverse impact on the wellbeing of three-quarters of parents while over 40% reported worsening of their pre-existing mental health issues [16]. This appears to be compounded by the fear of infection and the impact of lockdowns and consequent social isolation as well as adverse economic consequences. Similarly, a study in Italy using the same survey found that 58.5% of respondents found that their child’s overall health and wellbeing had been impacted by the pandemic, while 47.7% stated their own wellbeing had been affected [17].Despite the relatively few studies investigating the impact of the COVID-19 pandemic on the wellbeing of parents of children with NDDs, it is clear that these parents need support. Clarifying the clinical and sociodemographic contributors to parental psychological wellbeing may facilitate the implementation of mental health for families with children with NDDs [18]. To further explore the relationship between parental distress and children with NDDs, this study aimed to address. The study aims to address two primary research questions: (1) Which diagnosis of NDD was more likely to correlate with higher levels of parental distress during the COVID-19 pandemic? (2) Which diagnosis-specific symptomatic change was more likely to correlate with parental distress levels during the COVID-19 pandemic? We believe that the answers to these questions might pave the way for targeted prevention or intervention strategies for families with special needs. Further, we aimed to investigate how the magnitude of the disease burden of COVID-19 in society as a whole might impact parental distress differently. From January to June 2020, Australia was in the top 10% of countries with the lowest infection and mortality rates related to COVID-19, while Italy was one of the top 10% countries with the highest infection and mortality rates related to COVID-19 [19]. The disease burden has led to variable levels of stringency index values (an indicator for how strictly non-pharmaceutical intervention measures are implemented) across different countries [20].A cross-sectional design was used to investigate the impact of the COVID-19 pandemic on Italian and Australian families. Parents and caregivers of a child aged 3–18 years with an NDD were asked to participate in the surveys. The survey was promoted via disability service providers and support groups by emailing parents on their mailing lists with information about the study and the link to the survey, or by posting an advertisement with the link on social media. Interested parents could read the Participant Information Statement online and complete the survey via Research Electronic Data Capture (REDCap), a secure web-based survey tool [21,22]. Consent to participate was considered implied if parents elected to complete the survey. The survey was open for a period of approximately 6 weeks during May and June 2020.The survey consisted of questions about the participants’ socio-demographic characteristics and the relevant diagnoses the child had received, along with a rating of symptom change since the start of the pandemic. A 5-point scale with 1 = “symptoms much improved”, 2 = “symptoms somewhat improved”, 3 = “symptoms the same”, 4 = “symptoms somewhat worse” and 5 = “symptoms much worse” was used. We combined responses from 4 = “symptoms somewhat worse” and 5 = “symptoms much worse” into one group indicating worsening of the child’s condition. In addition, there were questions relating to: general wellbeing (perceived impact of the pandemic across the family), family health (employment, finances, food and housing), home-based learning (challenges in homeschooling) and child behaviors (including emotional responses and use of technology and devices). The above questions were rated on a 5-point Likert scale ranging from “strongly agree” to “strongly disagree”. Similarly, the two responses at the impacted end of the scale—”strongly agree” and “somewhat agree” were combined together to represent a negative impact felt by the respondents. Parental distress was assessed using the Kessler Psychological Distress Scale (K6), a robust measure of psychological distress for adult populations [23]. The K6 consists of 6 questions that ask about the frequency that participants felt sad, nervous, restless, hopeless, that everything was an effort, and worthless during the past 30 days.We first compiled the results of the survey for Italy and Australia separately to allow a direct comparison between the two countries. We compared the distributions of demographic and clinical variables between the two countries by carrying out chi-squared tests. Regression models were constructed to evaluate our hypotheses to determine which NDD diagnosis in the child can predict parental distress in Australia and Italy, as well as in the total population by merging the Australia and Italy datasets. Furthermore, we used semi-partial correlation coefficients to infer the proportion of variance in parental distress levels explained by each NDD to compare the relative contribution of each diagnosis. The two-sided alpha value = 0.05 was used to determine the relationship between the independent variable and the dependent variable.To prove that the two countries had different stringency index levels in 2020, we extracted the data from “Our World in Data—COVID-19 section” (https://ourworldindata.org/coronavirus, accessed on the 1 August 2020) and compared the stringency index values between Italy and Australia, and confirmed that the stringency index values statistically significantly varied between these two countries using a mixed model analysis (p < 0.0001). Therefore, we also compared the impacts on parental distress between Australia and Italy, the two countries that represent two groups of countries affected by the COVID-19 pandemic with different levels of stringency index values.Next, we carried out interaction analyses to compare the impact of the child’s diagnosis on parental distress between Australia and Italy. In addition, interaction analyses were also used to investigate whether the impact of diagnosis-specific symptomatic changes on parental distress varied by country. In the linear regression model, the outcome variable (i.e., Kessler-6 scores) was regressed against the primary predictor (i.e., either diagnosis or diagnosis-specific symptomatic change), country, and the product of the primary predictor and the country, adjusting for the parental age and educational level. All analyses were conducted in R version 4.00(R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria) [24].A total of 200 respondents participated in the survey in Italy, while 296 respondents participated in Australia. As shown in Table 1, the caregivers participating in the study were mostly between the ages of 40 and 49 in both Italy (53.0%) and Australia (54.8%). The mean age of the participants’ children was 10.7 and 10.1, in Italy and Australia, respectively, with the majority of the children male in both Italy (81.5%) and Australia (65.9%). In Australia, a higher percentage of caregivers had higher education compared to Italy (68.8% vs. 19.5%).Table 2 provides an overview of the NDD groups. In Italy, the most common NDDs were Tourette’s syndrome (TS) (54.5%), Autism Spectrum Disorder (ASD) (51.0%) and Intellectual or Learning Disorder (ILD) (37.5%). In Australia, the most common NDDs were ASD (61.3%), ADHD (40.7%) and Anxiety Disorder (33.7%). Parents were also asked whether there were any symptomatic changes in their children during the pandemic. Although significant differences in the symptomatic changes were observed between Italy and Australia across all NDD groups, at least half of the respondents in both Italy and Australia experienced worsening of the symptoms in children with ADHD (50.0% vs. 58.7%), Anxiety Disorder (50.0% vs. 68.0%), OCD (51.4% vs. 51.6%) and TS (52.3% vs. 64.3%). In Australia, children with ASD were perceived to have increased symptoms (58.0%) compared to children in Italy (37.3%). Less than half of the respondents in both Italy and Australia experienced worsening of the symptoms in children with ILD (42.7% vs. 37.5%).The K6 scores of the parents are presented in Table 3. On average, parents in Italy showed significantly higher distress levels compared to parents in Australia. For example, 36.0% of the respondents reported being nervous most or all of the time in Italy, whereas only 20.3% felt the same way in Australia. Similarly, 26.0% of parents in Italy reported that most or all of the time they felt so depressed that nothing could cheer them, while only 9.1% felt the same way in Australia. However, when asked “how often did you feel that everything was an effort?”, a greater percentage felt that way most or all the time in Australia, 25.7%, compared to only 20.0% of parents in Italy.The following results can be found in the Supplementary Materials. More than half of the participants in both Italy and Australia somewhat or strongly agreed that their child’s overall health and wellbeing had been impacted by the pandemic with 58.5% and 69.4%, respectively. However, just under half (45.5%) believed the pandemic had worsened preexisting health conditions for their child in Italy, with half of the respondents feeling the same way in Australia.With regards to support networks, caregivers in Australia were impacted more than those in Italy. For example, 77.4% of respondents felt their support networks had decreased compared to 53.0% in Italy. Similarly, 71.8% felt that the pandemic had disrupted caregivers’ support and services in Australia, whereas only 44.89% felt the same way in Italy. Almost three-quarters of the respondents in Italy and Australia, 74.3% and 72.2%, respectively, felt that the pandemic has significantly disrupted the allied health services their child accessed. Most of the respondents in Australia (97.1%) stated that their children took more medication than normal. In Italy, a smaller proportion (29.2%) of caregivers felt that their child’s ability to access specialists had been significantly impacted by the pandemic compared to a much larger proportion of caregivers in Australia (70.5%).With regards to contributors to overall family health, caregivers in Australia were more impacted than those in Italy with 92.9% stating that COVID had significantly disrupted their child’s routines compared to 69.0% in Italy. Similarly, 75.5% indicated that during the past two weeks, COVID restrictions had been stressful for their child compared to 47.0% in Italy. Moreover, when asked about balancing work with childcare and family responsibilities, the majority (80.9%) of caregivers in Australia reported it had been difficult, whereas less than half (40.0%), felt the same way in Italy. Nearly all the respondents in Italy were optimistic the COVID crisis would end soon (94.5%) with just under three-quarters (73.2%) feeling the same in Australia.The pandemic had a significant impact on the children’s education in both Italy and Australia, with respondents stating that COVID-19 had prevented their child from attending school with 90.3% and 91.1%, respectively. Although most of the respondents had adequate access to online resources, fewer than half in both Italy (47.5%) and Australia (44.4%) felt their child had adequate capacity to engage in home-based learning. Interestingly, the majority of caregivers in Italy (79.9%) felt that they had adequate capacity to support their child’s educational needs, whereas less than half (46.4%) felt the same way in Australia.A proportion of caregivers in Italy reported a reduction in sleep quality in their children (18.7%) compared to almost half (46.9%) of caregivers in Australia. In both Italy and Australia, more than half of the respondents reported that their child had become more easily annoyed since the start of the pandemic (Italy: 60.6% vs. Australia 65.2%). Moreover, the majority of the parents reported that their children spent more screen time and played significantly more video games since the outbreak (Italy: 58.7% vs. Australia 65.4%).The results from the linear regression analysis in Australia (Table 4) show the diagnosis of ILD in children is significantly positively correlated with parental distress after adjusting for the other NDDs, the child’s age, the caregivers’ age and education level. Interestingly, parents with children diagnosed with TS were significantly inversely correlated with parental distress levels during the pandemic after adjusting for the same confounders. Although there were no significant associations between NDDs and parental distress in Italy, children with ASD showed a positive correlation with parental distress. When merging the two datasets together, it became apparent that ILD could be a major contributor to parental distress during the pandemic (B = −1.2654, p = 0.038). However, children diagnosed with TS were inversely correlated with parental distress (B = 2.7612, p < 0.005). The proportion in the K6 score explained by each diagnosis of the child in each country is shown in Figure 1. The results indicate that, among all diagnoses, the largest contributor to parental stress in Australia was TS, while the largest contributor to parental stress in Italy was ASD.Figure 2 provides an overview of the interaction plots which illustrates how Italian and Australian parents experienced different levels of stress during the pandemic by the following child diagnoses: ASD, TS, ILD and OCD. The results confirm the earlier analyses that children diagnosed with TS showed lower parental distress levels compared to those without the diagnosis. There was no significant difference between Australia and Italy regarding the impact of NDDs on parental distress (Table 5).In Italy, higher parental distress levels were significantly associated with the worsening of the symptomatic changes in children with ILD, OCD and TS—which is also shown in Table 6. In Australia, only worsening of the symptoms in ASD or ILD was associated with increased parental distress levels, while worsening of symptoms in OCD and TS did not statistically significantly correlate with parental distress levels. These results imply that the impact of symptomatic changes in OCD or TS on parental distress might vary by country during the pandemic. However, none of the interaction effects between the diagnosis-specific symptomatic changes and the country on parental distress reached statistical significance levels (Table 7). Figure 3 shows the results of the effect of diagnosis-specific symptomatic change and its effect on parental distress.In this study, we investigated the predictors of parental distress pertaining to NDDs and diagnosis-specific symptomatic changes. Moreover, we aimed to compare the impact of COVID-19 factors between Italy and Australia.The results from this study provide important insights into the predictors of parental distress during the pandemic and suggest that ILD in children is a major contributor in both Australia and Italy. This is in accordance with two recent studies carried out in the UK which reported increased levels of mental health in caregivers of children with intellectual disabilities [25,26]. Children with ILD are more likely to require more face-to-face special educational support than other children and hence interruptions of such support during the pandemic may pose a greater challenge for parents compared with others. Our study did not show any significant results with regards to children diagnosed with ASD increasing parental distress during the pandemic. This is in accordance with a recent paper that evaluated parenting stress before and during the pandemic in children with NDDs and their families [27]. However, our finding is contrary to previous studies that reported increased distress levels of parents of children with ASD during the pandemic [28,29]. This result may be explained by the fact that our study only included parents of children with NDDs, whereas other studies compared parental distress levels of children with NDDs to those of typically developing children. Furthermore, the life routines of children with ASD, who have fewer needs for social activities than some non-ASD children, may be less likely to be affected by the pandemic compared with children without ASD. Therefore, social distancing could cause less stress among children with ASD than other non-ASD children although these children with ASD might perceive more stress when school-based services could become less accessible due to school closures. Taken together, the child’s diagnosis of ASD may not necessarily cause higher levels of distress compared with other non-ASD children with or without other NDDs.One unanticipated finding from this study was the diagnosis of TS in children appeared to be associated with lower levels of parental distress when compared to other NDDs. A possible explanation is that parents or caregivers of children with TS might receive decreased unpleasant attention in public because of lockdowns during the pandemic era. Furthermore, children with TS might have less need for face-to-face special educational support than children with some other NDDs, such as ASD and ILD. Therefore, compared with families of children with various types of NDDs, homeschooling and other anti-contagion measures seemed to yield a limited impact on parents of children with TS. Indeed, previous research has speculated some positive aspects of home quarantine [30].With respect to the symptomatic changes of children with NDDs, we found that the worsening of symptoms in children with ASD, ILD and TS were significantly correlated with increased levels of parental distress, which is consistent with recent findings [31,32] This could be explained by the impact of home quarantine disrupting children’s routines, given that most parents in both Italy and Australia stated that COVID significantly disrupted their child’s routines. It should be noted, however, that the levels varied between Australia and Italy. For example, in Australia, we observed high levels of parental distress in parents of children with OCD and TS irrespective of symptomatic changes. Although, it is unclear whether the difference in the impact of symptomatic changes associated with these two NDDs between Italy and Australia could be attributable to the level of stringency concerning anti-contagion measures at the country level.Indeed, our study found many differences between Italy and Australia with regards to the impact of the pandemic. On average, parents in Italy showed higher distress levels compared to Australia based on the K6 score, while parents in Australia self-reported a higher impact on aspects such as types of support and the worsening of symptoms in children. The differences in Italy and Australia are most likely due to both countries being in different stages of the pandemic at the time of the survey. Other different parental features that impact parental distress levels may account for differences in pandemic’s impact between Italy and Australia (e.g., educational attainment in Italy and parent–child relationship in Australia) may demand further investigations using larger samples.A number of limitations need due consideration while interpreting the findings. Firstly, this is a cross-sectional survey and hence the inter-relationships and directionality of the impact over time are unclear. This survey was conducted during the early phase of the pandemic in June/July 2020 and hence the long-lasting economic and social impact of the pandemic and the consequent effects on parental mental health and the parent–child interactions may not have been fully unfolded at that time. Secondly, parents were asked to participate in a survey which may have led to self-selection bias that might stem from any circumstances that distinguish the parents who participated in the survey from the other non-participating parents. Further, the survey was performed as a parent report and hence the self-report bias cannot be excluded. Thirdly, the modest sample size in both Italy and Australia was only intended to relate to a portion of the NDD community. Although cross-country comparisons could shed some light on the association between environmental factors and health issues, caution needs to be exercised in the interpretation of these findings for several reasons. First, multiple unmeasured confounders may be involved in differences between countries. Second, ecological fallacy may arise when we use aggregate data to infer individual data [33]. In the current study, we did not intend to use aggregate data to infer the relationship between parental distress and children’s features. However, the difference in parental distress levels in relation to children’s clinical features between the two countries may not be attributable to the difference in aggregate data, such as the magnitude of the disease burden associated with COVID-19 between these two countries. Finally, self-reported categorical data lack cross-population comparability [34]. Taken together, these reasons could limit the interpretations of our findings in the context of cross-country comparison. Future studies with larger representative samples are needed for ensuring the generalisability of the findings. Additionally, longitudinal data on changes in distress levels at multiple time points over the course of the pandemic may cast insights into the long-term impact of the pandemic on the wellbeing of parents of children with neurodevelopmental disorders.The current study is the first to examine cross-country comparisons of parents of children with NDDs by comparing Italy (longer lockdowns and school closures in 2020) versus Australia (shorter lockdowns and school closures in 2020). The results indicate that the pandemic might yield different impacts on the psychological wellbeing of parents (or caregivers) of children with different types of NDDs, and such differences might also vary between Italy and Australia. Although it appears that the difference in how stringently anti-contagion measures were implemented might yield, at best, a limited impact on how parental wellbeing could be affected by NDDs in children, more research is warranted to investigate the relationship between the level of stringency of lockdowns and school closures parental wellbeing. Family and social support measures aligned with the unique characteristics of each vulnerable group need to be revised to cater to the unique needs of families of children with different NDD diagnoses.The following are available online at https://www.mdpi.com/article/10.3390/ijerph182111066/s1, Table S1: Impacts on children and families (those who agree or strongly agree with each statement).Conceptualization, P.-I.L. and V.E.; formal analysis, D.B., P.-I.L.; methodology, D.B. and P.-I.L.; investigation, P.-I.L. and A.M.; data curation, A.M. and A.M.D.; resources, R.R. and V.E.; writing—original draft preparation, D.B. and P.-I.L.; writing—review and editing, D.B., P.-I.L., A.M., A.M.D., R.R. and V.E.; visualization, D.B. and P.-I.L.; supervision, P.-I.L.; project administration, V.E. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of University of New South Wales.Informed consent was obtained from all subjects involved in the study.The dataset has been uploaded to a public database: doi: 10.6084/m9.figshare.16528998.The authors declare no conflict of interest.Proportions in the variation in parental stress explained by the child’s NDD diagnosis.Parental distress levels stratified by the child’s diagnosis and country. Vertical bars (red and blue bars indicate Italian and Australian samples, respectively) refer to the 95% CIs.Comparing the effects of symptomatic change of NDDs on parental distress between Italy and Australia. SASD indicates the symptomatic change for ASD; SILD indicates the symptomatic change for ILD; SOCD indicates the symptomatic change for OCD; STS indicates the symptomatic change for TS. Lower K6_sum scores indicate higher distress levels. Higher scores in the symptomatic change indicate worsening of symptoms during the pandemic.Demographic features of the two samples.† TAFE: Technical and Further Education. ‡ Geographic regions outside major metropolitan areas. * Chi-squared test p-value < 0.05.Diagnoses and perceived effect of the pandemic on symptoms changes in Italy and Australia.* Chi-squared test p-value < 0.05.Caregivers’ mental health and wellbeing.* Chi-squared test p-value < 0.05.Linear regression results of NDDs on parental distress †.* p < 0.05, † After adjusting for the other NDDs, child age, caregivers’ age and educational level.Linear regression results of the total sample *,†.* p < 0.05, † After adjusting for the other NDDs, child age, caregivers’ age and educational level.Linear regression results of the symptomatic changes of specific NDDs on parental distress *,†.* p < 0.05, † After adjusting for caregivers’ age and educational level.Linear regression results of the symptomatic changes of specific NDDs on parental distress *,†.*p < 0.05, † After adjusting for caregivers’ age and educational level.Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Med-MDPI/ijerph_8/ijerph-18-21-11067.txt

@@ -0,0 +1 @@
+Background: The purpose of this study was to analyze the relationship between COVID-19 preventive behaviors, as the dependent variable, with risk perception, coping style and sense of coherence, as independent variables, in older people living in the community. Methods: An observational design for predictive model development. This study was reported following the STROBE statement. The subjects were people over 65 years of age living in the community. Data collection included sociodemographic variables related to COVID-19, risk perception and types, coping styles in the face of contagion, sense of coherence, and preventive behaviors in the face of COVID-19. The data collection period was from November 2020 to January 2021. Results: A total of 305 people participated in this study (71.5% women, mean age 71.34 years; 6.9% suffered from COVID-19 and 44.3% knew someone close to them who suffered from the virus). The coping style variables problem-focused, emotion-focused, and sense of coherence subscales Significance and manageability explained 17% of the variable preventive behaviors against COVID-19. There were statistically significant differences by gender in all subscales, with women scoring higher in all of them; Conclusions: Men with low risk perception, extrinsic risk perception, and low sense of coherence presented worse COVID-19 preventive behaviors. It would be interesting to develop specific prevention and health education campaigns for this population.The rapid spread of COVID-19 and the severity of the symptoms it can cause in a segment of infected individuals has stretched health systems to the limit [1]. The most vulnerable group and the one that is suffering the highest number of deaths globally, with exorbitant figures, is that of older people [2]. This group is mainly located in the community [3]. Experts worldwide warn that the problem is not going to disappear in the short term and that the security measures adopted by individuals will be key to ensure that the situation remains under control [4,5].From the epidemiological point of view, it would therefore be necessary to know, especially for the purpose of health promotion and prevention, what the behaviors of older people are towards the adoption of safety measures to avoid contagion, based on their beliefs and attitudes. This is an aspect that the WHO considers to be a priority [6]. Several recent studies in China and Italy have investigated the risk perception and coping strategies followed by older people [7,8]. Their results are conclusive: older people estimate their risk of COVID-19 to be lower than younger people. Women are more concerned about COVID-19 than men.These data are in line with the literature regarding the health belief model, in which older people, paradoxically, present higher illusory optimism and lower risk perception [9]. These are well-studied facts within processes such as treatment adherence or chronic disease management [10]. The psychological and cognitive processes underlying health belief models, already in place since the 1970s, clearly indicate that constructs such as risk perception, coping styles or adaptation to difficult situations justify the difficulties encountered by health professionals in ensuring good adherence to treatment and adequate control of one’s own health [11]. Of all the health models, this paper will focus on the salutogenic model [12]. This model relates the management of stressful situations (such as this pandemic) to the individual’s capacity for self-management of such situations. It develops concepts such as sense of coherence (SOC) which is directly related to the ability to employ cognitive, affective, and instrumental strategies that help improve the ability to adapt to difficult situations. In health care, the salutogenic paradigm can be developed either for the design of interventions or to reorient health care research [13].Clearly something has gone wrong with the instructions given to the general population to adopt safety measures in the face of COVID-19 disease progression as for example in Madrid [14]. Recent research points out that health systems have placed more financial effort and resources on hospital and clinical care and have decreased their focus on the community and this has taken a noticeable toll on the containment of the pandemic [15]. Studies related to etiology, clinical control of the disease, the search for valid diagnostic tests and finding an effective vaccine are undoubtedly necessary and important. However, experts already warn that the main key to control is to prevent the onset of the disease and not solely to treat it when its spread can no longer be contained [16,17]. We know that the greatest number of infections occur within the community [1,18]. Therefore, efforts should be made to implement strategies in the community, on behalf of Primary Care services. The Community Nurse is the professional of reference for many older people. If we were to explore older people from the perspective of health belief models, we could design specific strategies to ensure a self-effective control and a realistic coping style in the face of the pandemic [19,20].It was hypothesized that older people with low risk perception and ineffective coping style are at a higher risk of not adhering to preventive measures [7,8]. However, we incorporated the variable sense of coherence (SOC) and hypothesized that older people with low SOC have lower risk perception, worse coping style, and inadequate preventive behaviors against the risk of COVID-19 infection.To our knowledge, this is the first study in our country (xx) that specifically measures risk perception and coping styles for COVID-19 disease in community-dwelling older people. Its value is the development of a logistic regression model that analyzes the relationship between COVID-19 preventive behaviors and the variables SOC, risk perception and coping style. This is the first international study, to our knowledge, to carry out a predictive model incorporating the variable sense of coherence together with risk perception and coping style, in community-dwelling older people, within the salutogenic health framework.The contribution of this study will be an important aid for nurses because, based on the predictive model, we will be able to easily detect those older people with poorer adherence to preventive measures based on their SOC, coping style and risk perception. The results obtained will enable the design of specific health and promotion strategies to favor health behaviors in the face of COVID-19 in older people. The study is led and conducted by nurses, as health agents.This paper has rigorously followed The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement. Furthermore, the researchers have used structured research instruments. The results are based on larger sample sizes that are representative of the population. In addition, this research study can be replicated or repeated, given its high reliability. Researchers have a clearly defined research question to which objective answers are sought. This project can be used to generalize concepts more widely, predict future results, or investigate causal relationships. The researchers used tools, such as questionnaires or computer software, to collect numerical data.The purpose of this study was to analyze the relationship between COVID-19 preventive behaviors, as the dependent variable, with risk perception, coping style and sense of coherence, as independent variables, in older people living in the community.This study was based on an observational design for predictive model development. The data collection period was from November 2020 to January 2021.The target of the sampling survey was older individuals over 65 years old in XX (north of Spain). We excluded participants who were not cognitively competent, as well as participants who had language impairments, i.e., reading comprehension impairments. A random sample selection was made in Health Centers in XX. The sample size was calculated from the total number of people over 65 years of age in 2020 (128,494) for a confidence level of 95% and considering a beta error of 0.2 for an expected proportion of 80% of potential participants, we estimated a minimum requirement of 127 people. To make a logistic regression model with statistical robustness, in this type of study the literature advises a sample size ten times the number of independent variables to be estimated plus one. The sample size was estimated using the Granmo 7.11 program. Subject selection was randomized following a table of random numbers.Demographic characteristics were described by categorical variables (sex, age, place of residence and degree of independence using the Barthel Index [21]). COVID-19 variables were used based on a dichotomous response (yes/no) to the following questions: Have you suffered from the disease? Has anyone in your close environment suffered from the disease? Has anyone in your close environment died from COVID-19? * Preventive behaviors scale for COVID-19 (preventive behaviors, PB). This was created ad hoc by a panel of 11 experts (6 nurses specializing in Public Health and 5 physicians specializing in Public Health and Epidemiology). It was developed from the literature [22]. The survey consisted of 19 positive items on preventive behaviors against contagion (wearing a mask, hand washing, etc.), which were answered according to the degree of agreement (1 Strongly disagree to 5 Strongly agree). The factor analysis identified a single factor structure that was called preventive behaviors (Cronbach’s alpha .77). The higher the score, the greater the knowledge and the better the preventive behaviors against COVID-19. The range of scores is from 19 to 95 points.* Sense of Coherence was evaluated using the Orientation to Life Questionnaire-13 Items (OLQ-13 or SOC-13) [23] in this work the Spanish validated version [24] was used. The instrument aims to measure a global personality orientation that facilitates adaptive problem solving in stressful situations to which people are subjected throughout life. As in the extended questionnaire, the 13-item questionnaire also measures the dimensions of comprehensibility (with 5 items), manageability (with 4 items) and meaningfulness (with 4 items). The scores obtained express the strength in the sense of coherence of the person, the higher the score obtained, the greater the strength. The answers offer a continuum from agreement to disagreement in 7 response options, represented on a Likert-type scale, from 1 to 7, ranging from "Never", "Rarely" to "Very often" or "Always", both in the positive and negative sense of the question. The OLQ-13 scale has shown good internal consistency, with a Cronbach’s alpha between 0.70 and 0.92 [23,24] and retains the same psychometric qualities as the original 29-item version. Regarding our study, the internal consistency of the items was analyzed using Cronbach’s alpha, which was 0.71, for the comprehensibility subscale it was 0.81, for manageability it was 0.79 and for Significance it was 0.71.* For the assessment of risk perception, perceived risk factors and coping styles towards COVID-19, a specific questionnaire was designed by a panel of 9 experts (3 nurses specialized in Public Health, 4 psychologists specialized in behavior modification and health education and 2 physicians specialized in Public Health and Epidemiology) based on the literature [25, 26, 27]. The questionnaire was validated in a sample of 30 elderly, with good psychometric properties and consists of the following three scales:—Perceived Risk Scale. Consisting of 3 items, in which the person had to show his or her degree of agreement using a Likert scale (0 being No risk and 10 Maximum risk). The maximum score was 30 points, indicating that the higher the score, the greater the perception of risk of infection by COVID-19. The factor analysis identified a one-factor structure which we will call risk (Cronbach’s alpha .735).—Perceived risk factors for Contagiousness Scale. Composed of 16 items in which the person showed his or her degree of agreement according to a Likert scale (1, Strongly disagree to 5, Strongly agree). The factor analysis identified a two-factor structure (Cronbach’s alpha .781). The two factors identified correspond to how the person perceives the risk factors: as external or environment-dependent risk factors with 9 items (extrinsic risk factor, alpha .721,) or as factors that depend on the individual, with 7 items (intrinsic risk factor, alfa .841). The higher the score, the greater the weight of one risk factor over the other. The range for the extrinsic risk factor is 9–45 and for the intrinsic risk factor it is 7–35 points. The intrinsic factor is desirable because it speaks of the things I can do to protect myself, while the extrinsic factor speaks of the inevitability of the disease and of factors that are beyond my control and over which I can do nothing. The results are consistent with the literature [7,28].—Coping Styles Scale with Contagion. These were assessed based on 14 items, in which the person showed his or her degree of agreement according to a Likert scale (1 Strongly disagree to 5 Strongly agree). The factor analysis identified a two-factor structure (Cronbach’s alpha .793). The two factors identified correspond to two coping styles in relation to COVID-19: Problem-focused (7 items, alpha .801) and emotion-focused (7 items, alpha .785). The higher the score, the more one risk factor weighs against the other. The range for problem-focused and for emotion-focused is 7–35 points. Of the coping styles that coincide with the literature, problem-focused is preferable, as it is a style based on the active search for solutions [8,29]. The nurses participating in the study made a random selection of those persons from their Health Center who met the inclusion criteria. After an explanation of the project, participants signed an informed consent form. Participants completed all variables using an anonymous online survey. To control the questionnaire quality, the same IP address was only allowed to answer once. The questionnaire was designed to be answered from a single IP. In the case of persons without internet access to answer the questionnaire online or who only had a computer/tablet at home to answer the questionnaire, the nurse collected the data via telephone. The data collection period was from November 2020 to January 2021.The data were analyzed using IBM SPSS Statistics 22 (IBM, Boston, MA, USA), a p value of <0.05 (two-tailed) was considered statistically significant. For the descriptive analysis, all variables were analyzed to identify and correct for outliers and missing data. For the analysis of possible missing values, we used the EM (expected maximization) method. We adopted a bilateral contrast and a 95% confidence level. Descriptive statistics (means, standard deviations, and percentages) were used to describe the sample. A comparison was carried out between people who had experienced COVID-19 and those who had not for the variables SOC, coping styles, risk factors and preventive behaviors by means of the Student’s t-test for independent samples. Subsequently, a bivariate correlation analysis was carried out between all the variables in the study using the Pearson’s r test. Only those variables that showed significant correlations with the preventive behaviors scale were introduced in a stepwise, multiple linear regression analysis (MLR) to determine the best predictors. The assumptions of the MLR model were evaluated by means of the following analyses: (1) normality: Kolmogorov–Smirnov test and P–P Normal plots; (2) linearity: partial regression plots; (3) homoscedasticity: scatter plots of typed residuals and typed predictors; (4) independence of errors: Durbin–Watson statistic; and (5) noncollinearity: diagnostics of collinearity (Tolerance >.10 and IVF <.10). Considering the literature on the model of health beliefs and health-generating behaviors, we decided to choose a single predictive model that predicts the high-risk outcome of inappropriate knowledge behavior in the face of contagion risk.The study was designed according to the ethical standards of the Declaration of Helsinki and received approval from the ethical committee of Servicio xxx de Salud (Approval number: CE INNVAL 20/31) and by the Primary Health Care Management of xx according to the specific national guidelines and conformed to the principles of the Declaration of Helsinki. Prior to data collection, all the participants were provided information concerning the study and signed the informed consent. Participants were informed that all answers would be anonymous.A total of 305 people responded (exceeding 35% of the required sample size). Table 1 shows the sociodemographic variables, as well as the questions related to the COVID-19. A total of 67.9% suffered from some chronic disease (hypertension and diabetes mellitus being the most prevalent). The mean Barthel Index was 83.95 ± 8.39.Table 2 shows the mean values for the following scales preventive behaviors, risk and risk factors, by gender and COVID-19 variables. The score obtained on the preventive behaviors scale, which reached a mid-level in the general population (59.26 +/− 4.99 of a maximum score of 95, statistically significant compared to the mean of the scale itself, t = 5.23, p = 0.03), was higher and statistically significant in women and in people who had not had the disease, as well as in those who had had someone close to them sick or deceased. The risk scale score was higher than the sample mean (t = 0.12, p = 0 .03). Statistically significant differences were found by gender, with women scoring higher. In addition, among people who had had someone sick in their environment, those who had had someone sick scored higher (p < 0 .001). The sample analyzed showed a medium perception of the risk of infection by COVID-19 mean (t = 2.36, p = 0.02), with a greater weight of the extrinsic risk factor (t = 2.36, p = 0.02). There were statistically significant differences in the two subscales of risk factors between those who had experienced COVID-19 and those who had not: those who had experienced COVID-19 scored higher in the extrinsic factor and those who had not had COVID-19 scored higher in the intrinsic factor. There were also differences by gender in the intrinsic factor, with women scoring higher. In addition, also for the extrinsic factor, men scored higher.Table 3 shows the results for the coping styles subscales as a function of gender and illness-related questions. The problem-focused coping style scored the highest in the total sample. There were statistically significant differences by gender for both coping styles, with women scoring higher in both, but especially in the problem-focused style. We also found statistically significant differences between those who had suffered from the disease and those who had not for both coping styles, with the emotion-focused style scoring higher in those who had suffered from the disease. Differences were also found between people who had had a death from COVID-19 in their close environment, with those who had had a death from COVID-19 scoring higher in the emotion-focused style. In addition, the same was found for those who had had a patient close to them.When calculating the sense of coherence value for the entire sample, we obtained a total mean SOC of 50.58 ± 11.43 points (out of a total of 91), with the relative order of the dimensions, according to their percentage of each total, being comprehensibility, meaningfulness and manageability (Table 4). Women scored higher in all subscales except comprehensibility, although statistically significant differences were found only in total SOC. Statistically significant differences were found for the SOC Total and comprehensibility subscales between those who had and had not had the disease (higher scores for those who had). Statistically significant differences were also found in the manageability subscale among people who had experienced a death in their close environment.A statistically significant association was found for age and problem-focused coping style (r = 0.139, p < 0.001). However, no association was found between the other sociodemographic variables and the scales studied. Neither was there any association between the COVID-19 variables (having suffered from the virus oneself, someone in the patient’s environment suffering from it and/or someone close dying) and the scales studied. A statistically significant association was found between the variable preventive behaviors and the subscales SOC manageability (r = −0.201, p < 0.001), meaningfulness (r = −0.244, p < 0.001), and the two subscales of coping styles (problem-focused r = 0.041, p < 0.001; emotion-focused r = 0.381, p < 0.001. However, not for the remaining scales. For the analysis of the predictive capacity of the preventive behaviors variable of the study variables, Multiple Regression Analysis (MRA) models were applied using the forward stepwise method, after verifying that all the assumptions were met. Given the limitations posed by the sample size for the inclusion of many variables in the regression model, we first performed two MRAs considering the sociodemographic characteristics (age, sex, place of residence and Barthel Index) and the COVID-19 variables (having suffered from the virus oneself, someone close to the patient dying from COVID-19 or having suffered from the virus) as predictors. None of these models were significant: Sociodemographic characteristics (F = 0.22, R2 = 0.001, p = 0.1) and COVID-19 variables (F = 0.39, R2 = 0.045, p = 0.247). Consequently, the need to include any of these variables as control variables in the multiple regression model of the study variables was ruled out.A forward stepwise multiple regression analysis was carried out using the preventive behaviors variable as the dependent variable and the SOC manageability and meaningfulness subscales, and the problem-focused and emotion-focused coping styles subscales as predictor variables. The model was significant (F = 35.80; p < 0.001) explaining 33.5% of the variance of the criterion variable through the predictor variables problem-focused, emotion-focused, SOC Manageability and meaningfulness. Problem-focused coping style was the most relevant predictor (Beta = 0.20; p < 0.001) explaining on its own 17% of the variance of the dependent variable, followed by emotion-focused coping style (Beta= 0.41; p < 0.001), SOC meaningfulness (Beta = 0.23; p < 0.001) and SOC manageability (Beta = 0.17; p = 0.002) (Table 5).Older adults are the most vulnerable group in terms of morbidity and mortality during the COVID-19 pandemic [30]. Adopting preventive measures against contagion is the most effective measure in addition to vaccination [31]. Exploring risk perception, coping styles and preventive behaviors towards COVID-19 is crucial for the design of preventive strategies within the community. Incorporating another variable such as sense of coherence, within the salutogenic model of health, will help us to understand the underlying mechanisms of adherence to preventive measures. The main findings found in this study were: (i) older people presented average preventive behaviors, especially women and people who had not suffered from the disease and those who had a close relative who was ill and/or died from COVID-19 scored high. (ii) in a predictive model, the two coping styles problem-focused, emotion-focused, and the two subscales SOC Significance and manageability explained up to 17% of the variance of the variable preventive behaviors. These and other relevant results are discussed below.Older people in our sample presented good knowledge and preventive behaviors towards COVID-19. This is repeated in other similar studies, although the results are disparate [3,32]. There are studies indicating that older people present lower risk perception, due to illusory feelings of invulnerability and illusory optimism, among others [8]. In our study, the values have been average, which does seem to be consistent with their behavior compared to similar previous diseases [33]. In our case it seems that older people have preventive behaviors, although we do not know if this is due to fear [34], or to a greater responsibility compared to other groups of people with a different age or lack of chronic diseases [35]. Women scored higher, which is widely corroborated in other studies [3,7]. It is also notable that people who had not suffered from the disease presented better behaviors than those who had, a finding that speaks highly of the importance of preventive behaviors. The use of different scales largely conditions the ability to compare our findings with other studies. Nevertheless, our scale, preventive behaviors, was based on the literature, was developed by a panel of experts, and showed good properties.The most predominant risk perception was the extrinsic risk factor. This indicates that the people in the sample perceived that the risk factors depend more on causes external to them (contact with other people, inevitability of the disease, pandemic, etc.) [36]. This seems consistent considering that it is a global pandemic, with numerous infections and deaths. Uncertainty and ignorance of the cause generates a feeling of a lack of control over the situation [37].In terms of coping styles in the face of contagion, problem-focused scored the highest. It also showed a significant association with age. Older people were more problem-focused. This indicates that older people focused on the problem and how to deal with it to respond to it. The other style, emotion-focused, also with average values, although which were lower, indicated a style more centered on focusing on feelings as a means of avoiding the situation. Women scored statistically higher than men in both styles, for which problem-focused stood out. There is evidence indicating that women present a greater sense of responsibility in the face of health problems, with more preventive and reality-focused behaviors [7,37,38].The SOC variable presented mean values in all its subscales, with women scoring significantly higher in all subscales except for the comprehensibility subscale. Presenting a medium–high SOC is related to healthier aging, greater adaptation to stress and greater adherence to healthy behaviors [39,40]. This is something which according to the literature is more present in women than in men, although no conclusive explanations for this have been drawn [41]. Focusing on the differences between people who have had the disease and those who have not, this study reveals interesting findings. Statistically significant differences were found for the subscales of the perceived risk factors. People who had experienced COVID-19 scored higher on the extrinsic factor. This may be related to the fact that those who have been infected may not have strictly complied with safety measures. In part, this may be due to considering that the causes of the virus are alien, extrinsic and therefore uncontrollable [32]. There were also statistically significant differences in coping styles. Those who had suffered from the disease, or who had a family member who had suffered from or died from the virus, scored higher on emotion-focused. It seems consistent to suppose that having been in contact with the disease favors more avoidant coping styles as the person avoids thinking about the problem. These results are similar to other studies [7,37,42]. Regarding the sense of coherence values, we found statistically significant differences between those who had suffered from COVID and those who had not, with higher scores on the comprehensibility and total SOC subscales for those who had been ill. The SOC focuses on the ability to adapt to stressful situations. Comprehensibility refers to the degree to which people make cognitive sense and deal with situations by considering information in an orderly, consistent, structured, and clear manner. It seems clear that having suffered from the disease requires a cognitive structuring of the person to understand what happened, adapt to the consequences, and better understand the situation experienced [39,40].Undoubtedly, the interesting part of this study is the predictive model. In this model, the two coping styles and two subscales of the SOC, meaningfulness and manageability, stand out as predictors of preventive behaviors. Interestingly, coping styles are the strongest predictors, especially problem-focused. It seems logical that people who focus more on coping and take an active role in preventing the disease also have adequate preventive behaviors [35,43]. Regarding the SOC scale variables, the results seem consistent with other similar findings suggesting that those people who attach high meaning to their life events develop a sense of commitment and involvement with them and this involvement in turn favors them to assume them as structured, predictable, and explainable events [41,44]. In other words, medium–high levels of SOC seem to suggest that they experienced their situations as understandable and meaningful, developing effective preventive behaviors.The results of this study should be interpreted with caution, as the cross-sectional nature of the design prevents us from determining the direction of the relationships between variables. In addition, there are other limitations: all measures were self-report questionnaires; the selection of participants was based on convenience sampling and there was an absence of follow-up data. Although the established theoretical sample size was reached, perhaps a larger participation of subjects would improve the findings of the predictive model.To our knowledge, no research has been conducted in Spain analyzing risk perception, coping styles and preventive behaviors against COVID-19 in older people living in the community. Furthermore, no study has examined sense of coherence at a global level, framed within a predictive model.This exploratory study has enabled us to identify that coping styles based especially on focusing on the problem, as well as the subscales of SOC Significance and manageability, can explain up to 17% of the variance of preventive behaviors against COVID-19 in older people. Identifying elderly, especially men, with low risk perception, in addition to an extrinsic risk perception and low sense of coherence, seems key for implementing specific prevention measures.The study of risk perception, coping styles, sense of coherence and preventive behaviors in the older population is essential during emergencies to improve guidance regarding the approach and prevention strategies that nurses design in the Primary Care setting.Conceptualization, C.M.S.-C.; Methodology, C.M.S.-C.; Supervision, C.M.S.-C.; Data collection: A.D.U.; M.G.L.; E.I.P.; C.M.R.; Writing—original draft, C.M.S.-C.; Writing—review and editing, Á.F.-R.; A.D.U.; M.G.L.; E.I.P.; C.M.R. All authors have read and agreed to the published version of the manuscript.This research was funded by IDIVAL, Spain, grant number INNVAL 20/31.The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of CEICm, Servicio Cántabro de Salud (protocol code CE INNVAL 20/31, 22/12/2020).Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patient(s) to publish this paper.Not applicable.Acknowledge to all the participants.The authors declare no conflict of interest.Sociodemographic characteristics of the sample (N = 305).Gender variables and COVID-19 questions by risk scale, perceived risk factors and preventive behaviors (descriptive and differential analysis).* p value < 0.05. ** p value < 0.001. Extrinsic = extrinsic risk factor. Intrinsic = intrinsic risk factor. PB = preventive behaviors. a. Anyone close to you = family, close friends.Coping styles in the face of COVID-19.* p value < 0.05. ** p value < 0.001. Problem F = problem-focused. Emotion F = emotion-focused.Sense of Coherence (SOC) and subscales.* p value < 0.05. ** p value < 0.001. SOC1 (Comprehensibility); SOC2 (Manageability); SOC3 (Significance).Multiple Regression Analysis Model.Dependent variable: Having suffered from COVID-19. Problem f = problem-focused. Emotion f = emotion-focused. Significance = SOC Significance. Manageability = SOC Manageability. R2 total model = 0.337; R2 goodness-of-fit = 0.327 (F = 35.80; p < 0.001).Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.